quda Namespace Reference

Namespaces
	asym_clover
	blas
	clover
	colorspinor
	cublas
	domainwall
	domainwall4d
	dslash
	dslash_aux
	fermion_force
	gauge
	improvedstaggered
	linalg
	mobius
	ndegtwisted
	pack
	pool
	staggered
	twisted
	twistedclover
	wilson

Classes
struct	AllocType
struct	AllocType< false >
struct	AllocType< true >
class	ArpackArgs
class	Array
class	BiCGstab
class	BiCGstabL
class	BiCGstabLUpdate
struct	bridge_mapper
struct	bridge_mapper< double2, double2 >
struct	bridge_mapper< double2, float2 >
struct	bridge_mapper< double2, float4 >
struct	bridge_mapper< double2, short2 >
struct	bridge_mapper< double2, short4 >
struct	bridge_mapper< float2, double2 >
struct	bridge_mapper< float2, float2 >
struct	bridge_mapper< float2, short2 >
struct	bridge_mapper< float4, double2 >
struct	bridge_mapper< float4, float4 >
struct	bridge_mapper< float4, short4 >
class	CalculateY
struct	CalculateYArg
class	CalculateYhat
struct	CalculateYhatArg
class	CG
class	CGNE
class	CGNR
struct	ChecksumArg
struct	ChiralToNonRelBasis
class	Clover
struct	clover_mapper
struct	clover_mapper< double, N >
struct	clover_mapper< float, N >
struct	clover_mapper< short, N >
struct	clover_wrapper
	clover_wrapper is an internal class that is used to wrap instances of colorspinor accessors, currying in a specifc location and chirality on the field. The operator() accessors in clover-field accessors return instances to this class, allowing us to then use operator overloading upon this class to interact with the HMatrix class. As a result we can include clover-field accessors directly in HMatrix expressions in kernels without having to declare temporaries with explicit calls to the load/save methods in the clover-field accessors. More...
struct	CloverArg
	Parameteter structure for driving the clover and twist-clover application kernels. More...
class	CloverField
struct	CloverFieldParam
struct	ColorSpinor
struct	ColorSpinor< Float, Nc, 2 >
struct	ColorSpinor< Float, Nc, 4 >
struct	colorspinor_ghost_wrapper
	colorspinor_ghost_wrapper is an internal class that is used to wrap instances of colorspinor accessors, currying in a specifc location on the field. The Ghost() accessors in colorspinor-field accessors return instances to this class, allowing us to then use operator overloading upon this class to interact with the ColorSpinor class. As a result we can include colorspinor-field accessors directly in ColorSpinor expressions in kernels without having to declare temporaries with explicit calls to the loadGhost/saveGhost methods in the colorspinor-field accessors. More...
struct	colorspinor_mapper
struct	colorspinor_mapper< double, 1, Nc >
struct	colorspinor_mapper< double, 2, Nc >
struct	colorspinor_mapper< double, 4, Nc >
struct	colorspinor_mapper< float, 1, Nc >
struct	colorspinor_mapper< float, 2, Nc >
struct	colorspinor_mapper< float, 4, Nc >
struct	colorspinor_mapper< short, 1, Nc >
struct	colorspinor_mapper< short, 2, Nc >
struct	colorspinor_mapper< short, 4, Nc >
struct	colorspinor_order_mapper
struct	colorspinor_order_mapper< T, QUDA_FLOAT2_FIELD_ORDER, Ns, Nc >
struct	colorspinor_order_mapper< T, QUDA_SPACE_COLOR_SPIN_FIELD_ORDER, Ns, Nc >
struct	colorspinor_order_mapper< T, QUDA_SPACE_SPIN_COLOR_FIELD_ORDER, Ns, Nc >
struct	colorspinor_wrapper
	colorspinor_wrapper is an internal class that is used to wrap instances of colorspinor accessors, currying in a specifc location on the field. The operator() accessors in colorspinor-field accessors return instances to this class, allowing us to then use operator overloading upon this class to interact with the ColorSpinor class. As a result we can include colorspinor-field accessors directly in ColorSpinor expressions in kernels without having to declare temporaries with explicit calls to the load/save methods in the colorspinor-field accessors. More...
class	ColorSpinorField
class	ColorSpinorParam
struct	complex
struct	complex< double >
struct	complex< float >
struct	CompositeColorSpinorFieldDescriptor
class	CopyColorSpinor
class	CopyColorSpinor< FloatOut, FloatIn, 4, Nc, Arg >
struct	CopyColorSpinorArg
class	CopyGauge
struct	CopyGaugeArg
class	CopyGaugeEx
struct	CopyGaugeExArg
class	CopySpinor
class	CopySpinorEx
struct	CopySpinorExArg
class	CovD
class	cpuCloverField
class	cpuColorSpinorField
class	cpuGaugeField
class	cudaCloverField
class	cudaColorSpinorField
class	cudaGaugeField
struct	deflated_solver
class	Deflation
struct	DeflationParam
class	Dirac
class	DiracClover
class	DiracCloverPC
class	DiracCoarse
class	DiracCoarsePC
class	DiracDomainWall
class	DiracDomainWall4DPC
class	DiracDomainWallPC
class	DiracImprovedStaggered
class	DiracImprovedStaggeredPC
class	DiracM
class	DiracMatrix
class	DiracMdag
class	DiracMdagM
class	DiracMMdag
class	DiracMobius
class	DiracMobiusPC
class	DiracParam
class	DiracStaggered
class	DiracStaggeredPC
class	DiracTwistedClover
class	DiracTwistedCloverPC
class	DiracTwistedMass
class	DiracTwistedMassPC
class	DiracWilson
class	DiracWilsonPC
struct	DslashCoarseLaunch
class	DslashCoarsePolicyTune
struct	DslashConstant
	Constants used by dslash and packing kernels. More...
class	Eig_Solver
class	EigCGArgs
class	EigParam
class	ExtractGhost
struct	ExtractGhostArg
class	ExtractGhostEx
struct	ExtractGhostExArg
struct	FullClover
class	Gamma
struct	GammaArg
	Parameter structure for driving the Gamma operator. More...
struct	gauge_ghost_wrapper
	gauge_ghost_wrapper is an internal class that is used to wrap instances of gauge ghost accessors, currying in a specific location and dimension on the field. The Ghost() accessors in gauge-field accessors return instances to this class, allowing us to then use operator overloading upon this class to interact with the Matrix class. As a result we can include gauge-field ghost accessors directly in Matrix expressions in kernels without having to declare temporaries with explicit calls to the load/save methods in the gauge-field accessors. More...
struct	gauge_mapper
struct	gauge_mapper< double, QUDA_RECONSTRUCT_12, N, stag, huge_alloc >
struct	gauge_mapper< double, QUDA_RECONSTRUCT_13, N, stag, huge_alloc >
struct	gauge_mapper< double, QUDA_RECONSTRUCT_8, N, stag, huge_alloc >
struct	gauge_mapper< double, QUDA_RECONSTRUCT_9, N, stag, huge_alloc >
struct	gauge_mapper< double, QUDA_RECONSTRUCT_NO, N, stag, huge_alloc >
struct	gauge_mapper< float, QUDA_RECONSTRUCT_12, N, stag, huge_alloc >
struct	gauge_mapper< float, QUDA_RECONSTRUCT_13, N, stag, huge_alloc >
struct	gauge_mapper< float, QUDA_RECONSTRUCT_8, N, stag, huge_alloc >
struct	gauge_mapper< float, QUDA_RECONSTRUCT_9, N, stag, huge_alloc >
struct	gauge_mapper< float, QUDA_RECONSTRUCT_NO, N, stag, huge_alloc >
struct	gauge_mapper< short, QUDA_RECONSTRUCT_12, N, stag, huge_alloc >
struct	gauge_mapper< short, QUDA_RECONSTRUCT_13, N, stag, huge_alloc >
struct	gauge_mapper< short, QUDA_RECONSTRUCT_8, N, stag, huge_alloc >
struct	gauge_mapper< short, QUDA_RECONSTRUCT_9, N, stag, huge_alloc >
struct	gauge_mapper< short, QUDA_RECONSTRUCT_NO, N, stag, huge_alloc >
struct	gauge_order_mapper
struct	gauge_order_mapper< T, QUDA_BQCD_GAUGE_ORDER, Nc >
struct	gauge_order_mapper< T, QUDA_FLOAT2_GAUGE_ORDER, Nc >
struct	gauge_order_mapper< T, QUDA_MILC_GAUGE_ORDER, Nc >
struct	gauge_order_mapper< T, QUDA_QDP_GAUGE_ORDER, Nc >
struct	gauge_order_mapper< T, QUDA_QDPJIT_GAUGE_ORDER, Nc >
struct	gauge_order_mapper< T, QUDA_TIFR_GAUGE_ORDER, Nc >
struct	gauge_order_mapper< T, QUDA_TIFR_PADDED_GAUGE_ORDER, Nc >
struct	gauge_wrapper
	gauge_wrapper is an internal class that is used to wrap instances of gauge accessors, currying in a specific location on the field. The operator() accessors in gauge-field accessors return instances to this class, allowing us to then use operator overloading upon this class to interact with the Matrix class. As a result we can include gauge-field accessors directly in Matrix expressions in kernels without having to declare temporaries with explicit calls to the load/save methods in the gauge-field accessors. More...
class	GaugeCovDev
	Full Covariant Derivative operator. Although not a Dirac operator per se, it's a linear operator so it's conventient to put in the Dirac operator abstraction. More...
class	GaugeField
struct	GaugeFieldParam
class	GaugeLaplace
	Full Gauge Laplace operator. Although not a Dirac operator per se, it's a linear operator so it's conventient to put in the Dirac operator abstraction. More...
class	GaugeLaplacePC
	Even-odd preconditioned Gauge Laplace operator. More...
class	GaugeOvrImpSTOUT
struct	GaugeOvrImpSTOUTArg
class	GaussSpinor
class	GCR
class	GenericPackGhostLauncher
class	GMResDR
class	GMResDRArgs
class	HMatrix
	Specialized container for Hermitian matrices (e.g., used for wrapping clover matrices) More...
struct	HMatrix_wrapper
	wrapper class that enables us to write to Hmatrices in packed format More...
struct	Identity
class	IncEigCG
struct	Int2
struct	isHalf
struct	isHalf< short >
struct	isHalf< short2 >
struct	isHalf< short4 >
struct	KSForceArg
class	KSForceComplete
struct	KSLongLinkArg
class	KSLongLinkForce
class	Lanczos
class	Laplace
struct	LaplaceArg
	Parameter structure for driving the Laplace operator. More...
class	LatticeField
struct	LatticeFieldParam
struct	less_significant
struct	mapper
struct	mapper< double >
struct	mapper< double2 >
struct	mapper< double4 >
struct	mapper< float >
struct	mapper< float2 >
struct	mapper< float4 >
struct	mapper< short >
struct	mapper< short2 >
struct	mapper< short4 >
class	Matrix
class	MemAlloc
class	MG
struct	MGParam
class	MinResExt
class	MPBiCGstab
class	MPCG
class	MR
struct	multigrid_solver
class	MultiShiftCG
class	MultiShiftSolver
struct	NonRelBasis
struct	NonRelToChiralBasis
struct	norm_type
struct	norm_type< complex< T > >
struct	normal
struct	normal< double >
struct	normal< float >
class	Object
struct	PackGhostArg
class	PreconCG
class	PreconditionedSolver
class	PreserveBasis
class	ProjectSU3
struct	ProjectSU3Arg
struct	PromoteTypeId
struct	PromoteTypeId< complex< double >, double >
struct	PromoteTypeId< complex< float >, float >
struct	PromoteTypeId< double, complex< double > >
struct	PromoteTypeId< double, float >
struct	PromoteTypeId< double, int >
struct	PromoteTypeId< float, complex< float > >
struct	PromoteTypeId< float, double >
struct	PromoteTypeId< float, int >
struct	PromoteTypeId< int, double >
struct	PromoteTypeId< int, float >
class	QudaMatvec
class	QudaMemCopy
struct	RealType
struct	RealType< complex< double > >
struct	RealType< complex< float > >
struct	RealType< complex< short > >
struct	RealType< double >
struct	RealType< double2 >
struct	RealType< float >
struct	RealType< float2 >
struct	RealType< float4 >
struct	RealType< short >
struct	RealType< short2 >
struct	RealType< short4 >
struct	reduce_vector
struct	ReduceArg
struct	RelBasis
class	RitzMat
class	RNG
	Class declaration to initialize and hold CURAND RNG states. More...
struct	rngArg
struct	scalar
struct	scalar< double >
struct	scalar< double2 >
struct	scalar< double3 >
struct	scalar< double4 >
struct	scalar< float >
struct	scalar< float2 >
struct	scalar< float3 >
struct	scalar< float4 >
struct	scalar< short >
struct	scalar< short2 >
struct	scalar< short3 >
struct	scalar< short4 >
class	SD
struct	SharedMemory
class	ShiftColorSpinorField
struct	ShiftColorSpinorFieldArg
class	ShiftUpdate
class	SimpleBiCGstab
class	Solver
struct	SolverParam
struct	SortedEvals
struct	spin_mapper
struct	Summ
struct	Summ< double2 >
struct	Summ< double3 >
struct	Summ< double4 >
struct	TexVectorType
struct	TexVectorType< double, 1 >
struct	TexVectorType< double, 2 >
struct	TexVectorType< float, 1 >
struct	TexVectorType< float, 2 >
struct	TexVectorType< float, 4 >
struct	TexVectorType< short, 1 >
struct	TexVectorType< short, 2 >
struct	TexVectorType< short, 4 >
class	TimeProfile
struct	Timer
struct	TraceKey
class	Transfer
struct	Trig
struct	Trig< false, float >
struct	Trig< true, float >
class	Tunable
class	TunableLocalParity
class	TunableVectorY
class	TunableVectorYZ
struct	TuneKey
class	TuneParam
class	TwistClover
class	TwistGamma
struct	uniform
struct	uniform< double >
struct	uniform< float >
struct	vec_length
struct	vec_length< double >
struct	vec_length< double2 >
struct	vec_length< double4 >
struct	vec_length< float >
struct	vec_length< float2 >
struct	vec_length< float4 >
struct	vec_length< short >
struct	vec_length< short2 >
struct	vec_length< short4 >
struct	vector
struct	vector< double, 2 >
struct	vector< float, 2 >
struct	vector_type
struct	VectorType
struct	VectorType< double, 1 >
struct	VectorType< double, 2 >
struct	VectorType< double, 4 >
struct	VectorType< float, 1 >
struct	VectorType< float, 2 >
struct	VectorType< float, 4 >
struct	VectorType< short, 1 >
struct	VectorType< short, 2 >
struct	VectorType< short, 4 >
class	Worker
class	WuppertalSmearing
struct	WuppertalSmearingArg
class	XSD
struct	Zero

Typedefs
typedef std::vector< ColorSpinorField * >	CompositeColorSpinorField
typedef std::complex< double >	Complex
using	ColorSpinorFieldSet = ColorSpinorField
typedef struct curandStateMRG32k3a	cuRNGState
using	DynamicStride = Stride< Dynamic, Dynamic >
typedef std::map< TuneKey, TuneParam >	map
using	DenseMatrix = MatrixXcd
using	VectorSet = MatrixXcd
using	Vector = VectorXcd
using	RealVector = VectorXd
using	RowMajorDenseMatrix = Matrix< Complex, Dynamic, Dynamic, RowMajor >

Enumerations
enum	MemoryLocation { Device = 1, Host = 2, Remote = 4 }
enum	QudaProfileType {   QUDA_PROFILE_H2D, QUDA_PROFILE_D2H, QUDA_PROFILE_INIT, QUDA_PROFILE_PREAMBLE,   QUDA_PROFILE_COMPUTE, QUDA_PROFILE_COMMS, QUDA_PROFILE_EPILOGUE, QUDA_PROFILE_FREE,   QUDA_PROFILE_IO, QUDA_PROFILE_LOWER_LEVEL, QUDA_PROFILE_PACK_KERNEL, QUDA_PROFILE_DSLASH_KERNEL,   QUDA_PROFILE_GATHER, QUDA_PROFILE_SCATTER, QUDA_PROFILE_LAUNCH_KERNEL, QUDA_PROFILE_EVENT_RECORD,   QUDA_PROFILE_EVENT_QUERY, QUDA_PROFILE_STREAM_WAIT_EVENT, QUDA_PROFILE_FUNC_SET_ATTRIBUTE, QUDA_PROFILE_EVENT_SYNCHRONIZE,   QUDA_PROFILE_STREAM_SYNCHRONIZE, QUDA_PROFILE_DEVICE_SYNCHRONIZE, QUDA_PROFILE_MEMCPY_D2D_ASYNC, QUDA_PROFILE_MEMCPY_D2H_ASYNC,   QUDA_PROFILE_MEMCPY2D_D2H_ASYNC, QUDA_PROFILE_MEMCPY_H2D_ASYNC, QUDA_PROFILE_COMMS_START, QUDA_PROFILE_COMMS_QUERY,   QUDA_PROFILE_CONSTANT, QUDA_PROFILE_TOTAL, QUDA_PROFILE_COUNT }
enum	ComputeType {   COMPUTE_UV, COMPUTE_AV, COMPUTE_TMAV, COMPUTE_TMCAV,   COMPUTE_VUV, COMPUTE_COARSE_CLOVER, COMPUTE_REVERSE_Y, COMPUTE_COARSE_LOCAL,   COMPUTE_DIAGONAL, COMPUTE_TMDIAGONAL, COMPUTE_INVALID }
enum	DslashCoarsePolicy {   DSLASH_COARSE_BASIC, DSLASH_COARSE_ZERO_COPY_PACK, DSLASH_COARSE_ZERO_COPY_READ, DSLASH_COARSE_ZERO_COPY,   DSLASH_COARSE_GDR_SEND, DSLASH_COARSE_GDR_RECV, DSLASH_COARSE_GDR, DSLASH_COARSE_ZERO_COPY_PACK_GDR_RECV,   DSLASH_COARSE_GDR_SEND_ZERO_COPY_READ }
enum	BiCGstabLUpdateType { BICGSTABL_UPDATE_U = 0, BICGSTABL_UPDATE_R = 1 }
enum	libtype {   libtype::eigen_lib, libtype::magma_lib, libtype::lapack_lib, libtype::mkl_lib,   libtype::eigen_lib, libtype::magma_lib, libtype::lapack_lib, libtype::mkl_lib }
enum	libtype {   libtype::eigen_lib, libtype::magma_lib, libtype::lapack_lib, libtype::mkl_lib,   libtype::eigen_lib, libtype::magma_lib, libtype::lapack_lib, libtype::mkl_lib }
enum	AllocType {   DEVICE, HOST, PINNED, MAPPED,   N_ALLOC_TYPE }

Functions
std::ostream &	operator<< (std::ostream &output, const CloverFieldParam &param)
double	norm1 (const CloverField &u, bool inverse=false)
double	norm2 (const CloverField &a, bool inverse=false)
void	computeClover (CloverField &clover, const GaugeField &gauge, double coeff, QudaFieldLocation location)
void	copyGenericClover (CloverField &out, const CloverField &in, bool inverse, QudaFieldLocation location, void *Out=0, void *In=0, void *outNorm=0, void *inNorm=0)
	This generic function is used for copying the clover field where in the input and output can be in any order and location. More...
void	cloverInvert (CloverField &clover, bool computeTraceLog, QudaFieldLocation location)
	This function compute the Cholesky decomposition of each clover matrix and stores the clover inverse field. More...
void	cloverRho (CloverField &clover, double rho)
	This function adds a real scalar onto the clover diagonal (only to the direct field not the inverse) More...
void	computeCloverForce (GaugeField &force, const GaugeField &U, std::vector< ColorSpinorField *> &x, std::vector< ColorSpinorField *> &p, std::vector< double > &coeff)
	Compute the force contribution from the solver solution fields. More...
void	computeCloverSigmaOprod (GaugeField &oprod, std::vector< ColorSpinorField *> &x, std::vector< ColorSpinorField *> &p, std::vector< std::vector< double > > &coeff)
	Compute the outer product from the solver solution fields arising from the diagonal term of the fermion bilinear in direction mu,nu and sum to outer product field. More...
void	computeCloverSigmaTrace (GaugeField &output, const CloverField &clover, double coeff)
	Compute the matrix tensor field necessary for the force calculation from the clover trace action. This computes a tensor field [mu,nu]. More...
void	cloverDerivative (cudaGaugeField &force, cudaGaugeField &gauge, cudaGaugeField &oprod, double coeff, QudaParity parity)
	Compute the derivative of the clover matrix in the direction mu,nu and compute the resulting force given the outer-product field. More...
template<typename Float , int Nc, int Ns>
__device__ __host__ Matrix< complex< Float >, Nc >	outerProdSpinTrace (const ColorSpinor< Float, Nc, Ns > &a, const ColorSpinor< Float, Nc, Ns > &b)
template<typename Float , int Nc, int Ns>
__device__ __host__ ColorSpinor< Float, Nc, Ns >	operator+ (const ColorSpinor< Float, Nc, Ns > &x, const ColorSpinor< Float, Nc, Ns > &y)
	ColorSpinor addition operator. More...
template<typename Float , int Nc, int Ns>
__device__ __host__ ColorSpinor< Float, Nc, Ns >	operator- (const ColorSpinor< Float, Nc, Ns > &x, const ColorSpinor< Float, Nc, Ns > &y)
	ColorSpinor subtraction operator. More...
template<typename Float , int Nc, int Ns, typename S >
__device__ __host__ ColorSpinor< Float, Nc, Ns >	operator* (const S &a, const ColorSpinor< Float, Nc, Ns > &x)
	Compute the scalar-vector product y = a * x. More...
template<typename Float , int Nc, int Ns>
__device__ __host__ ColorSpinor< Float, Nc, Ns >	operator* (const Matrix< complex< Float >, Nc > &A, const ColorSpinor< Float, Nc, Ns > &x)
	Compute the matrix-vector product y = A * x. More...
template<typename Float , int Nc, int Ns>
__device__ __host__ ColorSpinor< Float, Nc, Ns >	operator* (const HMatrix< Float, Nc *Ns > &A, const ColorSpinor< Float, Nc, Ns > &x)
	Compute the matrix-vector product y = A * x. More...
void	copyGenericColorSpinor (ColorSpinorField &dst, const ColorSpinorField &src, QudaFieldLocation location, void *Dst=0, void *Src=0, void *dstNorm=0, void *srcNorm=0)
void	genericSource (cpuColorSpinorField &a, QudaSourceType sourceType, int x, int s, int c)
int	genericCompare (const cpuColorSpinorField &a, const cpuColorSpinorField &b, int tol)
void	genericPrintVector (cpuColorSpinorField &a, unsigned int x)
void	wuppertalStep (ColorSpinorField &out, const ColorSpinorField &in, int parity, const GaugeField &U, double A, double B)
void	wuppertalStep (ColorSpinorField &out, const ColorSpinorField &in, int parity, const GaugeField &U, double alpha)
void	exchangeExtendedGhost (cudaColorSpinorField *spinor, int R[], int parity, cudaStream_t *stream_p)
void	copyExtendedColorSpinor (ColorSpinorField &dst, const ColorSpinorField &src, QudaFieldLocation location, const int parity, void *Dst, void *Src, void *dstNorm, void *srcNorm)
void	genericPackGhost (void **ghost, const ColorSpinorField &a, QudaParity parity, int nFace, int dagger, MemoryLocation *destination=nullptr)
	Generic ghost packing routine. More...
void	spinorGauss (ColorSpinorField &src, int seed)
void	spinorGauss (ColorSpinorField &src, RNG &randstates)
template<typename ValueType >
__host__ __device__ ValueType	cos (ValueType x)
template<typename ValueType >
__host__ __device__ ValueType	sin (ValueType x)
template<typename ValueType >
__host__ __device__ ValueType	tan (ValueType x)
template<typename ValueType >
__host__ __device__ ValueType	acos (ValueType x)
template<typename ValueType >
__host__ __device__ ValueType	asin (ValueType x)
template<typename ValueType >
__host__ __device__ ValueType	atan (ValueType x)
template<typename ValueType >
__host__ __device__ ValueType	atan2 (ValueType x, ValueType y)
template<typename ValueType >
__host__ __device__ ValueType	cosh (ValueType x)
template<typename ValueType >
__host__ __device__ ValueType	sinh (ValueType x)
template<typename ValueType >
__host__ __device__ ValueType	tanh (ValueType x)
template<typename ValueType >
__host__ __device__ ValueType	exp (ValueType x)
template<typename ValueType >
__host__ __device__ ValueType	log (ValueType x)
template<typename ValueType >
__host__ __device__ ValueType	log10 (ValueType x)
template<typename ValueType , typename ExponentType >
__host__ __device__ ValueType	pow (ValueType x, ExponentType e)
template<typename ValueType >
__host__ __device__ ValueType	sqrt (ValueType x)
template<typename ValueType >
__host__ __device__ ValueType	abs (ValueType x)
template<typename ValueType >
__host__ __device__ ValueType	conj (ValueType x)
template<typename ValueType >
__host__ __device__ ValueType	abs (const complex< ValueType > &z)
	Returns the magnitude of z. More...
template<typename ValueType >
__host__ __device__ ValueType	arg (const complex< ValueType > &z)
	Returns the phase angle of z. More...
template<typename ValueType >
__host__ __device__ ValueType	norm (const complex< ValueType > &z)
	Returns the magnitude of z squared. More...
template<typename ValueType >
__host__ __device__ complex< ValueType >	conj (const complex< ValueType > &z)
	Returns the complex conjugate of z. More...
template<typename ValueType >
__host__ __device__ complex< ValueType >	polar (const ValueType &m, const ValueType &theta=0)
	Returns the complex with magnitude m and angle theta in radians. More...
template<typename ValueType >
__host__ __device__ complex< ValueType >	operator* (const complex< ValueType > &lhs, const complex< ValueType > &rhs)
template<typename ValueType >
__host__ __device__ complex< ValueType >	operator* (const complex< ValueType > &lhs, const ValueType &rhs)
template<typename ValueType >
__host__ __device__ complex< ValueType >	operator* (const ValueType &lhs, const complex< ValueType > &rhs)
template<typename ValueType >
__host__ __device__ complex< ValueType >	operator/ (const complex< ValueType > &lhs, const complex< ValueType > &rhs)
template<>
__host__ __device__ complex< float >	operator/ (const complex< float > &lhs, const complex< float > &rhs)
template<>
__host__ __device__ complex< double >	operator/ (const complex< double > &lhs, const complex< double > &rhs)
template<typename ValueType >
__host__ __device__ complex< ValueType >	operator+ (const complex< ValueType > &lhs, const complex< ValueType > &rhs)
template<typename ValueType >
__host__ __device__ complex< ValueType >	operator+ (const complex< ValueType > &lhs, const ValueType &rhs)
template<typename ValueType >
__host__ __device__ complex< ValueType >	operator+ (const ValueType &lhs, const complex< ValueType > &rhs)
template<typename ValueType >
__host__ __device__ complex< ValueType >	operator- (const complex< ValueType > &lhs, const complex< ValueType > &rhs)
template<typename ValueType >
__host__ __device__ complex< ValueType >	operator- (const complex< ValueType > &lhs, const ValueType &rhs)
template<typename ValueType >
__host__ __device__ complex< ValueType >	operator- (const ValueType &lhs, const complex< ValueType > &rhs)
template<typename ValueType >
__host__ __device__ complex< ValueType >	operator+ (const complex< ValueType > &rhs)
template<typename ValueType >
__host__ __device__ complex< ValueType >	operator- (const complex< ValueType > &rhs)
template<typename ValueType >
__host__ __device__ complex< ValueType >	cos (const complex< ValueType > &z)
template<typename ValueType >
__host__ __device__ complex< ValueType >	cosh (const complex< ValueType > &z)
template<typename ValueType >
__host__ __device__ complex< ValueType >	exp (const complex< ValueType > &z)
template<typename ValueType >
__host__ __device__ complex< ValueType >	log (const complex< ValueType > &z)
template<typename ValueType >
__host__ __device__ complex< ValueType >	log10 (const complex< ValueType > &z)
template<typename ValueType >
__host__ __device__ complex< ValueType >	pow (const complex< ValueType > &z, const int &n)
template<typename ValueType >
__host__ __device__ complex< ValueType >	pow (const complex< ValueType > &z, const ValueType &x)
template<typename ValueType >
__host__ __device__ complex< ValueType >	pow (const complex< ValueType > &z, const complex< ValueType > &z2)
template<typename ValueType >
__host__ __device__ complex< ValueType >	pow (const ValueType &x, const complex< ValueType > &z)
template<typename ValueType >
__host__ __device__ complex< ValueType >	sin (const complex< ValueType > &z)
template<typename ValueType >
__host__ __device__ complex< ValueType >	sinh (const complex< ValueType > &z)
template<typename ValueType >
__host__ __device__ complex< ValueType >	sqrt (const complex< ValueType > &z)
template<typename ValueType >
__host__ __device__ complex< ValueType >	tan (const complex< ValueType > &z)
template<typename ValueType >
__host__ __device__ complex< ValueType >	tanh (const complex< ValueType > &z)
template<typename ValueType >
__host__ __device__ complex< ValueType >	acos (const complex< ValueType > &z)
template<typename ValueType >
__host__ __device__ complex< ValueType >	asin (const complex< ValueType > &z)
template<typename ValueType >
__host__ __device__ complex< ValueType >	atan (const complex< ValueType > &z)
template<typename ValueType >
__host__ __device__ complex< ValueType >	acosh (const complex< ValueType > &z)
template<typename ValueType >
__host__ __device__ complex< ValueType >	asinh (const complex< ValueType > &z)
template<typename ValueType >
__host__ __device__ complex< ValueType >	atanh (const complex< ValueType > &z)
template<typename ValueType , class charT , class traits >
std::basic_ostream< charT, traits > &	operator<< (std::basic_ostream< charT, traits > &os, const complex< ValueType > &z)
template<typename ValueType , typename charT , class traits >
std::basic_istream< charT, traits > &	operator>> (std::basic_istream< charT, traits > &is, complex< ValueType > &z)
template<typename ValueType >
__host__ __device__ complex< ValueType >	operator+ (const volatile complex< ValueType > &lhs, const volatile complex< ValueType > &rhs)
template<typename ValueType >
__host__ __device__ complex< ValueType >	operator/ (const complex< ValueType > &lhs, const ValueType &rhs)
template<typename ValueType >
__host__ __device__ complex< ValueType >	operator/ (const ValueType &lhs, const complex< ValueType > &rhs)
template<>
__host__ __device__ complex< float >	operator/ (const float &lhs, const complex< float > &rhs)
template<>
__host__ __device__ complex< double >	operator/ (const double &lhs, const complex< double > &rhs)
template<typename ValueType >
__host__ __device__ bool	operator== (const complex< ValueType > &lhs, const complex< ValueType > &rhs)
template<typename ValueType >
__host__ __device__ bool	operator== (const ValueType &lhs, const complex< ValueType > &rhs)
template<typename ValueType >
__host__ __device__ bool	operator== (const complex< ValueType > &lhs, const ValueType &rhs)
template<typename ValueType >
__host__ __device__ bool	operator!= (const complex< ValueType > &lhs, const complex< ValueType > &rhs)
template<typename ValueType >
__host__ __device__ bool	operator!= (const ValueType &lhs, const complex< ValueType > &rhs)
template<typename ValueType >
__host__ __device__ bool	operator!= (const complex< ValueType > &lhs, const ValueType &rhs)
template<>
__host__ __device__ float	abs (const complex< float > &z)
template<>
__host__ __device__ double	abs (const complex< double > &z)
template<>
__host__ __device__ float	arg (const complex< float > &z)
template<>
__host__ __device__ double	arg (const complex< double > &z)
template<>
__host__ __device__ complex< float >	polar (const float &magnitude, const float &angle)
template<>
__host__ __device__ complex< double >	polar (const double &magnitude, const double &angle)
template<>
__host__ __device__ complex< float >	cos (const complex< float > &z)
template<>
__host__ __device__ complex< float >	cosh (const complex< float > &z)
template<>
__host__ __device__ complex< float >	exp (const complex< float > &z)
template<>
__host__ __device__ complex< float >	log (const complex< float > &z)
template<>
__host__ __device__ complex< float >	pow (const float &x, const complex< float > &exponent)
template<>
__host__ __device__ complex< float >	sin (const complex< float > &z)
template<>
__host__ __device__ complex< float >	sinh (const complex< float > &z)
template<typename ValueType >
__host__ __device__ complex< float >	sqrt (const complex< float > &z)
template<typename ValueType >
__host__ __device__ complex< float >	atanh (const complex< float > &z)
void	contractCuda (const cudaColorSpinorField &x, const cudaColorSpinorField &y, void *result, const QudaContractType contract_type, const QudaParity parity, TimeProfile &profile)
void	contractCuda (const cudaColorSpinorField &x, const cudaColorSpinorField &y, void *result, const QudaContractType contract_type, const int tSlice, const QudaParity parity, TimeProfile &profile)
void	covDev (cudaColorSpinorField *out, cudaGaugeField &gauge, const cudaColorSpinorField *in, const int parity, const int mu, TimeProfile &profile)
void	ApplyCovDev (ColorSpinorField &out, const ColorSpinorField &in, const GaugeField &U, int parity, int mu)
	Driver for applying the covariant derivative. More...
template<typename scalar , int n>
__device__ __host__ void	zero (vector_type< scalar, n > &v)
template<typename scalar , int n>
__device__ __host__ vector_type< scalar, n >	operator+ (const vector_type< scalar, n > &a, const vector_type< scalar, n > &b)
template<int block_size_x, int block_size_y, typename T >
__device__ void	reduce2d (ReduceArg< T > arg, const T &in, const int idx=0)
template<int block_size, typename T >
__device__ void	reduce (ReduceArg< T > arg, const T &in, const int idx=0)
template<int block_size_x, int block_size_y, typename T >
__device__ void	reduceRow (ReduceArg< T > arg, const T &in)
void	setDiracParam (DiracParam &diracParam, QudaInvertParam *inv_param, bool pc)
void	setDiracSloppyParam (DiracParam &diracParam, QudaInvertParam *inv_param, bool pc)
void	setKernelPackT (bool pack)
bool	getKernelPackT ()
void	setPackComms (const int *commDim)
bool	getDslashLaunch ()
void	createDslashEvents ()
void	destroyDslashEvents ()
void	wilsonDslashCuda (cudaColorSpinorField *out, const cudaGaugeField &gauge, const cudaColorSpinorField *in, const int oddBit, const int daggerBit, const cudaColorSpinorField *x, const double &k, const int *commDim, TimeProfile &profile)
void	cloverDslashCuda (cudaColorSpinorField *out, const cudaGaugeField &gauge, const FullClover &cloverInv, const cudaColorSpinorField *in, const int oddBit, const int daggerBit, const cudaColorSpinorField *x, const double &k, const int *commDim, TimeProfile &profile)
void	asymCloverDslashCuda (cudaColorSpinorField *out, const cudaGaugeField &gauge, const FullClover &cloverInv, const cudaColorSpinorField *in, const int oddBit, const int daggerBit, const cudaColorSpinorField *x, const double &k, const int *commDim, TimeProfile &profile)
void	ApplyClover (ColorSpinorField &out, const ColorSpinorField &in, const CloverField &clover, bool inverse, int parity)
	Apply clover-matrix field to a color-spinor field. More...
void	domainWallDslashCuda (cudaColorSpinorField *out, const cudaGaugeField &gauge, const cudaColorSpinorField *in, const int parity, const int dagger, const cudaColorSpinorField *x, const double &m_f, const double &k, const int *commDim, TimeProfile &profile)
void	domainWallDslashCuda (cudaColorSpinorField *out, const cudaGaugeField &gauge, const cudaColorSpinorField *in, const int parity, const int dagger, const cudaColorSpinorField *x, const double &m_f, const double &a, const double &b, const int *commDim, const int DS_type, TimeProfile &profile)
void	MDWFDslashCuda (cudaColorSpinorField *out, const cudaGaugeField &gauge, const cudaColorSpinorField *in, const int parity, const int dagger, const cudaColorSpinorField *x, const double &m_f, const double &k, const double *b5, const double *c_5, const double &m5, const int *commDim, const int DS_type, TimeProfile &profile)
void	staggeredDslashCuda (cudaColorSpinorField *out, const cudaGaugeField &gauge, const cudaColorSpinorField *in, const int parity, const int dagger, const cudaColorSpinorField *x, const double &k, const int *commDim, TimeProfile &profile)
void	improvedStaggeredDslashCuda (cudaColorSpinorField *out, const cudaGaugeField &fatGauge, const cudaGaugeField &longGauge, const cudaColorSpinorField *in, const int parity, const int dagger, const cudaColorSpinorField *x, const double &k, const int *commDim, TimeProfile &profile)
void	twistedMassDslashCuda (cudaColorSpinorField *out, const cudaGaugeField &gauge, const cudaColorSpinorField *in, const int parity, const int dagger, const cudaColorSpinorField *x, const QudaTwistDslashType type, const double &kappa, const double &mu, const double &epsilon, const double &k, const int *commDim, TimeProfile &profile)
void	ndegTwistedMassDslashCuda (cudaColorSpinorField *out, const cudaGaugeField &gauge, const cudaColorSpinorField *in, const int parity, const int dagger, const cudaColorSpinorField *x, const QudaTwistDslashType type, const double &kappa, const double &mu, const double &epsilon, const double &k, const int *commDim, TimeProfile &profile)
void	twistedCloverDslashCuda (cudaColorSpinorField *out, const cudaGaugeField &gauge, const FullClover *clover, const FullClover *cloverInv, const cudaColorSpinorField *in, const int parity, const int dagger, const cudaColorSpinorField *x, const QudaTwistCloverDslashType type, const double &kappa, const double &mu, const double &epsilon, const double &k, const int *commDim, TimeProfile &profile)
void	ApplyTwistGamma (ColorSpinorField &out, const ColorSpinorField &in, int d, double kappa, double mu, double epsilon, int dagger, QudaTwistGamma5Type type)
	Apply the twisted-mass gamma operator to a color-spinor field. More...
void	ApplyTwistClover (ColorSpinorField &out, const ColorSpinorField &in, const CloverField &clover, double kappa, double mu, double epsilon, int parity, int dagger, QudaTwistGamma5Type twist)
	Apply twisted clover-matrix field to a color-spinor field. More...
void	packFace (void *ghost_buf[2 *QUDA_MAX_DIM], cudaColorSpinorField &in, MemoryLocation location, const int nFace, const int dagger, const int parity, const int dim, const int face_num, const cudaStream_t &stream, const double a=0.0, const double b=0.0)
	Dslash face packing routine. More...
void	packFaceExtended (void *ghost_buf[2 *QUDA_MAX_DIM], cudaColorSpinorField &field, MemoryLocation location, const int nFace, const int R[], const int dagger, const int parity, const int dim, const int face_num, const cudaStream_t &stream, const bool unpack=false)
void	gamma5 (ColorSpinorField &out, const ColorSpinorField &in)
	Applies a gamma5 matrix to a spinor (wrapper to ApplyGamma) More...
__device__ __host__ void	zero (double &a)
__device__ __host__ void	zero (double2 &a)
__device__ __host__ void	zero (double3 &a)
__device__ __host__ void	zero (double4 &a)
__device__ __host__ void	zero (float &a)
__device__ __host__ void	zero (float2 &a)
__device__ __host__ void	zero (float3 &a)
__device__ __host__ void	zero (float4 &a)
__host__ __device__ double2	operator+ (const double2 &x, const double2 &y)
__host__ __device__ double2	operator- (const double2 &x, const double2 &y)
__host__ __device__ float2	operator- (const float2 &x, const float2 &y)
__host__ __device__ float4	operator- (const float4 &x, const float4 &y)
__host__ __device__ double3	operator+ (const double3 &x, const double3 &y)
__host__ __device__ double4	operator+ (const double4 &x, const double4 &y)
__host__ __device__ float4	operator* (const float a, const float4 x)
__host__ __device__ float2	operator* (const float a, const float2 x)
__host__ __device__ double2	operator* (const double a, const double2 x)
__host__ __device__ double4	operator* (const double a, const double4 x)
__host__ __device__ float2	operator+ (const float2 x, const float2 y)
__host__ __device__ float4	operator+ (const float4 x, const float4 y)
__host__ __device__ float4	operator+= (float4 &x, const float4 y)
__host__ __device__ float2	operator+= (float2 &x, const float2 y)
__host__ __device__ double2	operator+= (double2 &x, const double2 y)
__host__ __device__ double3	operator+= (double3 &x, const double3 y)
__host__ __device__ double4	operator+= (double4 &x, const double4 y)
__host__ __device__ float4	operator-= (float4 &x, const float4 y)
__host__ __device__ float2	operator-= (float2 &x, const float2 y)
__host__ __device__ double2	operator-= (double2 &x, const double2 y)
__host__ __device__ float2	operator*= (float2 &x, const float a)
__host__ __device__ double2	operator*= (double2 &x, const float a)
__host__ __device__ float4	operator*= (float4 &a, const float &b)
__host__ __device__ double2	operator*= (double2 &a, const double &b)
__host__ __device__ double4	operator*= (double4 &a, const double &b)
__host__ __device__ float2	operator- (const float2 &x)
__host__ __device__ double2	operator- (const double2 &x)
__forceinline__ __host__ __device__ float	max_fabs (const float4 &c)
__forceinline__ __host__ __device__ float	max_fabs (const float2 &b)
__forceinline__ __host__ __device__ double	max_fabs (const double4 &c)
__forceinline__ __host__ __device__ double	max_fabs (const double2 &b)
__forceinline__ __host__ __device__ float2	make_FloatN (const double2 &a)
__forceinline__ __host__ __device__ float4	make_FloatN (const double4 &a)
__forceinline__ __host__ __device__ double2	make_FloatN (const float2 &a)
__forceinline__ __host__ __device__ double4	make_FloatN (const float4 &a)
__forceinline__ __host__ __device__ short4	make_shortN (const float4 &a)
__forceinline__ __host__ __device__ short2	make_shortN (const float2 &a)
__forceinline__ __host__ __device__ short4	make_shortN (const double4 &a)
__forceinline__ __host__ __device__ short2	make_shortN (const double2 &a)
template<typename Float2 , typename Complex >
Float2	make_Float2 (const Complex &a)
template<>
double2	make_Float2 (const complex< double > &a)
template<>
double2	make_Float2 (const complex< float > &a)
template<>
float2	make_Float2 (const complex< double > &a)
template<>
float2	make_Float2 (const complex< float > &a)
template<>
double2	make_Float2 (const std::complex< double > &a)
template<>
double2	make_Float2 (const std::complex< float > &a)
template<>
float2	make_Float2 (const std::complex< double > &a)
template<>
float2	make_Float2 (const std::complex< float > &a)
complex< double >	make_Complex (const double2 &a)
complex< float >	make_Complex (const float2 &a)
std::ostream &	operator<< (std::ostream &output, const GaugeFieldParam &param)
double	norm1 (const GaugeField &u)
	This is a debugging function, where we cast a gauge field into a spinor field so we can compute its L1 norm. More...
double	norm2 (const GaugeField &u)
	This is a debugging function, where we cast a gauge field into a spinor field so we can compute its L2 norm. More...
void	ax (const double &a, GaugeField &u)
	Scale the gauge field by the scalar a. More...
void	copyGenericGauge (GaugeField &out, const GaugeField &in, QudaFieldLocation location, void *Out=0, void *In=0, void **ghostOut=0, void **ghostIn=0, int type=0)
void	copyExtendedGauge (GaugeField &out, const GaugeField &in, QudaFieldLocation location, void *Out=0, void *In=0)
void	extractGaugeGhost (const GaugeField &u, void **ghost, bool extract=true, int offset=0)
void	extractExtendedGaugeGhost (const GaugeField &u, int dim, const int *R, void **ghost, bool extract)
double	maxGauge (const GaugeField &u)
void	applyGaugePhase (GaugeField &u)
uint64_t	Checksum (const GaugeField &u, bool mini=false)
void	gaugeForce (GaugeField &mom, const GaugeField &u, double coeff, int ***input_path, int *length, double *path_coeff, int num_paths, int max_length)
	Compute the gauge-force contribution to the momentum. More...
double3	plaquette (const GaugeField &U, QudaFieldLocation location)
void	gaugeGauss (GaugeField &dataDs, RNG &rngstate)
void	APEStep (GaugeField &dataDs, const GaugeField &dataOr, double alpha)
void	STOUTStep (GaugeField &dataDs, const GaugeField &dataOr, double rho)
void	OvrImpSTOUTStep (GaugeField &dataDs, const GaugeField &dataOr, double rho, double epsilon)
void	gaugefixingOVR (cudaGaugeField &data, const int gauge_dir, const int Nsteps, const int verbose_interval, const double relax_boost, const double tolerance, const int reunit_interval, const int stopWtheta)
	Gauge fixing with overrelaxation with support for single and multi GPU. More...
void	gaugefixingFFT (cudaGaugeField &data, const int gauge_dir, const int Nsteps, const int verbose_interval, const double alpha, const int autotune, const double tolerance, const int stopWtheta)
	Gauge fixing with Steepest descent method with FFTs with support for single GPU only. More...
void	computeFmunu (GaugeField &Fmunu, const GaugeField &gauge, QudaFieldLocation location)
double	computeQCharge (GaugeField &Fmunu, QudaFieldLocation location)
void	updateGaugeField (GaugeField &out, double dt, const GaugeField &in, const GaugeField &mom, bool conj_mom, bool exact)
template<typename I , typename J , typename K >
static __device__ __host__ int	linkIndexShift (const I x[], const J dx[], const K X[4])
template<typename I , typename J , typename K >
static __device__ __host__ int	linkIndexShift (I y[], const I x[], const J dx[], const K X[4])
template<typename I >
static __device__ __host__ int	linkIndex (const int x[], const I X[4])
template<typename I >
static __device__ __host__ int	linkIndex (int y[], const int x[], const I X[4])
template<typename I >
static __device__ __host__ int	linkIndexM1 (const int x[], const I X[4], const int mu)
template<typename I >
static __device__ __host__ int	linkNormalIndexP1 (const int x[], const I X[4], const int mu)
template<typename I >
static __device__ __host__ int	linkIndexP1 (const int x[], const I X[4], const int mu)
template<typename I >
static __device__ __host__ void	getCoords (int x[], int cb_index, const I X[], int parity)
template<typename I , typename J >
static __device__ __host__ void	getCoordsExtended (I x[], int cb_index, const J X[], int parity, const int R[])
template<typename I >
static __device__ __host__ void	getCoords5 (int x[5], int cb_index, const I X[5], int parity, QudaDWFPCType pc_type)
template<typename I >
static __device__ __host__ int	getIndexFull (int cb_index, const I X[4], int parity)
template<int dir, typename I >
__device__ __host__ int	ghostFaceIndex (const int x[], const I X[], int dim, int nFace)
__device__ void	load_streaming_double2 (double2 &a, const double2 *addr)
__device__ void	load_streaming_float4 (float4 &a, const float4 *addr)
__device__ void	load_global_float4 (float4 &a, const float4 *addr)
__device__ void	store_streaming_float4 (float4 *addr, float x, float y, float z, float w)
__device__ void	store_streaming_short4 (short4 *addr, short x, short y, short z, short w)
__device__ void	store_streaming_double2 (double2 *addr, double x, double y)
__device__ void	store_streaming_float2 (float2 *addr, float x, float y)
__device__ void	store_streaming_short2 (short2 *addr, short x, short y)
void	completeKSForce (GaugeField &mom, const GaugeField &oprod, const GaugeField &gauge, QudaFieldLocation location, long long *flops=NULL)
std::ostream &	operator<< (std::ostream &output, const LatticeFieldParam &param)
QudaFieldLocation	Location_ (const char *func, const char *file, int line, const LatticeField &a, const LatticeField &b)
	Helper function for determining if the location of the fields is the same. More...
template<typename... Args>
QudaFieldLocation	Location_ (const char *func, const char *file, int line, const LatticeField &a, const LatticeField &b, const Args &... args)
	Helper function for determining if the location of the fields is the same. More...
QudaPrecision	Precision_ (const char *func, const char *file, int line, const LatticeField &a, const LatticeField &b)
	Helper function for determining if the precision of the fields is the same. More...
template<typename... Args>
QudaPrecision	Precision_ (const char *func, const char *file, int line, const LatticeField &a, const LatticeField &b, const Args &... args)
	Helper function for determining if the precision of the fields is the same. More...
QudaFieldLocation	reorder_location ()
	Return whether data is reordered on the CPU or GPU. This can set at QUDA initialization using the environment variable QUDA_REORDER_LOCATION. More...
void	reorder_location_set (QudaFieldLocation reorder_location_)
	Set whether data is reorderd on the CPU or GPU. This can set at QUDA initialization using the environment variable QUDA_REORDER_LOCATION. More...
void	fatLongKSLink (cudaGaugeField *fat, cudaGaugeField *lng, const cudaGaugeField &gauge, const double *coeff)
	Compute the fat and long links for an improved staggered (Kogut-Susskind) fermions. More...
void	printPeakMemUsage ()
void	assertAllMemFree ()
long	device_allocated_peak ()
long	pinned_allocated_peak ()
long	mapped_allocated_peak ()
long	host_allocated_peak ()
void *	device_malloc_ (const char *func, const char *file, int line, size_t size)
void *	device_pinned_malloc_ (const char *func, const char *file, int line, size_t size)
void *	safe_malloc_ (const char *func, const char *file, int line, size_t size)
void *	pinned_malloc_ (const char *func, const char *file, int line, size_t size)
void *	mapped_malloc_ (const char *func, const char *file, int line, size_t size)
void	device_free_ (const char *func, const char *file, int line, void *ptr)
void	device_pinned_free_ (const char *func, const char *file, int line, void *ptr)
void	host_free_ (const char *func, const char *file, int line, void *ptr)
constexpr const char *	str_end (const char *str)
constexpr bool	str_slant (const char *str)
constexpr const char *	r_slant (const char *str)
constexpr const char *	file_name (const char *str)
double	computeMomAction (const GaugeField &mom)
	Compute and return global the momentum action 1/2 mom^2. More...
void	updateMomentum (GaugeField &mom, double coeff, GaugeField &force)
void	applyU (GaugeField &force, GaugeField &U)
void	ApplyCoarse (ColorSpinorField &out, const ColorSpinorField &inA, const ColorSpinorField &inB, const GaugeField &Y, const GaugeField &X, double kappa, int parity=QUDA_INVALID_PARITY, bool dslash=true, bool clover=true, bool dagger=false)
void	CoarseOp (GaugeField &Y, GaugeField &X, GaugeField &Xinv, GaugeField &Yhat, const Transfer &T, const cudaGaugeField &gauge, const cudaCloverField *clover, double kappa, double mu, double mu_factor, QudaDiracType dirac, QudaMatPCType matpc)
	Coarse operator construction from a fine-grid operator (Wilson / Clover) More...
void	CoarseCoarseOp (GaugeField &Y, GaugeField &X, GaugeField &Xinv, GaugeField &Yhat, const Transfer &T, const GaugeField &gauge, const GaugeField &clover, const GaugeField &cloverInv, double kappa, double mu, double mu_factor, QudaDiracType dirac, QudaMatPCType matpc)
	Coarse operator construction from an intermediate-grid operator (Coarse) More...
void	Monte (cudaGaugeField &data, RNG &rngstate, double Beta, int nhb, int nover)
	Perform heatbath and overrelaxation. Performs nhb heatbath steps followed by nover overrelaxation steps. More...
void	InitGaugeField (cudaGaugeField &data)
	Perform a cold start to the gauge field, identity SU(3) matrix, also fills the ghost links in multi-GPU case (no need to exchange data) More...
void	InitGaugeField (cudaGaugeField &data, RNG &rngstate)
	Perform a hot start to the gauge field, random SU(3) matrix, followed by reunitarization, also exchange borders links in multi-GPU case. More...
void	PGaugeExchange (cudaGaugeField &data, const int dir, const int parity)
	Perform heatbath and overrelaxation. Performs nhb heatbath steps followed by nover overrelaxation steps. More...
void	PGaugeExchangeFree ()
	Release all allocated memory used to exchange data between nodes. More...
double2	getLinkDeterminant (cudaGaugeField &data)
	Calculate the Determinant. More...
double2	getLinkTrace (cudaGaugeField &data)
	Calculate the Trace. More...
void	arpackSolve (std::vector< ColorSpinorField *> &B, void *evals, DiracMatrix &matEigen, QudaPrecision matPrec, QudaPrecision arpackPrec, double tol, int nev, int ncv, char *target)
void	qudaMemcpy_ (void *dst, const void *src, size_t count, cudaMemcpyKind kind, const char *func, const char *file, const char *line)
	Wrapper around cudaMemcpy used for auto-profiling. Do not call directly, rather call macro below which will grab the location of the call. More...
void	qudaMemcpyAsync_ (void *dst, const void *src, size_t count, cudaMemcpyKind kind, const cudaStream_t &stream, const char *func, const char *file, const char *line)
	Wrapper around cudaMemcpyAsync or driver API equivalent Potentially add auto-profiling support. More...
void	qudaMemcpy2DAsync_ (void *dst, size_t dpitch, const void *src, size_t spitch, size_t width, size_t hieght, cudaMemcpyKind kind, const cudaStream_t &stream, const char *func, const char *file, const char *line)
	Wrapper around cudaMemcpy2DAsync or driver API equivalent Potentially add auto-profiling support. More...
cudaError_t	qudaLaunchKernel (const void *func, dim3 gridDim, dim3 blockDim, void **args, size_t sharedMem, cudaStream_t stream)
	Wrapper around cudaLaunchKernel. More...
cudaError_t	qudaEventQuery (cudaEvent_t &event)
	Wrapper around cudaEventQuery or cuEventQuery. More...
cudaError_t	qudaEventRecord (cudaEvent_t &event, cudaStream_t stream=0)
	Wrapper around cudaEventRecord or cuEventRecord. More...
cudaError_t	qudaStreamWaitEvent (cudaStream_t stream, cudaEvent_t event, unsigned int flags)
	Wrapper around cudaEventRecord or cuEventRecord. More...
cudaError_t	qudaStreamSynchronize (cudaStream_t &stream)
	Wrapper around cudaStreamSynchronize or cuStreamSynchronize. More...
cudaError_t	qudaEventSynchronize (cudaEvent_t &event)
	Wrapper around cudaEventSynchronize or cuEventSynchronize. More...
cudaError_t	qudaDeviceSynchronize ()
	Wrapper around cudaDeviceSynchronize or cuDeviceSynchronize. More...
void	printAPIProfile ()
	Print out the timer profile for CUDA API calls. More...
bool	canReuseResidentGauge (QudaInvertParam *inv_param)
template<class Real >
__device__ Real	Random (cuRNGState &state, Real a, Real b)
	Return a random number between a and b. More...
template<>
__device__ float	Random< float > (cuRNGState &state, float a, float b)
template<>
__device__ double	Random< double > (cuRNGState &state, double a, double b)
template<class Real >
__device__ Real	Random (cuRNGState &state)
	Return a random number between 0 and 1. More...
template<>
__device__ float	Random< float > (cuRNGState &state)
template<>
__device__ double	Random< double > (cuRNGState &state)
template<typename T1 , typename T2 >
__host__ __device__ void	copy (T1 &a, const T2 &b)
template<>
__host__ __device__ void	copy (double &a, const int2 &b)
template<>
__host__ __device__ void	copy (double2 &a, const int4 &b)
static __host__ __device__ float	s2f (const short &a)
static __host__ __device__ double	s2d (const short &a)
__device__ __host__ int	f2i (float f)
__device__ __host__ int	d2i (double d)
template<>
__host__ __device__ void	copy (float &a, const short &b)
template<>
__host__ __device__ void	copy (short &a, const float &b)
template<>
__host__ __device__ void	copy (float2 &a, const short2 &b)
template<>
__host__ __device__ void	copy (short2 &a, const float2 &b)
template<>
__host__ __device__ void	copy (float4 &a, const short4 &b)
template<>
__host__ __device__ void	copy (short4 &a, const float4 &b)
template<typename VectorType >
__device__ __host__ VectorType	vector_load (void *ptr, int idx)
template<typename VectorType >
__device__ __host__ void	vector_store (void *ptr, int idx, const VectorType &value)
template<>
__device__ __host__ void	vector_store (void *ptr, int idx, const double2 &value)
template<>
__device__ __host__ void	vector_store (void *ptr, int idx, const float4 &value)
template<>
__device__ __host__ void	vector_store (void *ptr, int idx, const float2 &value)
template<>
__device__ __host__ void	vector_store (void *ptr, int idx, const short4 &value)
template<>
__device__ __host__ void	vector_store (void *ptr, int idx, const short2 &value)
void	computeStaggeredOprod (GaugeField *out[], ColorSpinorField &in, const double coeff[], int nFace)
	Compute the outer-product field between the staggered quark field's one and (for HISQ and ASQTAD) three hop sites. E.g.,. More...
void	ApplyLaplace (ColorSpinorField &out, const ColorSpinorField &in, const GaugeField &U, double kappa, const ColorSpinorField *x, int parity)
	Driver for applying the Laplace stencil. More...
template<typename Float2 , typename Float >
__host__ __device__ int	checkUnitary (Matrix< Float2, 3 > &inv, Matrix< Float2, 3 > in, const Float tol)
	Check the unitarity of the input matrix to a given tolerance. More...
template<typename Float2 >
__host__ __device__ int	checkUnitaryPrint (Matrix< Float2, 3 > &inv, Matrix< Float2, 3 > in)
	Check the unitarity of the input matrix to a given tolerance (1e-14) and print out deviation for each component (used for debugging only). More...
template<typename Float >
__host__ __device__ void	polarSu3 (Matrix< complex< Float >, 3 > &in, Float tol)
	Project the input matrix on the SU(3) group. First unitarize the matrix and then project onto the special unitary group. More...
void	FillV (ColorSpinorField &V, const std::vector< ColorSpinorField *> &B, int Nvec)
void	BlockOrthogonalize (ColorSpinorField &V, int Nvec, const int *geo_bs, const int *fine_to_coarse, int spin_bs)
	Block orthogonnalize the matrix field, where the blocks are defined by lookup tables that map the fine grid points to the coarse grid points, and similarly for the spin degrees of freedom. More...
void	Prolongate (ColorSpinorField &out, const ColorSpinorField &in, const ColorSpinorField &v, int Nvec, const int *fine_to_coarse, const int *spin_map, int parity=QUDA_INVALID_PARITY)
	Apply the prolongation operator. More...
void	Restrict (ColorSpinorField &out, const ColorSpinorField &in, const ColorSpinorField &v, int Nvec, const int *fine_to_coarse, const int *coarse_to_fine, const int *spin_map, int parity=QUDA_INVALID_PARITY)
	Apply the restriction operator. More...
bool	activeTuning ()
	query if tuning is in progress More...
void	loadTuneCache ()
void	saveTuneCache ()
void	saveProfile (const std::string label="")
	Save profile to disk. More...
void	flushProfile ()
	Flush profile contents, setting all counts to zero. More...
TuneParam &	tuneLaunch (Tunable &tunable, QudaTune enabled, QudaVerbosity verbosity)
void	u32toa (char *buffer, uint32_t value)
void	i32toa (char *buffer, int32_t value)
void	u64toa (char *buffer, uint64_t value)
void	i64toa (char *buffer, int64_t value)
void	setUnitarizeLinksConstants (double unitarize_eps, double max_error, bool allow_svd, bool svd_only, double svd_rel_error, double svd_abs_error)
void	unitarizeLinksCPU (cpuGaugeField &outfield, const cpuGaugeField &infield)
void	unitarizeLinks (cudaGaugeField &outfield, const cudaGaugeField &infield, int *fails)
void	unitarizeLinks (cudaGaugeField &outfield, int *fails)
bool	isUnitary (const cpuGaugeField &field, double max_error)
void	projectSU3 (cudaGaugeField &U, double tol, int *fails)
	Project the input gauge field onto the SU(3) group. This is a destructive operation. The number of link failures is reported so appropriate action can be taken. More...
template<typename Arg >
__device__ __host__ uint64_t	siteChecksum (const Arg &arg, int d, int parity, int x_cb)
template<typename Arg >
uint64_t	ChecksumCPU (const Arg &arg)
template<typename real , typename Link >
__device__ void	axpy (real a, const real *x, Link &y)
template<typename real , typename Link >
__device__ void	operator+= (real *y, const Link &x)
template<typename real , typename Link >
__device__ void	operator-= (real *y, const Link &x)
ColorSpinorParam	colorSpinorParam (const CloverField &a, bool inverse)
template<bool from_coarse, typename Float , int dim, QudaDirection dir, int fineSpin, int fineColor, int coarseSpin, int coarseColor, typename Arg >
__device__ __host__ void	computeUV (Arg &arg, int parity, int x_cb, int ic_c)
template<bool from_coarse, typename Float , int dim, QudaDirection dir, int fineSpin, int fineColor, int coarseSpin, int coarseColor, typename Arg >
void	ComputeUVCPU (Arg &arg)
template<bool from_coarse, typename Float , int dim, QudaDirection dir, int fineSpin, int fineColor, int coarseSpin, int coarseColor, typename Arg >
__global__ void	ComputeUVGPU (Arg arg)
template<typename Float , int fineSpin, int fineColor, int coarseColor, typename Arg >
__device__ __host__ void	computeAV (Arg &arg, int parity, int x_cb, int ic_c)
template<typename Float , int fineSpin, int fineColor, int coarseColor, typename Arg >
void	ComputeAVCPU (Arg &arg)
template<typename Float , int fineSpin, int fineColor, int coarseColor, typename Arg >
__global__ void	ComputeAVGPU (Arg arg)
template<typename Float , int fineSpin, int fineColor, int coarseColor, typename Arg >
__device__ __host__ void	computeTMAV (Arg &arg, int parity, int x_cb, int v)
template<typename Float , int fineSpin, int fineColor, int coarseColor, typename Arg >
void	ComputeTMAVCPU (Arg &arg)
template<typename Float , int fineSpin, int fineColor, int coarseColor, typename Arg >
__global__ void	ComputeTMAVGPU (Arg arg)
template<typename Float , int fineSpin, int fineColor, int coarseColor, typename Arg >
__device__ __host__ void	computeTMCAV (Arg &arg, int parity, int x_cb)
template<typename Float , int fineSpin, int fineColor, int coarseColor, typename Arg >
void	ComputeTMCAVCPU (Arg &arg)
template<typename Float , int fineSpin, int fineColor, int coarseColor, typename Arg >
__global__ void	ComputeTMCAVGPU (Arg arg)
template<bool from_coarse, typename Float , int dim, QudaDirection dir, int fineSpin, int fineColor, int coarseSpin, int coarseColor, typename Arg >
__device__ __host__ void	multiplyVUV (complex< Float > vuv[], Arg &arg, int parity, int x_cb, int ic_c)
	Do a single (AV)^ * UV product, where for preconditioned clover, AV correspond to the clover inverse multiplied by the packed null space vectors, else AV is simply the packed null space vectors. More...
template<bool from_coarse, typename Float , int dim, QudaDirection dir, int fineSpin, int fineColor, int coarseSpin, int coarseColor, typename Arg >
__device__ __host__ void	computeVUV (Arg &arg, int parity, int x_cb, int c_row)
template<bool from_coarse, typename Float , int dim, QudaDirection dir, int fineSpin, int fineColor, int coarseSpin, int coarseColor, typename Arg >
void	ComputeVUVCPU (Arg arg)
template<bool from_coarse, typename Float , int dim, QudaDirection dir, int fineSpin, int fineColor, int coarseSpin, int coarseColor, typename Arg >
__global__ void	ComputeVUVGPU (Arg arg)
template<typename Float , int nSpin, int nColor, typename Arg >
__device__ __host__ void	computeYreverse (Arg &arg, int parity, int x_cb)
template<typename Float , int nSpin, int nColor, typename Arg >
void	ComputeYReverseCPU (Arg &arg)
template<typename Float , int nSpin, int nColor, typename Arg >
__global__ void	ComputeYReverseGPU (Arg arg)
template<bool bidirectional, typename Float , int nSpin, int nColor, typename Arg >
__device__ __host__ void	computeCoarseLocal (Arg &arg, int parity, int x_cb)
template<bool bidirectional, typename Float , int nSpin, int nColor, typename Arg >
void	ComputeCoarseLocalCPU (Arg &arg)
template<bool bidirectional, typename Float , int nSpin, int nColor, typename Arg >
__global__ void	ComputeCoarseLocalGPU (Arg arg)
template<bool from_coarse, typename Float , int fineSpin, int coarseSpin, int fineColor, int coarseColor, typename Arg >
__device__ __host__ void	computeCoarseClover (Arg &arg, int parity, int x_cb, int ic_c)
template<bool from_coarse, typename Float , int fineSpin, int coarseSpin, int fineColor, int coarseColor, typename Arg >
void	ComputeCoarseCloverCPU (Arg &arg)
template<bool from_coarse, typename Float , int fineSpin, int coarseSpin, int fineColor, int coarseColor, typename Arg >
__global__ void	ComputeCoarseCloverGPU (Arg arg)
template<typename Float , int nSpin, int nColor, typename Arg >
void	AddCoarseDiagonalCPU (Arg &arg)
template<typename Float , int nSpin, int nColor, typename Arg >
__global__ void	AddCoarseDiagonalGPU (Arg arg)
template<typename Float , int nSpin, int nColor, typename Arg >
void	AddCoarseTmDiagonalCPU (Arg &arg)
template<typename Float , int nSpin, int nColor, typename Arg >
__global__ void	AddCoarseTmDiagonalGPU (Arg arg)
template<typename Float , int n, typename Arg >
__device__ __host__ void	computeYhat (Arg &arg, int d, int x_cb, int parity, int i)
template<typename Float , int n, typename Arg >
void	CalculateYhatCPU (Arg &arg)
template<typename Float , int n, typename Arg >
__global__ void	CalculateYhatGPU (Arg arg)
template<bool from_coarse, typename Float , int fineSpin, int fineColor, int coarseSpin, int coarseColor, QudaGaugeFieldOrder gOrder, typename F , typename Ftmp , typename coarseGauge , typename fineGauge , typename fineClover >
void	calculateY (coarseGauge &Y, coarseGauge &X, coarseGauge &Xinv, Ftmp &UV, F &AV, F &V, fineGauge &G, fineClover &C, fineClover &Cinv, GaugeField &Y_, GaugeField &X_, GaugeField &Xinv_, GaugeField &Yhat_, ColorSpinorField &av, const ColorSpinorField &v, double kappa, double mu, double mu_factor, QudaDiracType dirac, QudaMatPCType matpc)
	Calculate the coarse-link field, include the clover field, and its inverse, and finally also compute the preconditioned coarse link field. More...
std::ostream &	operator<< (std::ostream &out, const ColorSpinorField &a)
template<typename Float , int Ns, int Ms, int Nc, int Mc, int nDim, typename Arg >
__device__ __host__ void	packGhost (Arg &arg, int cb_idx, int parity, int spinor_parity, int spin_block, int color_block)
template<typename Float , int Ns, int Ms, int Nc, int Mc, int nDim, typename Arg >
void	GenericPackGhost (Arg &arg)
template<typename Float , int Ns, int Ms, int Nc, int Mc, int nDim, typename Arg >
__global__ void	GenericPackGhostKernel (Arg arg)
template<class T >
void	random (T &t)
template<class T >
void	point (T &t, int x, int s, int c)
template<class T >
void	constant (T &t, int k, int s, int c)
template<class P >
void	sin (P &p, int d, int n, int offset)
template<class U , class V >
int	compareSpinor (const U &u, const V &v, const int tol)
template<class Order >
void	print_vector (const Order &o, unsigned int x)
template<typename Float , int Nc, typename Vector , typename Arg >
__device__ __host__ void	computeNeighborSum (Vector &out, Arg &arg, int x_cb, int parity)
template<typename Float , int Ns, int Nc, typename Arg >
__device__ __host__ void	computeWupperalStep (Arg &arg, int x_cb, int parity)
template<typename Float , int Ns, int Nc, typename Arg >
void	wuppertalStepCPU (Arg arg)
template<typename Float , int Ns, int Nc, typename Arg >
__global__ void	wuppertalStepGPU (Arg arg)
void	copyGenericColorSpinorDD (ColorSpinorField &, const ColorSpinorField &, QudaFieldLocation, void *, void *, void *a=0, void *b=0)
void	copyGenericColorSpinorDS (ColorSpinorField &, const ColorSpinorField &, QudaFieldLocation, void *, void *, void *a=0, void *b=0)
void	copyGenericColorSpinorDH (ColorSpinorField &, const ColorSpinorField &, QudaFieldLocation, void *, void *, void *a=0, void *b=0)
void	copyGenericColorSpinorSD (ColorSpinorField &, const ColorSpinorField &, QudaFieldLocation, void *, void *, void *a=0, void *b=0)
void	copyGenericColorSpinorSS (ColorSpinorField &, const ColorSpinorField &, QudaFieldLocation, void *, void *, void *a=0, void *b=0)
void	copyGenericColorSpinorSH (ColorSpinorField &, const ColorSpinorField &, QudaFieldLocation, void *, void *, void *a=0, void *b=0)
void	copyGenericColorSpinorHD (ColorSpinorField &, const ColorSpinorField &, QudaFieldLocation, void *, void *, void *a=0, void *b=0)
void	copyGenericColorSpinorHS (ColorSpinorField &, const ColorSpinorField &, QudaFieldLocation, void *, void *, void *a=0, void *b=0)
void	copyGenericColorSpinorHH (ColorSpinorField &, const ColorSpinorField &, QudaFieldLocation, void *, void *, void *a=0, void *b=0)
void	copyGenericColorSpinorMGDD (ColorSpinorField &, const ColorSpinorField &, QudaFieldLocation, void *, void *, void *a=0, void *b=0)
void	copyGenericColorSpinorMGDS (ColorSpinorField &, const ColorSpinorField &, QudaFieldLocation, void *, void *, void *a=0, void *b=0)
void	copyGenericColorSpinorMGSD (ColorSpinorField &, const ColorSpinorField &, QudaFieldLocation, void *, void *, void *a=0, void *b=0)
void	copyGenericColorSpinorMGSS (ColorSpinorField &, const ColorSpinorField &, QudaFieldLocation, void *, void *, void *a=0, void *b=0)
template<typename FloatOut , typename FloatIn , int Ns, int Nc, typename Arg , typename Basis >
void	copyColorSpinor (Arg &arg, const Basis &basis)
template<typename FloatOut , typename FloatIn , int Ns, int Nc, typename Arg , typename Basis >
__global__ void	copyColorSpinorKernel (Arg arg, Basis basis)
template<typename FloatOut , typename FloatIn , int Ns, int Nc, typename Out , typename In >
void	genericCopyColorSpinor (Out &outOrder, const In &inOrder, const ColorSpinorField &out, const ColorSpinorField &in, QudaFieldLocation location)
template<typename FloatOut , typename FloatIn , int Ns, int Nc, typename InOrder >
void	genericCopyColorSpinor (InOrder &inOrder, ColorSpinorField &out, const ColorSpinorField &in, QudaFieldLocation location, FloatOut *Out, float *outNorm)
template<typename FloatOut , typename FloatIn , int Ns, int Nc>
void	genericCopyColorSpinor (ColorSpinorField &out, const ColorSpinorField &in, QudaFieldLocation location, FloatOut *Out, FloatIn *In, float *outNorm, float *inNorm)
template<int Ns, int Nc, typename dstFloat , typename srcFloat >
void	copyGenericColorSpinor (ColorSpinorField &dst, const ColorSpinorField &src, QudaFieldLocation location, dstFloat *Dst, srcFloat *Src, float *dstNorm, float *srcNorm)
template<int Nc, typename dstFloat , typename srcFloat >
void	CopyGenericColorSpinor (ColorSpinorField &dst, const ColorSpinorField &src, QudaFieldLocation location, dstFloat *Dst, srcFloat *Src, float *dstNorm=0, float *srcNorm=0)
template<typename FloatOut , typename FloatIn , int Ns, int Nc, typename OutOrder , typename InOrder >
void	packSpinor (OutOrder &outOrder, const InOrder &inOrder, int volume)
template<typename FloatOut , typename FloatIn , int Ns, int Nc, typename OutOrder , typename InOrder >
__global__ void	packSpinorKernel (OutOrder outOrder, const InOrder inOrder, int volume)
template<typename FloatOut , typename FloatIn , int Ns, int Nc, typename OutOrder , typename InOrder >
void	genericCopyColorSpinor (OutOrder &outOrder, const InOrder &inOrder, const ColorSpinorField &out, QudaFieldLocation location)
template<typename FloatOut , typename FloatIn , int Ns, int Nc, typename InOrder >
void	genericCopyColorSpinor (InOrder &inOrder, ColorSpinorField &out, QudaFieldLocation location, FloatOut *Out)
template<typename FloatOut , typename FloatIn , int Ns, int Nc>
void	genericCopyColorSpinor (ColorSpinorField &out, const ColorSpinorField &in, QudaFieldLocation location, FloatOut *Out, FloatIn *In)
template<int Ns, int Nc, typename dstFloat , typename srcFloat >
void	copyGenericColorSpinor (ColorSpinorField &dst, const ColorSpinorField &src, QudaFieldLocation location, dstFloat *Dst, srcFloat *Src)
template<int Nc, typename dstFloat , typename srcFloat >
void	CopyGenericColorSpinor (ColorSpinorField &dst, const ColorSpinorField &src, QudaFieldLocation location, dstFloat *Dst, srcFloat *Src)
void	copyGenericGaugeDoubleOut (GaugeField &out, const GaugeField &in, QudaFieldLocation location, void *Out, void *In, void **ghostOut, void **ghostIn, int type)
void	copyGenericGaugeHalfOut (GaugeField &out, const GaugeField &in, QudaFieldLocation location, void *Out, void *In, void **ghostOut, void **ghostIn, int type)
void	copyGenericGaugeSingleOut (GaugeField &out, const GaugeField &in, QudaFieldLocation location, void *Out, void *In, void **ghostOut, void **ghostIn, int type)
void	copyGenericGaugeMG (GaugeField &out, const GaugeField &in, QudaFieldLocation location, void *Out, void *In, void **ghostOut, void **ghostIn, int type)
void	checkMomOrder (const GaugeField &u)
template<typename FloatOut , typename FloatIn , int length, typename OutOrder , typename InOrder , bool regularToextended>
__device__ __host__ void	copyGaugeEx (CopyGaugeExArg< OutOrder, InOrder > &arg, int X, int parity)
template<typename FloatOut , typename FloatIn , int length, typename OutOrder , typename InOrder , bool regularToextended>
void	copyGaugeEx (CopyGaugeExArg< OutOrder, InOrder > arg)
template<typename FloatOut , typename FloatIn , int length, typename OutOrder , typename InOrder , bool regularToextended>
__global__ void	copyGaugeExKernel (CopyGaugeExArg< OutOrder, InOrder > arg)
template<typename FloatOut , typename FloatIn , int length, typename OutOrder , typename InOrder >
void	copyGaugeEx (OutOrder outOrder, const InOrder inOrder, const int *E, const int *X, const int *faceVolumeCB, const GaugeField &meta, QudaFieldLocation location)
template<typename FloatOut , typename FloatIn , int length, typename InOrder >
void	copyGaugeEx (const InOrder &inOrder, const int *X, GaugeField &out, QudaFieldLocation location, FloatOut *Out)
template<typename FloatOut , typename FloatIn , int length>
void	copyGaugeEx (GaugeField &out, const GaugeField &in, QudaFieldLocation location, FloatOut *Out, FloatIn *In)
template<typename FloatOut , typename FloatIn >
void	copyGaugeEx (GaugeField &out, const GaugeField &in, QudaFieldLocation location, FloatOut *Out, FloatIn *In)
template<typename FloatOut , typename FloatIn , int length, typename OutOrder , typename InOrder >
void	copyGauge (CopyGaugeArg< OutOrder, InOrder > arg)
template<typename Float , int length, typename Arg >
void	checkNan (Arg arg)
template<typename FloatOut , typename FloatIn , int length, typename OutOrder , typename InOrder >
__global__ void	copyGaugeKernel (CopyGaugeArg< OutOrder, InOrder > arg)
template<typename FloatOut , typename FloatIn , int length, typename OutOrder , typename InOrder >
void	copyGhost (CopyGaugeArg< OutOrder, InOrder > arg)
template<typename FloatOut , typename FloatIn , int length, typename OutOrder , typename InOrder >
__global__ void	copyGhostKernel (CopyGaugeArg< OutOrder, InOrder > arg)
template<typename FloatOut , typename FloatIn , int length, typename OutOrder , typename InOrder >
void	copyGauge (OutOrder &&outOrder, const InOrder &inOrder, int volume, const int *faceVolumeCB, int nDim, int geometry, const GaugeField &out, const GaugeField &in, QudaFieldLocation location, int type)
template<typename FloatOut , typename FloatIn , int length, typename InOrder >
void	copyGauge (const InOrder &inOrder, const GaugeField &out, const GaugeField &in, QudaFieldLocation location, FloatOut *Out, FloatOut **outGhost, int type)
template<typename FloatOut , typename FloatIn , int length>
void	copyGauge (GaugeField &out, const GaugeField &in, QudaFieldLocation location, FloatOut *Out, FloatIn *In, FloatOut **outGhost, FloatIn **inGhost, int type)
template<typename FloatOut , typename FloatIn , int length, typename Out , typename In , typename Arg >
void	copyMom (Arg &arg, const GaugeField &out, const GaugeField &in, QudaFieldLocation location)
template<typename FloatOut , typename FloatIn >
void	copyGauge (GaugeField &out, const GaugeField &in, QudaFieldLocation location, FloatOut *Out, FloatIn *In, FloatOut **outGhost, FloatIn **inGhost, int type)
template<typename FloatOut , typename FloatIn , int length, typename InOrder >
void	copyGaugeMG (const InOrder &inOrder, GaugeField &out, const GaugeField &in, QudaFieldLocation location, FloatOut *Out, FloatOut **outGhost, int type)
template<typename FloatOut , typename FloatIn , int length>
void	copyGaugeMG (GaugeField &out, const GaugeField &in, QudaFieldLocation location, FloatOut *Out, FloatIn *In, FloatOut **outGhost, FloatIn **inGhost, int type)
template<typename FloatOut , typename FloatIn >
void	copyGaugeMG (GaugeField &out, const GaugeField &in, QudaFieldLocation location, FloatOut *Out, FloatIn *In, FloatOut **outGhost, FloatIn **inGhost, int type)
void *	create_gauge_buffer (size_t bytes, QudaGaugeFieldOrder order, QudaFieldGeometry geometry)
void **	create_ghost_buffer (size_t bytes[], QudaGaugeFieldOrder order, QudaFieldGeometry geometry)
void	free_gauge_buffer (void *buffer, QudaGaugeFieldOrder order, QudaFieldGeometry geometry)
void	free_ghost_buffer (void **buffer, QudaGaugeFieldOrder order, QudaFieldGeometry geometry)
std::ostream &	operator<< (std::ostream &out, const cudaColorSpinorField &a)
const map &	getTuneCache ()
void	disableProfileCount ()
void	enableProfileCount ()
void	setPolicyTuning (bool)
template<typename Float , int nColor, typename Arg >
void	gammaCPU (Arg arg)
template<typename Float , int nColor, int d, typename Arg >
__global__ void	gammaGPU (Arg arg)
template<typename Float , int nColor>
void	ApplyGamma (ColorSpinorField &out, const ColorSpinorField &in, int d)
template<typename Float >
void	ApplyGamma (ColorSpinorField &out, const ColorSpinorField &in, int d)
template<bool doublet, typename Float , int nColor, typename Arg >
void	twistGammaCPU (Arg arg)
template<bool doublet, typename Float , int nColor, int d, typename Arg >
__global__ void	twistGammaGPU (Arg arg)
template<typename Float , int nSpin, int nColor, typename Arg >
__device__ __host__ void	cloverApply (Arg &arg, int x_cb, int parity)
template<typename Float , int nSpin, int nColor, typename Arg >
void	cloverCPU (Arg &arg)
template<typename Float , int nSpin, int nColor, typename Arg >
__global__ void	cloverGPU (Arg arg)
template<bool inverse, typename Float , int nSpin, int nColor, typename Arg >
__device__ __host__ void	twistCloverApply (Arg &arg, int x_cb, int parity)
template<bool inverse, typename Float , int nSpin, int nColor, typename Arg >
void	twistCloverCPU (Arg &arg)
template<bool inverse, typename Float , int nSpin, int nColor, typename Arg >
__global__ void	twistCloverGPU (Arg arg)
static void	report (const char *type)
template<typename FloatOut , typename FloatIn , int Ns, int Nc, typename OutOrder , typename InOrder , typename Basis , bool extend>
__device__ __host__ void	copyInterior (CopySpinorExArg< OutOrder, InOrder, Basis > &arg, int X)
template<typename FloatOut , typename FloatIn , int Ns, int Nc, typename OutOrder , typename InOrder , typename Basis , bool extend>
__global__ void	copyInteriorKernel (CopySpinorExArg< OutOrder, InOrder, Basis > arg)
template<typename FloatOut , typename FloatIn , int Ns, int Nc, typename OutOrder , typename InOrder , typename Basis , bool extend>
void	copyInterior (CopySpinorExArg< OutOrder, InOrder, Basis > &arg)
template<typename FloatOut , typename FloatIn , int Ns, int Nc, typename OutOrder , typename InOrder , typename Basis >
void	copySpinorEx (OutOrder outOrder, const InOrder inOrder, const Basis basis, const int *E, const int *X, const int parity, const bool extend, const ColorSpinorField &meta, QudaFieldLocation location)
template<typename FloatOut , typename FloatIn , int Ns, int Nc, typename OutOrder , typename InOrder >
void	copySpinorEx (OutOrder outOrder, InOrder inOrder, const QudaGammaBasis outBasis, const QudaGammaBasis inBasis, const int *E, const int *X, const int parity, const bool extend, const ColorSpinorField &meta, QudaFieldLocation location)
template<typename FloatOut , typename FloatIn , int Ns, int Nc, typename InOrder >
void	extendedCopyColorSpinor (InOrder &inOrder, ColorSpinorField &out, QudaGammaBasis inBasis, const int *E, const int *X, const int parity, const bool extend, QudaFieldLocation location, FloatOut *Out, float *outNorm)
template<typename FloatOut , typename FloatIn , int Ns, int Nc>
void	extendedCopyColorSpinor (ColorSpinorField &out, const ColorSpinorField &in, const int parity, const QudaFieldLocation location, FloatOut *Out, FloatIn *In, float *outNorm, float *inNorm)
template<int Ns, typename dstFloat , typename srcFloat >
void	copyExtendedColorSpinor (ColorSpinorField &dst, const ColorSpinorField &src, const int parity, const QudaFieldLocation location, dstFloat *Dst, srcFloat *Src, float *dstNorm, float *srcNorm)
template<typename dstFloat , typename srcFloat >
void	CopyExtendedColorSpinor (ColorSpinorField &dst, const ColorSpinorField &src, const int parity, const QudaFieldLocation location, dstFloat *Dst, srcFloat *Src, float *dstNorm=0, float *srcNorm=0)
template<typename Float >
void	extractGhost (const GaugeField &u, Float **Ghost, bool extract, int offset)
void	extractGaugeGhostMG (const GaugeField &u, void **ghost, bool extract, int offset)
template<typename Float , int length, int dim, typename Arg >
__device__ __host__ void	extractor (Arg &arg, int dir, int a, int b, int c, int d, int g, int parity)
template<typename Float , int length, int dim, typename Arg >
__device__ __host__ void	injector (Arg &arg, int dir, int a, int b, int c, int d, int g, int parity)
template<typename Float , int length, int nDim, int dim, typename Order , bool extract>
void	extractGhostEx (ExtractGhostExArg< Order, nDim, dim > arg)
template<typename Float , int length, int nDim, int dim, typename Order , bool extract>
__global__ void	extractGhostExKernel (ExtractGhostExArg< Order, nDim, dim > arg)
template<typename Float , int length, typename Order >
void	extractGhostEx (Order order, const int dim, const int *surfaceCB, const int *E, const int *R, bool extract, const GaugeField &u, QudaFieldLocation location)
template<typename Float >
void	extractGhostEx (const GaugeField &u, int dim, const int *R, Float **Ghost, bool extract)
template<typename Float , int length, int nDim, typename Order , bool extract>
void	extractGhost (ExtractGhostArg< Order, nDim > arg)
template<typename Float , int length, int nDim, typename Order , bool extract>
__global__ void	extractGhostKernel (ExtractGhostArg< Order, nDim > arg)
template<typename Float , int length, typename Order >
void	extractGhost (Order order, const GaugeField &u, QudaFieldLocation location, bool extract, int offset)
template<typename Float , int Nc>
void	extractGhostMG (const GaugeField &u, Float **Ghost, bool extract, int offset)
template<typename Float >
void	extractGhostMG (const GaugeField &u, Float **Ghost, bool extract, int offset)
ColorSpinorParam	colorSpinorParam (const GaugeField &a)
template<int NCOLORS>
static __host__ __device__ void	IndexBlock (int block, int &p, int &q)
template<int blockSize, typename Float , int gauge_dir, int NCOLORS>
__forceinline__ __device__ void	GaugeFixHit_AtomicAdd (Matrix< complex< Float >, NCOLORS > &link, const Float relax_boost, const int tid)
template<int blockSize, typename Float , int gauge_dir, int NCOLORS>
__forceinline__ __device__ void	GaugeFixHit_NoAtomicAdd (Matrix< complex< Float >, NCOLORS > &link, const Float relax_boost, const int tid)
template<int blockSize, typename Float , int gauge_dir, int NCOLORS>
__forceinline__ __device__ void	GaugeFixHit_NoAtomicAdd_LessSM (Matrix< complex< Float >, NCOLORS > &link, const Float relax_boost, const int tid)
template<int blockSize, typename Float , int gauge_dir, int NCOLORS>
__forceinline__ __device__ void	GaugeFixHit_AtomicAdd (Matrix< complex< Float >, NCOLORS > &link, Matrix< complex< Float >, NCOLORS > &link1, const Float relax_boost, const int tid)
template<int blockSize, typename Float , int gauge_dir, int NCOLORS>
__forceinline__ __device__ void	GaugeFixHit_NoAtomicAdd (Matrix< complex< Float >, NCOLORS > &link, Matrix< complex< Float >, NCOLORS > &link1, const Float relax_boost, const int tid)
template<int blockSize, typename Float , int gauge_dir, int NCOLORS>
__forceinline__ __device__ void	GaugeFixHit_NoAtomicAdd_LessSM (Matrix< complex< Float >, NCOLORS > &link, Matrix< complex< Float >, NCOLORS > &link1, const Float relax_boost, const int tid)
template<typename Float , typename GaugeOr , typename GaugeDs , typename Float2 >
__host__ __device__ void	computeStapleRectangle (GaugeOvrImpSTOUTArg< Float, GaugeOr, GaugeDs > &arg, int idx, int parity, int dir, Matrix< Float2, 3 > &staple, Matrix< Float2, 3 > &rectangle)
template<typename Float , typename GaugeOr , typename GaugeDs >
__global__ void	computeOvrImpSTOUTStep (GaugeOvrImpSTOUTArg< Float, GaugeOr, GaugeDs > arg)
template<typename Float , typename GaugeOr , typename GaugeDs >
void	OvrImpSTOUTStep (GaugeOr origin, GaugeDs dest, const GaugeField &dataOr, Float rho, Float epsilon)
template<typename Float >
void	OvrImpSTOUTStep (GaugeField &dataDs, const GaugeField &dataOr, Float rho, Float epsilon)
void	printLaunchTimer ()
void	setDiracPreParam (DiracParam &diracParam, QudaInvertParam *inv_param, const bool pc, bool comms)
void	createDirac (Dirac *&d, Dirac *&dSloppy, Dirac *&dPre, QudaInvertParam &param, const bool pc_solve)
void	massRescale (cudaColorSpinorField &b, QudaInvertParam &param)
void	fillInnerSolveParam (SolverParam &inner, const SolverParam &outer)
int	reliable (double &rNorm, double &maxrx, double &maxrr, const double &r2, const double &delta)
template<libtype which_lib>
void	ComputeRitz (EigCGArgs &args)
template<>
void	ComputeRitz< libtype::eigen_lib > (EigCGArgs &args)
template<>
void	ComputeRitz< libtype::magma_lib > (EigCGArgs &args)
static void	fillEigCGInnerSolverParam (SolverParam &inner, const SolverParam &outer, bool use_sloppy_partial_accumulator=true)
static void	fillInitCGSolverParam (SolverParam &inner, const SolverParam &outer)
double	timeInterval (struct timeval start, struct timeval end)
void	computeBeta (Complex **beta, std::vector< ColorSpinorField *> Ap, int i, int N, int k)
void	updateAp (Complex **beta, std::vector< ColorSpinorField *> Ap, int begin, int size, int k)
void	orthoDir (Complex **beta, std::vector< ColorSpinorField *> Ap, int k, int pipeline)
void	backSubs (const Complex *alpha, Complex **const beta, const double *gamma, Complex *delta, int n)
void	updateSolution (ColorSpinorField &x, const Complex *alpha, Complex **const beta, double *gamma, int k, std::vector< ColorSpinorField *> p)
template<libtype which_lib>
void	ComputeHarmonicRitz (GMResDRArgs &args)
template<>
void	ComputeHarmonicRitz< libtype::magma_lib > (GMResDRArgs &args)
template<>
void	ComputeHarmonicRitz< libtype::eigen_lib > (GMResDRArgs &args)
template<libtype which_lib>
void	ComputeEta (GMResDRArgs &args)
template<>
void	ComputeEta< libtype::magma_lib > (GMResDRArgs &args)
template<>
void	ComputeEta< libtype::eigen_lib > (GMResDRArgs &args)
void	fillFGMResDRInnerSolveParam (SolverParam &inner, const SolverParam &outer)
template<typename T >
static void	applyT (T d_out[], const T d_in[], const T gamma[], const T rho[], int N)
template<typename T >
static void	applyB (T d_out[], const T d_in[], int N)
void	print (const double d[], int n)
template<typename T >
static void	zero (T d[], int N)
template<typename T >
static void	applyThirdTerm (T d_out[], const T d_in[], int k, int j, int s, const T gamma[], const T rho[], const T gamma_kprev[], const T rho_kprev[])
template<typename T >
static void	computeCoeffs (T d_out[], const T d_p1[], const T d_p2[], int k, int j, int s, const T gamma[], const T rho[], const T gamma_kprev[], const T rho_kprev[])
void	solve (Complex *psi, std::vector< ColorSpinorField *> &p, std::vector< ColorSpinorField *> &q, ColorSpinorField &b)
	Solve the equation A p_k psi_k = b by minimizing the residual and using Gaussian elimination. More...
void	updateAlphaZeta (double *alpha, double *zeta, double *zeta_old, const double *r2, const double *beta, const double pAp, const double *offset, const int nShift, const int j_low)
static void	fillInnerSolverParam (SolverParam &inner, const SolverParam &outer)
template<typename Float , typename Oprod , typename Gauge , typename Mom >
__host__ __device__ void	completeKSForceCore (KSForceArg< Oprod, Gauge, Mom > &arg, int idx)
template<typename Float , typename Oprod , typename Gauge , typename Mom >
__global__ void	completeKSForceKernel (KSForceArg< Oprod, Gauge, Mom > arg)
template<typename Float , typename Oprod , typename Gauge , typename Mom >
void	completeKSForceCPU (KSForceArg< Oprod, Gauge, Mom > &arg)
template<typename Float , typename Oprod , typename Gauge , typename Mom >
void	completeKSForce (Oprod oprod, Gauge gauge, Mom mom, int dim[4], const GaugeField &meta, QudaFieldLocation location, long long *flops)
template<typename Float , typename Result , typename Oprod , typename Gauge >
__host__ __device__ void	computeKSLongLinkForceCore (KSLongLinkArg< Result, Oprod, Gauge > &arg, int idx)
template<typename Float , typename Result , typename Oprod , typename Gauge >
__global__ void	computeKSLongLinkForceKernel (KSLongLinkArg< Result, Oprod, Gauge > arg)
template<typename Float , typename Result , typename Oprod , typename Gauge >
void	computeKSLongLinkForceCPU (KSLongLinkArg< Result, Oprod, Gauge > &arg)
template<typename Float , typename Result , typename Oprod , typename Gauge >
void	computeKSLongLinkForce (Result res, Oprod oprod, Gauge gauge, int dim[4], const GaugeField &meta, QudaFieldLocation location)
template<typename Float >
void	computeKSLongLinkForce (GaugeField &result, const GaugeField &oprod, const GaugeField &gauge, QudaFieldLocation location)
template<typename Float , int nDim, int nColor, typename Vector , typename Arg >
__device__ __host__ void	applyLaplace (Vector &out, Arg &arg, int x_cb, int parity)
template<typename Float , int nDim, int nColor, typename Arg >
__device__ __host__ void	laplace (Arg &arg, int x_cb, int parity)
template<typename Float , int nDim, int nColor, typename Arg >
void	laplaceCPU (Arg arg)
template<typename Float , int nDim, int nColor, typename Arg >
__global__ void	laplaceGPU (Arg arg)
static void	print_trace (void)
static void	print_alloc_header ()
static void	print_alloc (AllocType type)
static void	track_malloc (const AllocType &type, const MemAlloc &a, void *ptr)
static void	track_free (const AllocType &type, void *ptr)
static void *	aligned_malloc (MemAlloc &a, size_t size)
template<typename Float , int Nc, typename Order >
double	maxGauge (const Order order, int volume, int nDim)
template<typename Float >
void	arpack_solve (std::vector< ColorSpinorField *> &B, void *evals, DiracMatrix &matEigen, QudaPrecision matPrec, QudaPrecision arpackPrec, double tol, int nev, int ncv, char *target)
template<class T >
__device__ __host__ T	getTrace (const Matrix< T, 3 > &a)
template<template< typename, int > class Mat, class T >
__device__ __host__ T	getDeterminant (const Mat< T, 3 > &a)
template<template< typename, int > class Mat, class T , int N>
__device__ __host__ Mat< T, N >	operator+ (const Mat< T, N > &a, const Mat< T, N > &b)
template<template< typename, int > class Mat, class T , int N>
__device__ __host__ Mat< T, N >	operator+= (Mat< T, N > &a, const Mat< T, N > &b)
template<template< typename, int > class Mat, class T , int N>
__device__ __host__ Mat< T, N >	operator+= (Mat< T, N > &a, const T &b)
template<template< typename, int > class Mat, class T , int N>
__device__ __host__ Mat< T, N >	operator-= (Mat< T, N > &a, const Mat< T, N > &b)
template<template< typename, int > class Mat, class T , int N>
__device__ __host__ Mat< T, N >	operator- (const Mat< T, N > &a, const Mat< T, N > &b)
template<template< typename, int > class Mat, class T , int N, class S >
__device__ __host__ Mat< T, N >	operator* (const S &scalar, const Mat< T, N > &a)
template<template< typename, int > class Mat, class T , int N, class S >
__device__ __host__ Mat< T, N >	operator* (const Mat< T, N > &a, const S &scalar)
template<template< typename, int > class Mat, class T , int N, class S >
__device__ __host__ Mat< T, N >	operator*= (Mat< T, N > &a, const S &scalar)
template<template< typename, int > class Mat, class T , int N>
__device__ __host__ Mat< T, N >	operator- (const Mat< T, N > &a)
template<template< typename, int > class Mat, class T , int N>
__device__ __host__ Mat< T, N >	operator* (const Mat< T, N > &a, const Mat< T, N > &b)
	Generic implementation of matrix multiplication. More...
template<template< typename > class complex, typename T , int N>
__device__ __host__ Matrix< complex< T >, N >	operator* (const Matrix< complex< T >, N > &a, const Matrix< complex< T >, N > &b)
	Specialization of complex matrix multiplication that will issue optimal fma instructions. More...
template<class T , int N>
__device__ __host__ Matrix< T, N >	operator*= (Matrix< T, N > &a, const Matrix< T, N > &b)
template<class T , class U , int N>
__device__ __host__ Matrix< typename PromoteTypeId< T, U >::Type, N >	operator* (const Matrix< T, N > &a, const Matrix< U, N > &b)
template<class T >
__device__ __host__ Matrix< T, 2 >	operator* (const Matrix< T, 2 > &a, const Matrix< T, 2 > &b)
template<class T , int N>
__device__ __host__ Matrix< T, N >	conj (const Matrix< T, N > &other)
template<class T >
__device__ __host__ void	computeMatrixInverse (const Matrix< T, 3 > &u, Matrix< T, 3 > *uinv)
template<class T , int N>
__device__ __host__ void	setIdentity (Matrix< T, N > *m)
template<int N>
__device__ __host__ void	setIdentity (Matrix< float2, N > *m)
template<int N>
__device__ __host__ void	setIdentity (Matrix< double2, N > *m)
template<class T , int N>
__device__ __host__ void	setZero (Matrix< T, N > *m)
template<int N>
__device__ __host__ void	setZero (Matrix< float2, N > *m)
template<int N>
__device__ __host__ void	setZero (Matrix< double2, N > *m)
template<typename Complex , int N>
__device__ __host__ void	makeAntiHerm (Matrix< Complex, N > &m)
template<class T , int N>
__device__ __host__ void	copyColumn (const Matrix< T, N > &m, int c, Array< T, N > *a)
template<class T , int N>
__device__ __host__ void	outerProd (const Array< T, N > &a, const Array< T, N > &b, Matrix< T, N > *m)
template<class T , int N>
__device__ __host__ void	outerProd (const T(&a)[N], const T(&b)[N], Matrix< T, N > *m)
template<class T , int N>
std::ostream &	operator<< (std::ostream &os, const Matrix< T, N > &m)
template<class T , int N>
std::ostream &	operator<< (std::ostream &os, const Array< T, N > &a)
template<class T , class U >
__device__ void	loadLinkVariableFromArray (const T *const array, const int dir, const int idx, const int stride, Matrix< U, 3 > *link)
template<class T , class U , int N>
__device__ void	loadMatrixFromArray (const T *const array, const int idx, const int stride, Matrix< U, N > *mat)
__device__ void	loadLinkVariableFromArray (const float2 *const array, const int dir, const int idx, const int stride, Matrix< complex< double >, 3 > *link)
template<class T , int N, class U >
__device__ void	writeMatrixToArray (const Matrix< T, N > &mat, const int idx, const int stride, U *const array)
__device__ void	appendMatrixToArray (const Matrix< complex< double >, 3 > &mat, const int idx, const int stride, double2 *const array)
__device__ void	appendMatrixToArray (const Matrix< complex< float >, 3 > &mat, const int idx, const int stride, float2 *const array)
template<class T , class U >
__device__ void	writeLinkVariableToArray (const Matrix< T, 3 > &link, const int dir, const int idx, const int stride, U *const array)
__device__ void	writeLinkVariableToArray (const Matrix< complex< double >, 3 > &link, const int dir, const int idx, const int stride, float2 *const array)
template<class T >
__device__ void	loadMomentumFromArray (const T *const array, const int dir, const int idx, const int stride, Matrix< T, 3 > *mom)
template<class T , class U >
__device__ void	writeMomentumToArray (const Matrix< T, 3 > &mom, const int dir, const int idx, const U coeff, const int stride, T *const array)
template<class Cmplx >
__device__ __host__ void	computeLinkInverse (Matrix< Cmplx, 3 > *uinv, const Matrix< Cmplx, 3 > &u)
void	copyArrayToLink (Matrix< float2, 3 > *link, float *array)
template<class Cmplx , class Real >
void	copyArrayToLink (Matrix< Cmplx, 3 > *link, Real *array)
void	copyLinkToArray (float *array, const Matrix< float2, 3 > &link)
template<class Cmplx , class Real >
void	copyLinkToArray (Real *array, const Matrix< Cmplx, 3 > &link)
template<class T >
__device__ __host__ Matrix< T, 3 >	getSubTraceUnit (const Matrix< T, 3 > &a)
template<class T >
__device__ __host__ void	SubTraceUnit (Matrix< T, 3 > &a)
template<class T >
__device__ __host__ double	getRealTraceUVdagger (const Matrix< T, 3 > &a, const Matrix< T, 3 > &b)
template<class Cmplx >
__host__ __device__ void	printLink (const Matrix< Cmplx, 3 > &link)
template<class Cmplx >
__device__ __host__ bool	isUnitary (const Matrix< Cmplx, 3 > &matrix, double max_error)
template<class Cmplx >
__device__ __host__ double	ErrorSU3 (const Matrix< Cmplx, 3 > &matrix)
template<class T >
__device__ __host__ void	exponentiate_iQ (const Matrix< T, 3 > &Q, Matrix< T, 3 > *exp_iQ)
dim3	GetBlockDim (size_t threads, size_t size)
__global__ void	kernel_random (cuRNGState *state, int seed, int rng_size, int node_offset)
	CUDA kernel to initialize CURAND RNG states. More...
__global__ void	kernel_random (cuRNGState *state, int seed, int rng_size, int node_offset, rngArg arg)
void	launch_kernel_random (cuRNGState *state, int seed, int rng_size, int node_offset, int X[4])
	Call CUDA kernel to initialize CURAND RNG states. More...
template<IndexType idxType, typename Int >
__device__ __forceinline__ int	neighborIndex (const unsigned int &cb_idx, const int(&shift)[4], const bool(&partitioned)[4], const unsigned int &parity)
template<typename FloatN , int N, typename Output , typename Input >
__global__ void	shiftColorSpinorFieldKernel (ShiftQuarkArg< Output, Input > arg)
template<typename FloatN , int N, typename Output , typename Input >
__global__ void	shiftColorSpinorFieldExternalKernel (ShiftQuarkArg< Output, Input > arg)
void	shiftColorSpinorField (cudaColorSpinorField &dst, const cudaColorSpinorField &src, const unsigned int parity, const unsigned int dim, const int shift)
static void	report (const char *type)
template<typename InOrder , typename FloatIn >
__device__ __host__ void	genGauss (InOrder &inOrder, cuRNGState &localState, int x, int s, int c)
template<typename FloatIn , int Ns, int Nc, typename InOrder >
void	gaussSpinor (InOrder &inOrder, int volume, RNG rngstate)
template<typename FloatIn , int Ns, int Nc, typename InOrder >
__global__ void	gaussSpinorKernel (InOrder inOrder, int volume, RNG rngstate)
template<typename FloatIn , int Ns, int Nc, typename InOrder >
void	gaussSpinor (InOrder &inOrder, const ColorSpinorField &meta, RNG &rngstate)
template<typename FloatIn , int Ns, int Nc>
void	gaussSpinor (ColorSpinorField &in, RNG &rngstate)
void	computeStaggeredOprod (GaugeField &outA, GaugeField &outB, ColorSpinorField &inEven, ColorSpinorField &inOdd, const unsigned int parity, const double coeff[2], int nFace)
bool	traceEnabled ()
static void	deserializeTuneCache (std::istream &in)
static void	serializeTuneCache (std::ostream &out)
static void	serializeProfile (std::ostream &out, std::ostream &async_out)
static void	serializeTrace (std::ostream &out)
static void	broadcastTuneCache ()
bool	policyTuning ()
template<typename Float , typename G >
__global__ void	ProjectSU3kernel (ProjectSU3Arg< Float, G > arg)
void	setTransferGPU (bool)

Variables
__device__ unsigned int	count [QUDA_MAX_MULTI_REDUCE] = { }
__shared__ bool	isLastBlockDone
__shared__ volatile bool	isLastWarpDone [16]
const int	Nstream = 9
static const char	gDigitsLut [200]
static bool	bidirectional_debug = false
cudaStream_t *	stream
static bool	complete_recv_fwd [QUDA_MAX_DIM] = { }
static bool	complete_recv_back [QUDA_MAX_DIM] = { }
static bool	complete_send_fwd [QUDA_MAX_DIM] = { }
static bool	complete_send_back [QUDA_MAX_DIM] = { }
static auto	pinned_allocator = [] (size_t bytes ) { return static_cast<Complex*>(pool_pinned_malloc(bytes)); }
static auto	pinned_deleter = [] (Complex *hptr) { pool_pinned_free(hptr); }
static bool	dslash_init = false
static std::vector< DslashCoarsePolicy >	policy
static int	config = 0
static bool	kernelPackT = false
static double	unscaled_shifts [QUDA_MAX_MULTI_SHIFT]
static int	max_eigcg_cycles = 4
static QudaFieldLocation	reorder_location_ = QUDA_CUDA_FIELD_LOCATION
static std::map< void *, MemAlloc >	alloc [N_ALLOC_TYPE]
static long	total_bytes [N_ALLOC_TYPE] = {0}
static long	max_total_bytes [N_ALLOC_TYPE] = {0}
static long	total_host_bytes
static long	max_total_host_bytes
static long	total_pinned_bytes
static long	max_total_pinned_bytes
static bool	debug = false
static TimeProfile	apiTimer ("CUDA API calls (driver)")
static TuneKey	last_key
static std::list< TraceKey >	trace_list
static bool	enable_trace = false
static const std::string	quda_hash = QUDA_HASH
static std::string	resource_path
static map	tunecache
static map::iterator	it
static size_t	initial_cache_size = 0
static const std::string	quda_version = STR(QUDA_VERSION_MAJOR) "." STR(QUDA_VERSION_MINOR) "." STR(QUDA_VERSION_SUBMINOR)
static bool	tuning = false
static bool	profile_count = true
static bool	policy_tuning = false
static TimeProfile	launchTimer ("tuneLaunch")

Detailed Description

This is the covariant derivative based on the basic gauged Laplace operator

This code has not been checked. In particular, I suspect it is erroneous in multi-GPU since it looks like the halo ghost region isn't being treated here.

Generic Multi Shift Solver

For staggered, the mass is folded into the dirac operator Otherwise the matrix mass is 'unmodified'.

The lowest offset is in offsets[0]

This is a basic gauged Laplace operator

Typedef Documentation

ColorSpinorFieldSet

using quda::ColorSpinorFieldSet = typedef ColorSpinorField

Definition at line 808 of file invert_quda.h.

Complex

typedef std::complex< double > quda::Complex

Definition at line 13 of file eig_variables.h.

CompositeColorSpinorField

typedef std::vector<ColorSpinorField*> quda::CompositeColorSpinorField

Typedef for a set of spinors. Can be further divided into subsets ,e.g., with different precisions (not implemented currently)

Definition at line 17 of file color_spinor_field.h.

cuRNGState

typedef struct curandStateMRG32k3a quda::cuRNGState

Definition at line 17 of file random_quda.h.

DenseMatrix

typedef MatrixXcd quda::DenseMatrix

Definition at line 36 of file inv_eigcg_quda.cpp.

DynamicStride

typedef Stride< Dynamic, Dynamic > quda::DynamicStride

Definition at line 22 of file deflation.cpp.

map

typedef std::map< TuneKey, TuneParam > quda::map

Definition at line 948 of file dslash_coarse.cu.

RealVector

using quda::RealVector = typedef VectorXd

Definition at line 39 of file inv_eigcg_quda.cpp.

RowMajorDenseMatrix

typedef Matrix< Complex, Dynamic, Dynamic, RowMajor > quda::RowMajorDenseMatrix

Definition at line 42 of file inv_eigcg_quda.cpp.

Vector

typedef VectorXcd quda::Vector

Definition at line 38 of file inv_eigcg_quda.cpp.

VectorSet

typedef MatrixXcd quda::VectorSet

Definition at line 37 of file inv_eigcg_quda.cpp.

Enumeration Type Documentation

AllocType

enum quda::AllocType

Enumerator
DEVICE
HOST
PINNED
MAPPED
N_ALLOC_TYPE

Definition at line 15 of file malloc.cpp.

BiCGstabLUpdateType

enum quda::BiCGstabLUpdateType

The following code is based on Kate's worker class in Multi-CG.

This worker class is used to update most of the u and r vectors. On BiCG iteration j, r[0] through r[j] and u[0] through u[j] all get updated, but the subsequent mat-vec operation only gets applied to r[j] and u[j]. Thus, we can hide updating r[0] through r[j-1] and u[0] through u[j-1], respectively, in the comms for the matvec on r[j] and u[j]. This results in improved strong scaling for BiCGstab-L.

See paragraphs 2 and 3 in the comments on the Worker class in Multi-CG for more remarks.

Enumerator
BICGSTABL_UPDATE_U
BICGSTABL_UPDATE_R

Definition at line 181 of file inv_bicgstabl_quda.cpp.

ComputeType

enum quda::ComputeType

Enumerator
COMPUTE_UV
COMPUTE_AV
COMPUTE_TMAV
COMPUTE_TMCAV
COMPUTE_VUV
COMPUTE_COARSE_CLOVER
COMPUTE_REVERSE_Y
COMPUTE_COARSE_LOCAL
COMPUTE_DIAGONAL
COMPUTE_TMDIAGONAL
COMPUTE_INVALID

Definition at line 916 of file coarse_op.cuh.

DslashCoarsePolicy

enum quda::DslashCoarsePolicy

Enumerator
DSLASH_COARSE_BASIC
DSLASH_COARSE_ZERO_COPY_PACK
DSLASH_COARSE_ZERO_COPY_READ
DSLASH_COARSE_ZERO_COPY
DSLASH_COARSE_GDR_SEND
DSLASH_COARSE_GDR_RECV
DSLASH_COARSE_GDR
DSLASH_COARSE_ZERO_COPY_PACK_GDR_RECV
DSLASH_COARSE_GDR_SEND_ZERO_COPY_READ

Definition at line 863 of file dslash_coarse.cu.

libtype [1/2]

enum quda::libtype

strong

Enumerator
eigen_lib
magma_lib
lapack_lib
mkl_lib
eigen_lib
magma_lib
lapack_lib
mkl_lib

Definition at line 47 of file inv_eigcg_quda.cpp.

libtype [2/2]

enum quda::libtype

strong

Enumerator
eigen_lib
magma_lib
lapack_lib
mkl_lib
eigen_lib
magma_lib
lapack_lib
mkl_lib

Definition at line 57 of file inv_gmresdr_quda.cpp.

MemoryLocation

enum quda::MemoryLocation

Enumerator
Device
Host
Remote

Definition at line 15 of file color_spinor_field.h.

QudaProfileType

enum quda::QudaProfileType

Enumerator
QUDA_PROFILE_H2D	host -> device transfers
QUDA_PROFILE_D2H	The time in seconds for device -> host transfers
QUDA_PROFILE_INIT	The time in seconds taken for initiation
QUDA_PROFILE_PREAMBLE	The time in seconds taken for any preamble
QUDA_PROFILE_COMPUTE	The time in seconds taken for the actual computation
QUDA_PROFILE_COMMS	synchronous communication
QUDA_PROFILE_EPILOGUE	The time in seconds taken for any epilogue
QUDA_PROFILE_FREE	The time in seconds for freeing resources
QUDA_PROFILE_IO	time spent on file i/o
QUDA_PROFILE_LOWER_LEVEL	dummy timer to mark beginning of lower level timers which do not count towrads global time
QUDA_PROFILE_PACK_KERNEL	face packing kernel
QUDA_PROFILE_DSLASH_KERNEL	dslash kernel
QUDA_PROFILE_GATHER	gather (device -> host)
QUDA_PROFILE_SCATTER	scatter (host -> device)
QUDA_PROFILE_LAUNCH_KERNEL	cudaLaunchKernel
QUDA_PROFILE_EVENT_RECORD	cuda event record
QUDA_PROFILE_EVENT_QUERY	cuda event querying
QUDA_PROFILE_STREAM_WAIT_EVENT	stream waiting for event completion
QUDA_PROFILE_FUNC_SET_ATTRIBUTE	set function attribute
QUDA_PROFILE_EVENT_SYNCHRONIZE	event synchronization
QUDA_PROFILE_STREAM_SYNCHRONIZE	stream synchronization
QUDA_PROFILE_DEVICE_SYNCHRONIZE	device synchronization
QUDA_PROFILE_MEMCPY_D2D_ASYNC	device to device async copy
QUDA_PROFILE_MEMCPY_D2H_ASYNC	device to host async copy
QUDA_PROFILE_MEMCPY2D_D2H_ASYNC	device to host 2-d memcpy async copy
QUDA_PROFILE_MEMCPY_H2D_ASYNC	host to device async copy
QUDA_PROFILE_COMMS_START	initiating communication
QUDA_PROFILE_COMMS_QUERY	querying communication
QUDA_PROFILE_CONSTANT	time spent setting CUDA constant parameters
QUDA_PROFILE_TOTAL	The total time in seconds for the algorithm. Must be the penultimate type.
QUDA_PROFILE_COUNT	The total number of timers we have. Must be last enum type.

Definition at line 167 of file quda_internal.h.

Function Documentation

abs() [1/4]

template<typename ValueType >

__host__ __device__ ValueType quda::abs ( ValueType  x )

inline

Definition at line 110 of file complex_quda.h.

References abs(), and x.

Referenced by ComputeHarmonicRitz< libtype::eigen_lib >(), ComputeHarmonicRitz< libtype::magma_lib >(), log(), maxGauge(), quda::BiCGstab::operator()(), solve(), sqrt(), and test().

Here is the call graph for this function:

Here is the caller graph for this function:

abs() [2/4]

template<typename ValueType >

__host__ __device__ ValueType quda::abs ( const complex< ValueType > &  z )

inline

Returns the magnitude of z.

Definition at line 864 of file complex_quda.h.

References hypot(), and z.

Here is the call graph for this function:

abs() [3/4]

template<>

__host__ __device__ float quda::abs ( const complex< float > &  z )

inline

Definition at line 869 of file complex_quda.h.

References hypotf(), and z.

Here is the call graph for this function:

abs() [4/4]

template<>

__host__ __device__ double quda::abs ( const complex< double > &  z )

inline

Definition at line 874 of file complex_quda.h.

References hypot(), and z.

Referenced by abs().

Here is the call graph for this function:

Here is the caller graph for this function:

acos() [1/2]

template<typename ValueType >

__host__ __device__ ValueType quda::acos ( ValueType  x )

inline

Definition at line 50 of file complex_quda.h.

References acos(), and x.

Referenced by exponentiate_iQ().

Here is the call graph for this function:

Here is the caller graph for this function:

acos() [2/2]

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::acos ( const complex< ValueType > &  z )

inline

Definition at line 1078 of file complex_quda.h.

References asin(), ret, and z.

Referenced by acos().

Here is the call graph for this function:

Here is the caller graph for this function:

acosh()

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::acosh ( const complex< ValueType > &  z )

inline

Definition at line 1099 of file complex_quda.h.

References log(), ret, sqrt(), and z.

Here is the call graph for this function:

activeTuning()

bool quda::activeTuning

(

)

query if tuning is in progress

Returns

tuning in progress?

Definition at line 103 of file tune.cpp.

References tuning.

Referenced by qudaLaunchKernel().

Here is the caller graph for this function:

AddCoarseDiagonalCPU()

template<typename Float , int nSpin, int nColor, typename Arg >

void quda::AddCoarseDiagonalCPU

(

Arg &

arg

)

Definition at line 846 of file coarse_op.cuh.

References arg(), c, nColor, parity, and s.

Here is the call graph for this function:

AddCoarseDiagonalGPU()

template<typename Float , int nSpin, int nColor, typename Arg >

__global__ void quda::AddCoarseDiagonalGPU

(

Arg

arg

)

Definition at line 861 of file coarse_op.cuh.

References arg(), blockDim, c, nColor, parity, and s.

Here is the call graph for this function:

AddCoarseTmDiagonalCPU()

template<typename Float , int nSpin, int nColor, typename Arg >

void quda::AddCoarseTmDiagonalCPU

(

Arg &

arg

)

Definition at line 875 of file coarse_op.cuh.

References arg(), c, mu, nColor, parity, and s.

Here is the call graph for this function:

AddCoarseTmDiagonalGPU()

template<typename Float , int nSpin, int nColor, typename Arg >

__global__ void quda::AddCoarseTmDiagonalGPU

(

Arg

arg

)

Definition at line 897 of file coarse_op.cuh.

References arg(), blockDim, mu, nColor, parity, and s.

Here is the call graph for this function:

aligned_malloc()

static void* quda::aligned_malloc ( MemAlloc &  a, size_t  size  )

static

Under CUDA 4.0, cudaHostRegister seems to require that both the beginning and end of the buffer be aligned on page boundaries. This local function takes care of the alignment and gets called by pinned_malloc_() and mapped_malloc_()

Definition at line 139 of file malloc.cpp.

References a, errorQuda, malloc(), posix_memalign(), printfQuda, ptr, and size.

Referenced by mapped_malloc_(), and pinned_malloc_().

Here is the call graph for this function:

Here is the caller graph for this function:

APEStep()

void quda::APEStep	(	GaugeField &	dataDs,
		const GaugeField &	dataOr,
		double	alpha
	)

Apply APE smearing to the gauge field

Parameters

dataDs	Output smeared field
dataOr	Input gauge field
alpha	smearing parameter

Definition at line 240 of file gauge_ape.cu.

References errorQuda, float, quda::GaugeField::isNative(), quda::GaugeField::Order(), quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, QUDA_HALF_PRECISION, QUDA_SINGLE_PRECISION, and quda::GaugeField::Reconstruct().

Referenced by performAPEnStep().

Here is the call graph for this function:

Here is the caller graph for this function:

appendMatrixToArray() [1/2]

__device__ void quda::appendMatrixToArray ( const Matrix< complex< double >, 3 > &  mat, const int  idx, const int  stride, double2 *const  array  )

inline

Definition at line 794 of file quda_matrix.h.

References array, fused_exterior_ndeg_tm_dslash_cuda_gen::i, idx, and mat().

Here is the call graph for this function:

appendMatrixToArray() [2/2]

__device__ void quda::appendMatrixToArray ( const Matrix< complex< float >, 3 > &  mat, const int  idx, const int  stride, float2 *const  array  )

inline

Definition at line 804 of file quda_matrix.h.

References array, fused_exterior_ndeg_tm_dslash_cuda_gen::i, idx, and mat().

Here is the call graph for this function:

applyB()

template<typename T >

static void quda::applyB ( T  d_out[], const T  d_in[], int  N  )

static

Definition at line 37 of file inv_mpcg_quda.cpp.

References fused_exterior_ndeg_tm_dslash_cuda_gen::i.

Referenced by applyThirdTerm().

Here is the caller graph for this function:

ApplyClover()

void quda::ApplyClover	(	ColorSpinorField &	out,
		const ColorSpinorField &	in,
		const CloverField &	clover,
		bool	inverse,
		int	parity
	)

Apply clover-matrix field to a color-spinor field.

Parameters

[out]	out	Result color-spinor field
[in]	in	Input color-spinor field
[in]	clover	Clover-matrix field
[in]	inverse	Whether we are applying the inverse or not
[in]	Field	parity (if color-spinor field is single parity)

Definition at line 557 of file dslash_quda.cu.

References quda::Clover< Float, nSpin, nColor, Arg >::apply(), arg(), checkCudaError, dslash_cuda_gen::clover, errorQuda, in, quda::ColorSpinorField::Nspin(), Nstream, out, parity, and streams.

Referenced by quda::DiracClover::Clover(), and quda::DiracCloverPC::CloverInv().

Here is the call graph for this function:

Here is the caller graph for this function:

ApplyCoarse()

void quda::ApplyCoarse	(	ColorSpinorField &	out,
		const ColorSpinorField &	inA,
		const ColorSpinorField &	inB,
		const GaugeField &	Y,
		const GaugeField &	X,
		double	kappa,
		int	parity = QUDA_INVALID_PARITY,
		bool	dslash = true,
		bool	clover = true,
		bool	dagger = false
	)

Definition at line 1096 of file dslash_coarse.cu.

References dslash_cuda_gen::clover, deg_tm_dslash_cuda_gen::dagger, kappa, out, parity, policy, and X.

Referenced by quda::DiracCoarse::Clover(), quda::DiracCoarse::CloverInv(), quda::DiracCoarse::Dslash(), quda::DiracCoarsePC::Dslash(), quda::DiracCoarse::DslashXpay(), and quda::DiracCoarse::M().

Here is the caller graph for this function:

ApplyCovDev()

void quda::ApplyCovDev	(	ColorSpinorField &	out,
		const ColorSpinorField &	in,
		const GaugeField &	U,
		int	parity,
		int	mu
	)

Driver for applying the covariant derivative.

out = U * in

where U is the gauge field in a particular direction.

This operator can be applied to both single parity (checker-boarded) fields, or to full fields.

Parameters

[out]	out	The output result field
[in]	in	The input field
[in]	U	The gauge field used for the covariant derivative
[in]	mu	Direction of the derivative. For mu > 3 it goes backwards

Definition at line 264 of file covDev.cu.

References quda::Worker::apply(), quda::dslash::aux_worker, quda::LatticeField::bufferIndex, checkLocation, checkPrecision, errorQuda, quda::cpuColorSpinorField::exchangeGhost(), quda::ColorSpinorField::FieldOrder(), in, mu, out, parity, quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, QUDA_SINGLE_PRECISION, and quda::ColorSpinorField::V().

Referenced by quda::GaugeCovDev::DslashCD().

Here is the call graph for this function:

Here is the caller graph for this function:

ApplyGamma() [1/2]

template<typename Float , int nColor>

void quda::ApplyGamma	(	ColorSpinorField &	out,
		const ColorSpinorField &	in,
		int	d
	)

Definition at line 262 of file dslash_quda.cu.

References arg(), d, gamma(), in, Nstream, out, and streams.

Referenced by gamma5().

Here is the call graph for this function:

Here is the caller graph for this function:

ApplyGamma() [2/2]

template<typename Float >

void quda::ApplyGamma	(	ColorSpinorField &	out,
		const ColorSpinorField &	in,
		int	d
	)

Definition at line 271 of file dslash_quda.cu.

References d, errorQuda, in, quda::ColorSpinorField::Ncolor(), and out.

Here is the call graph for this function:

applyGaugePhase()

void quda::applyGaugePhase

(

GaugeField &

u

)

Apply the staggered phase factor to the gauge field.

Parameters

[in]

u

The gauge field to which we apply the staggered phase factors

Definition at line 244 of file gauge_phase.cu.

References errorQuda, quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, and QUDA_SINGLE_PRECISION.

Referenced by quda::GaugeField::applyStaggeredPhase(), and quda::GaugeField::removeStaggeredPhase().

Here is the call graph for this function:

Here is the caller graph for this function:

ApplyLaplace()

void quda::ApplyLaplace	(	ColorSpinorField &	out,
		const ColorSpinorField &	in,
		const GaugeField &	U,
		double	kappa,
		const ColorSpinorField *	x,
		int	parity
	)

Driver for applying the Laplace stencil.

out = - kappa * A * in

where A is the gauge laplace linear operator.

If x is defined, the operation is given by out = x - kappa * A in. This operator can be applied to both single parity (checker-boarded) fields, or to full fields.

Parameters

[out]	out	The output result field
[in]	in	The input field
[in]	U	The gauge field used for the gauge Laplace
[in]	kappa	Scale factor applied
[in]	x	Vector field we accumulate onto to

Definition at line 210 of file laplace.cu.

References arg(), in, kappa, laplace(), out, parity, and x.

Referenced by quda::GaugeLaplace::Dslash(), and quda::GaugeLaplace::DslashXpay().

Here is the call graph for this function:

Here is the caller graph for this function:

applyLaplace()

template<typename Float , int nDim, int nColor, typename Vector , typename Arg >

__device__ __host__ void quda::applyLaplace ( Vector &  out, Arg &  arg, int  x_cb, int  parity  )

inline

Applies the off-diagonal part of the Laplace operator

Parameters

[out]	out	The out result field
[in]	U	The gauge field
[in]	kappa	Kappa value
[in]	in	The input field
[in]	parity	The site parity
[in]	x_cb	The checkerboarded site index

Definition at line 59 of file laplace.cu.

References arg(), conj(), coord, d, getCoords(), in, linkIndexM1(), linkIndexP1(), nColor, out, and parity.

Here is the call graph for this function:

applyT()

template<typename T >

static void quda::applyT ( T  d_out[], const T  d_in[], const T  gamma[], const T  rho[], int  N  )

static

Definition at line 18 of file inv_mpcg_quda.cpp.

References gamma(), and fused_exterior_ndeg_tm_dslash_cuda_gen::i.

Referenced by applyThirdTerm().

Here is the call graph for this function:

Here is the caller graph for this function:

applyThirdTerm()

template<typename T >

static void quda::applyThirdTerm ( T  d_out[], const T  d_in[], int  k, int  j, int  s, const T  gamma[], const T  rho[], const T  gamma_kprev[], const T  rho_kprev[]  )

static

Definition at line 57 of file inv_mpcg_quda.cpp.

References applyB(), applyT(), dim, gamma(), fused_exterior_ndeg_tm_dslash_cuda_gen::i, s, and zero().

Referenced by computeCoeffs().

Here is the call graph for this function:

Here is the caller graph for this function:

ApplyTwistClover()

void quda::ApplyTwistClover	(	ColorSpinorField &	out,
		const ColorSpinorField &	in,
		const CloverField &	clover,
		double	kappa,
		double	mu,
		double	epsilon,
		int	parity,
		int	dagger,
		QudaTwistGamma5Type	twist
	)

Apply twisted clover-matrix field to a color-spinor field.

Parameters

[out]	out	Result color-spinor field
[in]	in	Input color-spinor field
[in]	clover	Clover-matrix field
[in]	kappa	kappa parameter
[in]	mu	mu parameter
[in]	epsilon	epsilon parameter
[in]	Field	parity (if color-spinor field is single parity)
[in]	dagger	Whether we are applying the dagger or not
[in]	twist	The type of kernel we are doing if (twist == QUDA_TWIST_GAMMA5_DIRECT) apply (Clover + i*a*gamma_5) to the input spinor else if (twist == QUDA_TWIST_GAMMA5_INVERSE) apply (Clover + i*a*gamma_5)/(Clover^2 + a^2) to the input spinor

Definition at line 708 of file dslash_quda.cu.

References quda::TwistClover< Float, nSpin, nColor, Arg >::apply(), arg(), checkCudaError, dslash_cuda_gen::clover, deg_tm_dslash_cuda_gen::dagger, errorQuda, in, kappa, mu, quda::ColorSpinorField::Nspin(), Nstream, out, parity, QUDA_TWIST_GAMMA5_DIRECT, streams, and deg_tm_dslash_cuda_gen::twist.

Referenced by quda::DiracTwistedClover::twistedCloverApply().

Here is the call graph for this function:

Here is the caller graph for this function:

ApplyTwistGamma()

void quda::ApplyTwistGamma	(	ColorSpinorField &	out,
		const ColorSpinorField &	in,
		int	d,
		double	kappa,
		double	mu,
		double	epsilon,
		int	dagger,
		QudaTwistGamma5Type	type
	)

Apply the twisted-mass gamma operator to a color-spinor field.

Parameters

[out]	out	Result color-spinor field
[in]	in	Input color-spinor field
[in]	d	Which gamma matrix we are applying (C counting, so gamma_5 has d=4)
[in]	kappa	kappa parameter
[in]	mu	mu parameter
[in]	epsilon	epsilon parameter
[in]	dagger	Whether we are applying the dagger or not
[in]	twist	The type of kernel we are doing

Definition at line 384 of file dslash_quda.cu.

References arg(), checkCudaError, d, deg_tm_dslash_cuda_gen::dagger, gamma(), in, kappa, mu, Nstream, out, and streams.

Referenced by quda::DiracTwistedMassPC::Dslash(), quda::DiracTwistedMassPC::DslashXpay(), and quda::DiracTwistedMass::twistedApply().

Here is the call graph for this function:

Here is the caller graph for this function:

applyU()

void quda::applyU	(	GaugeField &	force,
		GaugeField &	U
	)

Left multiply the force field by the gauge field

force = U * force

Parameters

force	Force field
U	Gauge field

Definition at line 340 of file momentum.cu.

References checkCudaError, errorQuda, quda::GaugeField::Order(), quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, and QUDA_FLOAT2_GAUGE_ORDER.

Referenced by computeStaggeredForceQuda().

Here is the call graph for this function:

Here is the caller graph for this function:

arg() [1/3]

template<typename ValueType >

__host__ __device__ ValueType quda::arg ( const complex< ValueType > &  z )

inline

Returns the phase angle of z.

Definition at line 880 of file complex_quda.h.

References atan2(), and z.

Referenced by AddCoarseDiagonalCPU(), AddCoarseDiagonalGPU(), AddCoarseTmDiagonalCPU(), AddCoarseTmDiagonalGPU(), quda::CopyGaugeEx< FloatOut, FloatIn, length, OutOrder, InOrder >::apply(), quda::KSForceComplete< Float, Oprod, Gauge, Mom >::apply(), quda::CopyColorSpinor< FloatOut, FloatIn, Ns, Nc, Arg >::apply(), quda::ExtractGhostEx< Float, length, nDim, dim, Order >::apply(), quda::CopyGauge< FloatOut, FloatIn, length, OutOrder, InOrder, isGhost >::apply(), quda::CopyColorSpinor< FloatOut, FloatIn, 4, Nc, Arg >::apply(), quda::CopySpinorEx< FloatOut, FloatIn, Ns, Nc, OutOrder, InOrder, Basis, extend >::apply(), quda::KSLongLinkForce< Float, Result, Oprod, Gauge >::apply(), ApplyClover(), ApplyGamma(), ApplyLaplace(), applyLaplace(), ApplyTwistClover(), ApplyTwistGamma(), arpack_solve(), blasKernel(), quda::CopyGaugeEx< FloatOut, FloatIn, length, OutOrder, InOrder >::bytes(), quda::CopyColorSpinor< FloatOut, FloatIn, Ns, Nc, Arg >::bytes(), quda::CopyGauge< FloatOut, FloatIn, length, OutOrder, InOrder, isGhost >::bytes(), quda::ExtractGhostEx< Float, length, nDim, dim, Order >::bytes(), quda::CopyColorSpinor< FloatOut, FloatIn, 4, Nc, Arg >::bytes(), quda::CopySpinorEx< FloatOut, FloatIn, Ns, Nc, OutOrder, InOrder, Basis, extend >::bytes(), calculateY(), CalculateYhatCPU(), CalculateYhatGPU(), checkNan(), Checksum(), ChecksumCPU(), cloverApply(), cloverCPU(), cloverGPU(), completeKSForce(), completeKSForceCore(), completeKSForceCPU(), completeKSForceKernel(), compute(), computeAV(), ComputeAVCPU(), ComputeAVGPU(), computeCoarseClover(), ComputeCoarseCloverCPU(), ComputeCoarseCloverGPU(), computeCoarseLocal(), ComputeCoarseLocalCPU(), ComputeCoarseLocalGPU(), computeKSLongLinkForce(), computeKSLongLinkForceCPU(), computeKSLongLinkForceKernel(), computeNeighborSum(), computeOvrImpSTOUTStep(), computeStapleRectangle(), computeTMAV(), ComputeTMAVCPU(), ComputeTMAVGPU(), computeTMCAV(), ComputeTMCAVCPU(), ComputeTMCAVGPU(), computeUV(), ComputeUVCPU(), ComputeUVGPU(), computeVUV(), ComputeVUVCPU(), ComputeVUVGPU(), computeWupperalStep(), computeYhat(), computeYreverse(), ComputeYReverseCPU(), ComputeYReverseGPU(), copyColorSpinor(), quda::CopyColorSpinor< FloatOut, FloatIn, Ns, Nc, Arg >::CopyColorSpinor(), quda::CopyColorSpinor< FloatOut, FloatIn, 4, Nc, Arg >::CopyColorSpinor(), copyColorSpinorKernel(), copyGauge(), quda::CopyGauge< FloatOut, FloatIn, length, OutOrder, InOrder, isGhost >::CopyGauge(), copyGaugeEx(), quda::CopyGaugeEx< FloatOut, FloatIn, length, OutOrder, InOrder >::CopyGaugeEx(), copyGaugeExKernel(), copyGaugeKernel(), copyGhost(), copyGhostKernel(), copyInterior(), copyInteriorKernel(), copyMom(), quda::CopySpinorEx< FloatOut, FloatIn, Ns, Nc, OutOrder, InOrder, Basis, extend >::CopySpinorEx(), copySpinorEx(), extractGhost(), extractGhostEx(), quda::ExtractGhostEx< Float, length, nDim, dim, Order >::ExtractGhostEx(), extractGhostExKernel(), extractGhostKernel(), extractor(), quda::KSForceComplete< Float, Oprod, Gauge, Mom >::flops(), gammaCPU(), gammaGPU(), genericCopyColorSpinor(), GenericPackGhost(), genericPackGhost(), GenericPackGhostKernel(), quda::gauge::Reconstruct< 13, Float >::getPhase(), quda::gauge::Reconstruct< 9, Float >::getPhase(), injector(), kernel_random(), quda::KSForceComplete< Float, Oprod, Gauge, Mom >::KSForceComplete(), quda::KSLongLinkForce< Float, Result, Oprod, Gauge >::KSLongLinkForce(), laplace(), laplaceCPU(), laplaceGPU(), launch_kernel_random(), log(), quda::CopyGaugeEx< FloatOut, FloatIn, length, OutOrder, InOrder >::minThreads(), quda::KSForceComplete< Float, Oprod, Gauge, Mom >::minThreads(), quda::CopySpinorEx< FloatOut, FloatIn, Ns, Nc, OutOrder, InOrder, Basis, extend >::minThreads(), quda::KSLongLinkForce< Float, Result, Oprod, Gauge >::minThreads(), multiblasKernel(), multiplyVUV(), multiReduceKernel(), multiReduceLaunch(), OvrImpSTOUTStep(), packGhost(), projectSU3(), ProjectSU3kernel(), reduce(), reduce2d(), reduceKernel(), reduceLaunch(), reduceRow(), shiftColorSpinorField(), shiftColorSpinorFieldExternalKernel(), shiftColorSpinorFieldKernel(), siteChecksum(), sqrt(), twistCloverApply(), twistCloverCPU(), twistCloverGPU(), twistGammaCPU(), twistGammaGPU(), wuppertalStep(), wuppertalStepCPU(), and wuppertalStepGPU().

Here is the call graph for this function:

arg() [2/3]

template<>

__host__ __device__ float quda::arg ( const complex< float > &  z )

inline

Definition at line 885 of file complex_quda.h.

References atan2f(), and z.

Here is the call graph for this function:

arg() [3/3]

template<>

__host__ __device__ double quda::arg ( const complex< double > &  z )

inline

Definition at line 890 of file complex_quda.h.

References atan2(), and z.

Here is the call graph for this function:

arpack_solve()

template<typename Float >

void quda::arpack_solve	(	std::vector< ColorSpinorField *> &	B,
		void *	evals,
		DiracMatrix &	matEigen,
		QudaPrecision	matPrec,
		QudaPrecision	arpackPrec,
		double	tol,
		int	nev,
		int	ncv,
		char *	target
	)

Definition at line 357 of file quda_arpack_interface.cpp.

References arg(), nev, and tol.

Here is the call graph for this function:

arpackSolve()

void quda::arpackSolve	(	std::vector< ColorSpinorField *> &	B,
		void *	evals,
		DiracMatrix &	matEigen,
		QudaPrecision	matPrec,
		QudaPrecision	arpackPrec,
		double	tol,
		int	nev,
		int	ncv,
		char *	target
	)

Interface function to the external ARPACK library. This function utilizes ARPACK implemntation of the Implicitly Restarted Arnoldi Method to compute a number of eigenvectors/eigenvalues with user specified features such as those with small real part, small magnitude etc. Parallel version is also supported.

Parameters

	[in/out]	B Container of eigenvectors
	[in/out]	evals A pointer to eigenvalue array.
[in]	matEigen	Any QUDA implementation of the matrix-vector operation
[in]	matPrec	Precision of the matrix-vector operation
[in]	arpackPrec	Precision of IRAM procedures.
[in]	tol	tolerance for computing eigenvalues with ARPACK
[in]	nev	number of eigenvectors
[in]	ncv	size of the subspace used by IRAM. ncv must satisfy the two inequalities 2 <= ncv-nev and ncv <= *B[0].Length()
[in]	target	eigenvector selection criteria: 'LM' -> want the nev eigenvalues of largest magnitude. 'SM' -> want the nev eigenvalues of smallest magnitude. 'LR' -> want the nev eigenvalues of largest real part. 'SR' -> want the nev eigenvalues of smallest real part. 'LI' -> want the nev eigenvalues of largest imaginary part. 'SI' -> want the nev eigenvalues of smallest imaginary part.

Definition at line 367 of file quda_arpack_interface.cpp.

References errorQuda, nev, QUDA_DOUBLE_PRECISION, and tol.

Referenced by quda::MG::verify().

Here is the caller graph for this function:

asin() [1/2]

template<typename ValueType >

__host__ __device__ ValueType quda::asin ( ValueType  x )

inline

Definition at line 55 of file complex_quda.h.

References asin(), and x.

Referenced by acos().

Here is the call graph for this function:

Here is the caller graph for this function:

asin() [2/2]

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::asin ( const complex< ValueType > &  z )

inline

Definition at line 1085 of file complex_quda.h.

References asinh(), fused_exterior_ndeg_tm_dslash_cuda_gen::i, and z.

Referenced by asin().

Here is the call graph for this function:

Here is the caller graph for this function:

asinh()

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::asinh ( const complex< ValueType > &  z )

inline

Definition at line 1124 of file complex_quda.h.

References log(), sqrt(), and z.

Referenced by asin().

Here is the call graph for this function:

Here is the caller graph for this function:

assertAllMemFree()

void quda::assertAllMemFree

(

)

Definition at line 379 of file malloc.cpp.

References alloc, DEVICE, HOST, MAPPED, PINNED, print_alloc(), print_alloc_header(), printfQuda, and warningQuda.

Referenced by endQuda().

Here is the call graph for this function:

Here is the caller graph for this function:

asymCloverDslashCuda()

void quda::asymCloverDslashCuda	(	cudaColorSpinorField *	out,
		const cudaGaugeField &	gauge,
		const FullClover &	cloverInv,
		const cudaColorSpinorField *	in,
		const int	oddBit,
		const int	daggerBit,
		const cudaColorSpinorField *	x,
		const double &	k,
		const int *	commDim,
		TimeProfile &	profile
	)

Definition at line 156 of file dslash_clover_asym.cu.

References a, dslash_cuda_gen::clover, deg_tm_dslash_cuda_gen::dagger, deg_tm_dslash_cuda_gen::dslash, errorQuda, in, out, parity, QUDA_DOUBLE_PRECISION, QUDA_HALF_PRECISION, QUDA_SINGLE_PRECISION, and x.

Referenced by quda::DiracClover::DslashXpay().

Here is the caller graph for this function:

atan() [1/2]

template<typename ValueType >

__host__ __device__ ValueType quda::atan ( ValueType  x )

inline

Definition at line 60 of file complex_quda.h.

References atan(), and x.

Here is the call graph for this function:

atan() [2/2]

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::atan ( const complex< ValueType > &  z )

inline

Definition at line 1092 of file complex_quda.h.

References atanh(), fused_exterior_ndeg_tm_dslash_cuda_gen::i, and z.

Referenced by atan().

Here is the call graph for this function:

Here is the caller graph for this function:

atan2()

template<typename ValueType >

__host__ __device__ ValueType quda::atan2 ( ValueType  x, ValueType  y  )

inline

Definition at line 65 of file complex_quda.h.

References x, and y.

Referenced by arg(), quda::Trig< isHalf, T >::Atan2(), atanh(), new_save_half(), and polarSu3().

Here is the caller graph for this function:

atanh() [1/2]

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::atanh ( const complex< ValueType > &  z )

inline

Definition at line 1130 of file complex_quda.h.

References atan2(), d, log(), n, ret, and z.

Referenced by atan().

Here is the call graph for this function:

Here is the caller graph for this function:

atanh() [2/2]

template<typename ValueType >

__host__ __device__ complex<float> quda::atanh ( const complex< float > &  z )

inline

Definition at line 1148 of file complex_quda.h.

References atan2f(), d, float, logf(), n, ret, and z.

Here is the call graph for this function:

ax()

void quda::ax	(	const double &	a,
		GaugeField &	u
	)

Scale the gauge field by the scalar a.

Parameters

[in]	a	scalar multiplier
[in]	u	The gauge field we want to multiply

Definition at line 322 of file gauge_field.cpp.

References a, quda::blas::ax(), b, colorSpinorParam(), and quda::ColorSpinorField::Create().

Referenced by computeHISQForceQuda(), dslashReference_5th(), dslashReference_5th_inv(), and quda::MG::generateNullVectors().

Here is the call graph for this function:

Here is the caller graph for this function:

axpy()

template<typename real , typename Link >

__device__ void quda::axpy ( real  a, const real *  x, Link &  y  )

inline

Definition at line 76 of file clover_deriv_quda.cu.

References a, deg_tm_dslash_cuda_gen::block(), blockDim, for(), fused_exterior_ndeg_tm_dslash_cuda_gen::i, x, and y.

Referenced by dslashReference_5th_inv(), quda::RitzMat::operator()(), quda::Lanczos::operator()(), quda::PreconCG::operator()(), and quda::SD::operator()().

Here is the call graph for this function:

Here is the caller graph for this function:

backSubs()

void quda::backSubs	(	const Complex *	alpha,
		Complex **const	beta,
		const double *	gamma,
		Complex *	delta,
		int	n
	)

Definition at line 131 of file inv_gcr_quda.cpp.

References delta, gamma(), and n.

Referenced by updateSolution().

Here is the call graph for this function:

Here is the caller graph for this function:

BlockOrthogonalize()

void quda::BlockOrthogonalize	(	ColorSpinorField &	V,
		int	Nvec,
		const int *	geo_bs,
		const int *	fine_to_coarse,
		int	spin_bs
	)

Block orthogonnalize the matrix field, where the blocks are defined by lookup tables that map the fine grid points to the coarse grid points, and similarly for the spin degrees of freedom.

Parameters

[in,out]	V	Matrix field to be orthgonalized
[in]	Nvec	Vector length
[in]	geo_bs	Geometric block size
[in]	fine_to_coarse	Fine-to-coarse lookup table (linear indices)
[in]	spin_bs	Spin block size

Definition at line 664 of file transfer_util.cu.

References errorQuda, QUDA_DOUBLE_PRECISION, QUDA_SINGLE_PRECISION, and V.

Referenced by quda::Transfer::Transfer().

Here is the caller graph for this function:

broadcastTuneCache()

static void quda::broadcastTuneCache ( )

static

Distribute the tunecache from node 0 to all other nodes.

Definition at line 270 of file tune.cpp.

References comm_broadcast(), comm_rank(), deserializeTuneCache(), serializeTuneCache(), and size.

Referenced by loadTuneCache(), and tuneLaunch().

Here is the call graph for this function:

Here is the caller graph for this function:

calculateY()

template<bool from_coarse, typename Float , int fineSpin, int fineColor, int coarseSpin, int coarseColor, QudaGaugeFieldOrder gOrder, typename F , typename Ftmp , typename coarseGauge , typename fineGauge , typename fineClover >

void quda::calculateY	(	coarseGauge &	Y,
		coarseGauge &	X,
		coarseGauge &	Xinv,
		Ftmp &	UV,
		F &	AV,
		F &	V,
		fineGauge &	G,
		fineClover &	C,
		fineClover &	Cinv,
		GaugeField &	Y_,
		GaugeField &	X_,
		GaugeField &	Xinv_,
		GaugeField &	Yhat_,
		ColorSpinorField &	av,
		const ColorSpinorField &	v,
		double	kappa,
		double	mu,
		double	mu_factor,
		QudaDiracType	dirac,
		QudaMatPCType	matpc
	)

Calculate the coarse-link field, include the clover field, and its inverse, and finally also compute the preconditioned coarse link field.

Parameters

Y[out]	Coarse link field accessor
X[out]	Coarse clover field accessor
Xinv[out]	Coarse clover inverse field accessor
UV[out]	Temporary accessor used to store fine link field * null space vectors
AV[out]	Temporary accessor use to store fine clover inverse * null space vectors (only applicable when fine-grid operator is the preconditioned clover operator else in general this just aliases V
V[in]	Packed null-space vector accessor
G[in]	Fine grid link / gauge field accessor
C[in]	Fine grid clover field accessor
Cinv[in]	Fine grid clover inverse field accessor
Y_[out]	Coarse link field
X_[out]	Coarse clover field
Xinv_[out]	Coarse clover field
Yhat_[out]	Preconditioned coarse link field
v[in]	Packed null-space vectors
kappa[in]	Kappa parameter
mu[in]	Twisted-mass parameter
matpc[in]	The type of preconditioning of the source fine-grid operator

Definition at line 1487 of file coarse_op.cuh.

References quda::CalculateYhat< Float, n, Arg >::apply(), arg(), quda::cublas::BatchInvertMatrix(), bidirectional_debug, quda::LatticeField::bufferIndex, checkCudaError, checkLocation, comm_dim(), comm_dim_partitioned(), COMPUTE_AV, COMPUTE_COARSE_CLOVER, COMPUTE_COARSE_LOCAL, COMPUTE_DIAGONAL, COMPUTE_REVERSE_Y, COMPUTE_TMAV, COMPUTE_TMCAV, COMPUTE_TMDIAGONAL, COMPUTE_UV, COMPUTE_VUV, quda::GaugeField::copy(), d, dirac, errorQuda, quda::GaugeField::exchangeGhost(), quda::ColorSpinorField::exchangeGhost(), quda::blas::flops, quda::cudaGaugeField::Gauge_p(), quda::cpuGaugeField::Gauge_p(), quda::ColorSpinorField::Ghost(), fused_exterior_ndeg_tm_dslash_cuda_gen::i, quda::GaugeField::injectGhost(), kappa, quda::LatticeField::Location(), matpc(), mu, mu_factor, n, quda::GaugeField::Ncolor(), quda::GaugeField::Order(), param, quda::LatticeField::Precision(), printfQuda, QUDA_BACKWARDS, QUDA_CLOVER_DIRAC, QUDA_CLOVERPC_DIRAC, QUDA_COARSE_DIRAC, QUDA_COARSEPC_DIRAC, QUDA_CPU_FIELD_LOCATION, QUDA_CUDA_FIELD_LOCATION, QUDA_FLOAT2_GAUGE_ORDER, QUDA_FORWARDS, QUDA_INVALID_PARITY, QUDA_LINK_BACKWARDS, QUDA_LINK_BIDIRECTIONAL, QUDA_LINK_FORWARDS, QUDA_MATPC_EVEN_EVEN, QUDA_MATPC_EVEN_EVEN_ASYMMETRIC, QUDA_MATPC_ODD_ODD, QUDA_MATPC_ODD_ODD_ASYMMETRIC, QUDA_MAX_DIM, QUDA_MILC_GAUGE_ORDER, QUDA_QDP_GAUGE_ORDER, QUDA_TWISTED_CLOVER_DIRAC, QUDA_TWISTED_CLOVERPC_DIRAC, QUDA_TWISTED_MASS_DIRAC, QUDA_TWISTED_MASSPC_DIRAC, V, quda::LatticeField::Volume(), quda::ColorSpinorField::X(), X, quda::LatticeField::X(), X_h, Xinv_h, and y.

Referenced by CoarseOp().

Here is the call graph for this function:

Here is the caller graph for this function:

CalculateYhatCPU()

template<typename Float , int n, typename Arg >

void quda::CalculateYhatCPU

(

Arg &

arg

)

Definition at line 1390 of file coarse_op.cuh.

References arg(), d, fused_exterior_ndeg_tm_dslash_cuda_gen::i, and parity.

Here is the call graph for this function:

CalculateYhatGPU()

template<typename Float , int n, typename Arg >

__global__ void quda::CalculateYhatGPU

(

Arg

arg

)

Definition at line 1402 of file coarse_op.cuh.

References arg(), blockDim, d, fused_exterior_ndeg_tm_dslash_cuda_gen::i, n, and parity.

Here is the call graph for this function:

canReuseResidentGauge()

bool quda::canReuseResidentGauge

(

QudaInvertParam *

inv_param

)

Check that the resident gauge field is compatible with the requested inv_param

Parameters

inv_param

Contains all metadata regarding host and device storage

Definition at line 1997 of file interface_quda.cpp.

References QudaGaugeParam_s::cuda_prec, gaugePrecise, param, and quda::LatticeField::Precision().

Here is the call graph for this function:

checkMomOrder()

void quda::checkMomOrder

(

const GaugeField &

u

)

Definition at line 19 of file copy_gauge.cu.

References errorQuda, quda::GaugeField::Order(), QUDA_FLOAT2_GAUGE_ORDER, QUDA_MILC_GAUGE_ORDER, QUDA_MILC_SITE_GAUGE_ORDER, QUDA_RECONSTRUCT_10, QUDA_RECONSTRUCT_NO, QUDA_TIFR_GAUGE_ORDER, QUDA_TIFR_PADDED_GAUGE_ORDER, and quda::GaugeField::Reconstruct().

Referenced by copyGauge().

Here is the call graph for this function:

Here is the caller graph for this function:

checkNan()

template<typename Float , int length, typename Arg >

void quda::checkNan

(

Arg

arg

)

Check whether the field contains Nans

Definition at line 62 of file copy_gauge_helper.cuh.

References arg(), d, errorQuda, fused_exterior_ndeg_tm_dslash_cuda_gen::i, length, quda::gauge::Ncolor(), parity, and x.

Here is the call graph for this function:

Checksum()

uint64_t quda::Checksum	(	const GaugeField &	u,
		bool	mini = false
	)

Compute XOR-based checksum of this gauge field: each gauge field entry is converted to type uint64_t, and compute the cummulative XOR of these values.

Parameters

[in]

mini

Whether to compute a mini checksum or global checksum. A mini checksum only computes over a subset of the lattice sites and is to be used for online comparisons, e.g., checking a field has changed with a global update algorithm.

Returns

checksum value

Definition at line 34 of file checksum.cu.

References arg(), ChecksumCPU(), errorQuda, quda::GaugeField::Order(), QUDA_BQCD_GAUGE_ORDER, QUDA_MILC_GAUGE_ORDER, QUDA_QDP_GAUGE_ORDER, QUDA_QDPJIT_GAUGE_ORDER, QUDA_TIFR_GAUGE_ORDER, and QUDA_TIFR_PADDED_GAUGE_ORDER.

Referenced by quda::GaugeField::checksum().

Here is the call graph for this function:

Here is the caller graph for this function:

ChecksumCPU()

template<typename Arg >

uint64_t quda::ChecksumCPU

(

const Arg &

arg

)

Definition at line 23 of file checksum.cu.

References arg(), d, parity, and siteChecksum().

Referenced by Checksum().

Here is the call graph for this function:

Here is the caller graph for this function:

checkUnitary()

template<typename Float2 , typename Float >

__host__ __device__ int quda::checkUnitary	(	Matrix< Float2, 3 > &	inv,
		Matrix< Float2, 3 >	in,
		const Float	tol
	)

Check the unitarity of the input matrix to a given tolerance.

Parameters

inv	The inverse of the input matrix
in	The input matrix to which we're reporting its unitarity
tol	Tolerance to which this check is applied

Definition at line 24 of file su3_project.cuh.

References computeMatrixInverse(), fabs(), fused_exterior_ndeg_tm_dslash_cuda_gen::i, in, tol, x, and y.

Referenced by polarSu3().

Here is the call graph for this function:

Here is the caller graph for this function:

checkUnitaryPrint()

template<typename Float2 >

__host__ __device__ int quda::checkUnitaryPrint	(	Matrix< Float2, 3 > &	inv,
		Matrix< Float2, 3 >	in
	)

Check the unitarity of the input matrix to a given tolerance (1e-14) and print out deviation for each component (used for debugging only).

Parameters

inv	The inverse of the input matrix
in	The input matrix to which we're reporting its unitarity

Definition at line 47 of file su3_project.cuh.

References computeMatrixInverse(), e, fabs(), fused_exterior_ndeg_tm_dslash_cuda_gen::i, in, printf(), x, and y.

Here is the call graph for this function:

cloverApply()

template<typename Float , int nSpin, int nColor, typename Arg >

__device__ __host__ void quda::cloverApply ( Arg &  arg, int  x_cb, int  parity  )

inline

Definition at line 485 of file dslash_quda.cu.

References arg(), in, nColor, out, and parity.

Here is the call graph for this function:

cloverCPU()

template<typename Float , int nSpin, int nColor, typename Arg >

void quda::cloverCPU

(

Arg &

arg

)

Definition at line 505 of file dslash_quda.cu.

References arg(), for(), and parity.

Here is the call graph for this function:

cloverDerivative()

void quda::cloverDerivative	(	cudaGaugeField &	force,
		cudaGaugeField &	gauge,
		cudaGaugeField &	oprod,
		double	coeff,
		QudaParity	parity
	)

Compute the derivative of the clover matrix in the direction mu,nu and compute the resulting force given the outer-product field.

Parameters

force	The computed force field (read/write update)
gauge	The input gauge field
oprod	The input outer-product field (tensor matrix field)
coeff	Multiplicative coefficient (e.g., clover coefficient)
parity	The field parity we are working on

Definition at line 519 of file clover_deriv_quda.cu.

References dw_dslash_4D_cuda_gen::coeff(), d, errorQuda, quda::GaugeField::Geometry(), parity, quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, QUDA_EVEN_PARITY, QUDA_SINGLE_PRECISION, QUDA_TENSOR_GEOMETRY, QUDA_VECTOR_GEOMETRY, and quda::LatticeField::X().

Referenced by computeCloverForceQuda().

Here is the call graph for this function:

Here is the caller graph for this function:

cloverDslashCuda()

void quda::cloverDslashCuda	(	cudaColorSpinorField *	out,
		const cudaGaugeField &	gauge,
		const FullClover &	cloverInv,
		const cudaColorSpinorField *	in,
		const int	oddBit,
		const int	daggerBit,
		const cudaColorSpinorField *	x,
		const double &	k,
		const int *	commDim,
		TimeProfile &	profile
	)

Definition at line 175 of file dslash_clover.cu.

References a, deg_tm_dslash_cuda_gen::dagger, deg_tm_dslash_cuda_gen::dslash, errorQuda, in, out, parity, QUDA_DOUBLE_PRECISION, QUDA_HALF_PRECISION, QUDA_SINGLE_PRECISION, and x.

Referenced by quda::DiracCloverPC::Dslash(), and quda::DiracCloverPC::DslashXpay().

Here is the caller graph for this function:

cloverGPU()

template<typename Float , int nSpin, int nColor, typename Arg >

__global__ void quda::cloverGPU

(

Arg

arg

)

Definition at line 513 of file dslash_quda.cu.

References arg(), blockDim, if(), and parity.

Here is the call graph for this function:

cloverInvert()

void quda::cloverInvert	(	CloverField &	clover,
		bool	computeTraceLog,
		QudaFieldLocation	location
	)

This function compute the Cholesky decomposition of each clover matrix and stores the clover inverse field.

Parameters

clover	The clover field (contains both the field itself and its inverse)
computeTraceLog	Whether to compute the trace logarithm of the clover term
location	The location of the field

Definition at line 183 of file clover_invert.cu.

References dslash_cuda_gen::clover, errorQuda, QUDA_DOUBLE_PRECISION, QUDA_HALF_PRECISION, and QUDA_SINGLE_PRECISION.

Referenced by loadCloverQuda().

Here is the caller graph for this function:

cloverRho()

void quda::cloverRho	(	CloverField &	clover,
		double	rho
	)

This function adds a real scalar onto the clover diagonal (only to the direct field not the inverse)

Parameters

clover	The clover field
rho	Real scalar to be added on

CoarseCoarseOp()

void quda::CoarseCoarseOp	(	GaugeField &	Y,
		GaugeField &	X,
		GaugeField &	Xinv,
		GaugeField &	Yhat,
		const Transfer &	T,
		const GaugeField &	gauge,
		const GaugeField &	clover,
		const GaugeField &	cloverInv,
		double	kappa,
		double	mu,
		double	mu_factor,
		QudaDiracType	dirac,
		QudaMatPCType	matpc
	)

Coarse operator construction from an intermediate-grid operator (Coarse)

Parameters

Y[out]	Coarse link field
X[out]	Coarse clover field
Xinv[out]	Coarse clover inverse field
Y[out]	Preconditioned coarse link field
T[in]	Transfer operator that defines the new coarse space
gauge[in]	Link field from fine grid
clover[in]	Clover field on fine grid
cloverInv[in]	Clover inverse field on fine grid
kappa[in]	Kappa parameter
mu[in]	Mu parameter (set to non-zero for twisted-mass/twisted-clover)
mu_factor[in]	Multiplicative factor for the mu parameter
matpc[in]	The type of even-odd preconditioned fine-grid operator we are constructing the coarse grid operator from. If matpc==QUDA_MATPC_INVALID then we assume the operator is not even-odd preconditioned and we coarsen the full operator.

Definition at line 169 of file coarsecoarse_op.cu.

References checkLocation, dslash_cuda_gen::clover, quda::ColorSpinorParam::create, quda::ColorSpinorField::Create(), dirac, errorQuda, kappa, matpc(), mu, mu_factor, quda::LatticeField::Precision(), QUDA_ZERO_FIELD_CREATE, quda::Transfer::Vectors(), and X.

Referenced by quda::DiracCoarse::createCoarseOp(), and quda::DiracCoarsePC::createCoarseOp().

Here is the call graph for this function:

Here is the caller graph for this function:

CoarseOp()

void quda::CoarseOp	(	GaugeField &	Y,
		GaugeField &	X,
		GaugeField &	Xinv,
		GaugeField &	Yhat,
		const Transfer &	T,
		const cudaGaugeField &	gauge,
		const cudaCloverField *	clover,
		double	kappa,
		double	mu,
		double	mu_factor,
		QudaDiracType	dirac,
		QudaMatPCType	matpc
	)

Coarse operator construction from a fine-grid operator (Wilson / Clover)

Parameters

Y[out]	Coarse link field
X[out]	Coarse clover field
Xinv[out]	Coarse clover inverse field
Yhat[out]	Preconditioned coarse link field
T[in]	Transfer operator that defines the coarse space
gauge[in]	Gauge field from fine grid
clover[in]	Clover field on fine grid (optional)
kappa[in]	Kappa parameter
mu[in]	Mu parameter (set to non-zero for twisted-mass/twisted-clover)
mu_factor[in]	Multiplicative factor for the mu parameter
matpc[in]	The type of even-odd preconditioned fine-grid operator we are constructing the coarse grid operator from. If matpc==QUDA_MATPC_INVALID then we assume the operator is not even-odd preconditioned and we coarsen the full operator.

Definition at line 170 of file coarse_op.cu.

References quda::GaugeField::Anisotropy(), calculateY(), checkLocation, quda::CloverFieldParam::clover, dslash_cuda_gen::clover, quda::CloverFieldParam::cloverInv, quda::GaugeField::copy(), quda::CloverFieldParam::create, quda::ColorSpinorParam::create, quda::ColorSpinorField::Create(), dirac, quda::CloverFieldParam::direct, errorQuda, quda::GaugeField::GaugeFixed(), quda::GaugeField::Geometry(), fused_exterior_ndeg_tm_dslash_cuda_gen::i, quda::CloverFieldParam::inverse, quda::CloverFieldParam::invNorm, kappa, quda::GaugeField::LinkType(), matpc(), mu, mu_factor, quda::LatticeFieldParam::nDim, quda::CloverFieldParam::norm, quda::GaugeFieldParam::order, quda::CloverFieldParam::order, quda::LatticeFieldParam::pad, quda::LatticeFieldParam::precision, quda::LatticeField::Precision(), QUDA_CPU_FIELD_LOCATION, QUDA_CUDA_FIELD_LOCATION, QUDA_FULL_SITE_SUBSET, QUDA_GHOST_EXCHANGE_PAD, QUDA_INVALID_CLOVER_ORDER, QUDA_INVALID_PRECISION, QUDA_MATPC_INVALID, QUDA_NULL_FIELD_CREATE, QUDA_PACKED_CLOVER_ORDER, QUDA_QDP_GAUGE_ORDER, QUDA_RECONSTRUCT_NO, QUDA_TWISTED_MASSPC_DIRAC, QUDA_ZERO_FIELD_CREATE, quda::GaugeFieldParam::reconstruct, quda::GaugeField::Reconstruct(), quda::cudaGaugeField::saveCPUField(), quda::GaugeFieldParam::setPrecision(), quda::LatticeFieldParam::siteSubset, quda::GaugeField::TBoundary(), quda::Transfer::Vectors(), quda::LatticeFieldParam::x, X, and quda::LatticeField::X().

Referenced by quda::DiracWilson::createCoarseOp(), quda::DiracClover::createCoarseOp(), quda::DiracCloverPC::createCoarseOp(), quda::DiracTwistedMass::createCoarseOp(), quda::DiracTwistedMassPC::createCoarseOp(), quda::DiracTwistedClover::createCoarseOp(), and quda::DiracTwistedCloverPC::createCoarseOp().

Here is the call graph for this function:

Here is the caller graph for this function:

colorSpinorParam() [1/2]

ColorSpinorParam quda::colorSpinorParam

(

const GaugeField &

a

)

Definition at line 277 of file gauge_field.cpp.

References a, quda::ColorSpinorParam::create, d, errorQuda, quda::ColorSpinorParam::fieldOrder, quda::ColorSpinorParam::gammaBasis, quda::ColorSpinorParam::location, quda::ColorSpinorParam::nColor, quda::LatticeFieldParam::nDim, quda::ColorSpinorParam::nSpin, quda::LatticeFieldParam::pad, quda::LatticeFieldParam::precision, QUDA_COARSE_LINKS, QUDA_DOUBLE_PRECISION, QUDA_EVEN_ODD_SITE_ORDER, QUDA_FLOAT2_FIELD_ORDER, QUDA_FLOAT4_FIELD_ORDER, QUDA_FULL_SITE_SUBSET, QUDA_HALF_PRECISION, QUDA_QDP_GAUGE_ORDER, QUDA_QDPJIT_GAUGE_ORDER, QUDA_REFERENCE_FIELD_CREATE, QUDA_UKQCD_GAMMA_BASIS, quda::ColorSpinorParam::siteOrder, quda::LatticeFieldParam::siteSubset, quda::ColorSpinorParam::v, and quda::LatticeFieldParam::x.

colorSpinorParam() [2/2]

ColorSpinorParam quda::colorSpinorParam	(	const CloverField &	a,
		bool	inverse
	)

Definition at line 422 of file clover_field.cpp.

References a, quda::ColorSpinorParam::create, d, errorQuda, quda::ColorSpinorParam::fieldOrder, quda::ColorSpinorParam::gammaBasis, quda::ColorSpinorParam::location, quda::ColorSpinorParam::nColor, quda::LatticeFieldParam::nDim, quda::ColorSpinorParam::nSpin, quda::LatticeFieldParam::pad, quda::LatticeFieldParam::precision, QUDA_DOUBLE_PRECISION, QUDA_EVEN_ODD_SITE_ORDER, QUDA_FLOAT2_FIELD_ORDER, QUDA_FLOAT4_FIELD_ORDER, QUDA_FULL_SITE_SUBSET, QUDA_HALF_PRECISION, QUDA_REFERENCE_FIELD_CREATE, QUDA_UKQCD_GAMMA_BASIS, quda::ColorSpinorParam::siteOrder, quda::LatticeFieldParam::siteSubset, quda::ColorSpinorParam::v, and quda::LatticeFieldParam::x.

Referenced by ax(), norm1(), and norm2().

Here is the caller graph for this function:

compareSpinor()

template<class U , class V >

int quda::compareSpinor	(	const U &	u,
		const V &	v,
		const int	tol
	)

Definition at line 147 of file color_spinor_util.cu.

References c, comm_allreduce_int(), comm_size(), e, f, fabs(), fused_exterior_ndeg_tm_dslash_cuda_gen::i, parity, pow(), printfQuda, s, tol, total, and z.

Referenced by genericCompare().

Here is the call graph for this function:

Here is the caller graph for this function:

completeKSForce() [1/2]

void quda::completeKSForce	(	GaugeField &	mom,
		const GaugeField &	oprod,
		const GaugeField &	gauge,
		QudaFieldLocation	location,
		long long *	flops = NULL
	)

Definition at line 177 of file ks_force_quda.cu.

References errorQuda, quda::blas::flops, QUDA_CUDA_FIELD_LOCATION, QUDA_RECONSTRUCT_10, QUDA_RECONSTRUCT_NO, quda::GaugeField::Reconstruct(), and quda::LatticeField::X().

Here is the call graph for this function:

completeKSForce() [2/2]

template<typename Float , typename Oprod , typename Gauge , typename Mom >

void quda::completeKSForce	(	Oprod	oprod,
		Gauge	gauge,
		Mom	mom,
		int	dim[4],
		const GaugeField &	meta,
		QudaFieldLocation	location,
		long long *	flops
	)

Definition at line 166 of file ks_force_quda.cu.

References quda::KSForceComplete< Float, Oprod, Gauge, Mom >::apply(), arg(), dim, quda::blas::flops, quda::KSForceComplete< Float, Oprod, Gauge, Mom >::flops(), and qudaDeviceSynchronize().

Here is the call graph for this function:

completeKSForceCore()

template<typename Float , typename Oprod , typename Gauge , typename Mom >

__host__ __device__ void quda::completeKSForceCore	(	KSForceArg< Oprod, Gauge, Mom > &	arg,
		int	idx
	)

Definition at line 44 of file ks_force_quda.cu.

References arg(), quda::Matrix< T, N >::data, getCoords(), getTrace(), idx, linkIndexShift(), parity, sub(), X, and x.

Here is the call graph for this function:

completeKSForceCPU()

template<typename Float , typename Oprod , typename Gauge , typename Mom >

void quda::completeKSForceCPU

(

KSForceArg< Oprod, Gauge, Mom > &

arg

)

Definition at line 116 of file ks_force_quda.cu.

References arg(), and idx.

Here is the call graph for this function:

completeKSForceKernel()

template<typename Float , typename Oprod , typename Gauge , typename Mom >

__global__ void quda::completeKSForceKernel

(

KSForceArg< Oprod, Gauge, Mom >

arg

)

Definition at line 104 of file ks_force_quda.cu.

References arg(), blockDim, and idx.

Here is the call graph for this function:

computeAV()

template<typename Float , int fineSpin, int fineColor, int coarseColor, typename Arg >

__device__ __host__ void quda::computeAV ( Arg &  arg, int  parity, int  x_cb, int  ic_c  )

inline

Calculates the matrix A V^{s,c'}(x) = A^{c}(x) * V^{s,c}(x) Where: s = fine spin, c' = coarse color, c = fine color

Definition at line 157 of file coarse_op.cuh.

References arg(), c, parity, and s.

Here is the call graph for this function:

ComputeAVCPU()

template<typename Float , int fineSpin, int fineColor, int coarseColor, typename Arg >

void quda::ComputeAVCPU

(

Arg &

arg

)

Definition at line 184 of file coarse_op.cuh.

References arg(), and parity.

Here is the call graph for this function:

ComputeAVGPU()

template<typename Float , int fineSpin, int fineColor, int coarseColor, typename Arg >

__global__ void quda::ComputeAVGPU

(

Arg

arg

)

Definition at line 194 of file coarse_op.cuh.

References arg(), blockDim, and parity.

Here is the call graph for this function:

computeBeta()

void quda::computeBeta	(	Complex **	beta,
		std::vector< ColorSpinorField *>	Ap,
		int	i,
		int	N,
		int	k
	)

Definition at line 50 of file inv_gcr_quda.cpp.

References a, b, quda::blas::cDotProduct(), fused_exterior_ndeg_tm_dslash_cuda_gen::i, and printfQuda.

Referenced by orthoDir().

Here is the call graph for this function:

Here is the caller graph for this function:

computeClover()

void quda::computeClover	(	CloverField &	clover,
		const GaugeField &	gauge,
		double	coeff,
		QudaFieldLocation	location
	)

Definition at line 204 of file clover_quda.cu.

References dslash_cuda_gen::clover, errorQuda, f, QUDA_DOUBLE_PRECISION, and QUDA_SINGLE_PRECISION.

Referenced by quda::cudaCloverField::compute(), and createCloverQuda().

Here is the caller graph for this function:

computeCloverForce()

void quda::computeCloverForce	(	GaugeField &	force,
		const GaugeField &	U,
		std::vector< ColorSpinorField *> &	x,
		std::vector< ColorSpinorField *> &	p,
		std::vector< double > &	coeff
	)

Compute the force contribution from the solver solution fields.

Force(x, mu) = U(x, mu) * sum_i=1^nvec ( P_mu^+ x(x+mu) p(x)^ + P_mu^- p(x+mu) x(x)^ )

M = A_even - kappa^2 * Dslash * A_odd^{-1} * Dslash x(even) = M^{-1} b(even) x(odd) = A_odd^{-1} * Dslash * x(even) p(even) = M * x(even) p(odd) = A_odd^{-1} * Dslash^dag * M * x(even).

Parameters

force[out,in]	The resulting force field
U	The input gauge field
x	Solution field (both parities)
p	Intermediate vectors (both parities)
coeff	Multiplicative coefficient (e.g., dt * residue)

Definition at line 468 of file clover_outer_product.cu.

References checkCudaError, dw_dslash_4D_cuda_gen::coeff(), errorQuda, quda::ColorSpinorField::GhostFace(), fused_exterior_ndeg_tm_dslash_cuda_gen::i, quda::GaugeField::Order(), p, parity, quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, QUDA_FLOAT2_GAUGE_ORDER, QUDA_RECONSTRUCT_12, QUDA_RECONSTRUCT_NO, QUDA_SINGLE_PRECISION, quda::GaugeField::Reconstruct(), and x.

Referenced by computeCloverForceQuda().

Here is the call graph for this function:

Here is the caller graph for this function:

computeCloverSigmaOprod()

void quda::computeCloverSigmaOprod	(	GaugeField &	oprod,
		std::vector< ColorSpinorField *> &	x,
		std::vector< ColorSpinorField *> &	p,
		std::vector< std::vector< double > > &	coeff
	)

Compute the outer product from the solver solution fields arising from the diagonal term of the fermion bilinear in direction mu,nu and sum to outer product field.

Parameters

oprod[out,in]	Computed outer product field (tensor matrix field)
x[in]	Solution field (both parities)
p[in]	Intermediate vectors (both parities) coeff[in] Multiplicative coefficient (e.g., dt * residiue), one for each parity

Definition at line 178 of file clover_sigma_outer_product.cu.

References checkCudaError, dw_dslash_4D_cuda_gen::coeff(), errorQuda, fused_exterior_ndeg_tm_dslash_cuda_gen::i, quda::GaugeField::Order(), p, quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, QUDA_FLOAT2_GAUGE_ORDER, Spinor< RegType, StoreType, N, write, tex_id >::set(), and x.

Referenced by computeCloverForceQuda().

Here is the call graph for this function:

Here is the caller graph for this function:

computeCloverSigmaTrace()

void quda::computeCloverSigmaTrace	(	GaugeField &	output,
		const CloverField &	clover,
		double	coeff
	)

Compute the matrix tensor field necessary for the force calculation from the clover trace action. This computes a tensor field [mu,nu].

Parameters

output	The computed matrix field (tensor matrix field)
clover	The input clover field
coeff	Scalar coefficient multiplying the result (e.g., stepsize)

Definition at line 242 of file clover_trace_quda.cu.

References dslash_cuda_gen::clover, dw_dslash_4D_cuda_gen::coeff(), errorQuda, QUDA_DOUBLE_PRECISION, and QUDA_SINGLE_PRECISION.

Referenced by computeCloverForceQuda().

Here is the call graph for this function:

Here is the caller graph for this function:

computeCoarseClover()

template<bool from_coarse, typename Float , int fineSpin, int coarseSpin, int fineColor, int coarseColor, typename Arg >

__device__ __host__ void quda::computeCoarseClover	(	Arg &	arg,
		int	parity,
		int	x_cb,
		int	ic_c
	)

Definition at line 748 of file coarse_op.cuh.

References arg(), conj(), coord, d, for(), getCoords(), fused_exterior_ndeg_tm_dslash_cuda_gen::i, parity, QUDA_MAX_DIM, s, and X.

Here is the call graph for this function:

ComputeCoarseCloverCPU()

template<bool from_coarse, typename Float , int fineSpin, int coarseSpin, int fineColor, int coarseColor, typename Arg >

void quda::ComputeCoarseCloverCPU

(

Arg &

arg

)

Definition at line 822 of file coarse_op.cuh.

References arg(), and parity.

Here is the call graph for this function:

ComputeCoarseCloverGPU()

template<bool from_coarse, typename Float , int fineSpin, int coarseSpin, int fineColor, int coarseColor, typename Arg >

__global__ void quda::ComputeCoarseCloverGPU

(

Arg

arg

)

Definition at line 833 of file coarse_op.cuh.

References arg(), blockDim, and parity.

Here is the call graph for this function:

computeCoarseLocal()

template<bool bidirectional, typename Float , int nSpin, int nColor, typename Arg >

__device__ __host__ void quda::computeCoarseLocal	(	Arg &	arg,
		int	parity,
		int	x_cb
	)

Adds the reverse links to the coarse local term, which is just the conjugate of the existing coarse local term but with plus/minus signs for off-diagonal spin components so multiply by the appropriate factor of -kappa.

Definition at line 686 of file coarse_op.cuh.

References arg(), conj(), nColor, parity, and deg_tm_dslash_cuda_gen::sign().

Here is the call graph for this function:

ComputeCoarseLocalCPU()

template<bool bidirectional, typename Float , int nSpin, int nColor, typename Arg >

void quda::ComputeCoarseLocalCPU

(

Arg &

arg

)

Definition at line 729 of file coarse_op.cuh.

References arg(), and parity.

Here is the call graph for this function:

ComputeCoarseLocalGPU()

template<bool bidirectional, typename Float , int nSpin, int nColor, typename Arg >

__global__ void quda::ComputeCoarseLocalGPU

(

Arg

arg

)

Definition at line 738 of file coarse_op.cuh.

References arg(), blockDim, and parity.

Here is the call graph for this function:

computeCoeffs()

template<typename T >

static void quda::computeCoeffs ( T  d_out[], const T  d_p1[], const T  d_p2[], int  k, int  j, int  s, const T  gamma[], const T  rho[], const T  gamma_kprev[], const T  rho_kprev[]  )

static

Definition at line 79 of file inv_mpcg_quda.cpp.

References applyThirdTerm(), gamma(), fused_exterior_ndeg_tm_dslash_cuda_gen::i, and s.

Referenced by quda::MPCG::operator()().

Here is the call graph for this function:

Here is the caller graph for this function:

ComputeEta()

template<libtype which_lib>

void quda::ComputeEta

(

GMResDRArgs &

args

)

Definition at line 157 of file inv_gmresdr_quda.cpp.

References errorQuda.

ComputeEta< libtype::eigen_lib >()

template<>

void quda::ComputeEta< libtype::eigen_lib >

(

GMResDRArgs &

args

)

Definition at line 179 of file inv_gmresdr_quda.cpp.

References args.

ComputeEta< libtype::magma_lib >()

template<>

void quda::ComputeEta< libtype::magma_lib >

(

GMResDRArgs &

args

)

Definition at line 159 of file inv_gmresdr_quda.cpp.

References args, errorQuda, magma_Xgels(), memcpy(), and memset().

Here is the call graph for this function:

computeFmunu()

void quda::computeFmunu	(	GaugeField &	Fmunu,
		const GaugeField &	gauge,
		QudaFieldLocation	location
	)

Compute the Fmunu tensor

Parameters

Fmunu	The Fmunu tensor
gauge	The gauge field upon which to compute the Fmnu tensor
location	The location of where to do the computation

Definition at line 283 of file field_strength_tensor.cu.

References errorQuda, quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, and QUDA_SINGLE_PRECISION.

Referenced by createCloverQuda(), and qChargeCuda().

Here is the call graph for this function:

Here is the caller graph for this function:

ComputeHarmonicRitz()

template<libtype which_lib>

void quda::ComputeHarmonicRitz

(

GMResDRArgs &

args

)

Definition at line 88 of file inv_gmresdr_quda.cpp.

References errorQuda.

ComputeHarmonicRitz< libtype::eigen_lib >()

template<>

void quda::ComputeHarmonicRitz< libtype::eigen_lib >

(

GMResDRArgs &

args

)

Definition at line 127 of file inv_gmresdr_quda.cpp.

References abs(), args, e, memcpy(), quda::blas::norm(), and quda::SortedEvals::SelectSmall().

Here is the call graph for this function:

ComputeHarmonicRitz< libtype::magma_lib >()

template<>

void quda::ComputeHarmonicRitz< libtype::magma_lib >

(

GMResDRArgs &

args

)

Definition at line 90 of file inv_gmresdr_quda.cpp.

References abs(), args, e, errorQuda, magma_Xgeev(), magma_Xgesv(), memcpy(), quda::blas::norm(), and quda::SortedEvals::SelectSmall().

Here is the call graph for this function:

computeKSLongLinkForce() [1/2]

template<typename Float , typename Result , typename Oprod , typename Gauge >

void quda::computeKSLongLinkForce	(	Result	res,
		Oprod	oprod,
		Gauge	gauge,
		int	dim[4],
		const GaugeField &	meta,
		QudaFieldLocation	location
	)

Definition at line 378 of file ks_force_quda.cu.

References quda::KSLongLinkForce< Float, Result, Oprod, Gauge >::apply(), arg(), dim, and qudaDeviceSynchronize().

Here is the call graph for this function:

computeKSLongLinkForce() [2/2]

template<typename Float >

void quda::computeKSLongLinkForce	(	GaugeField &	result,
		const GaugeField &	oprod,
		const GaugeField &	gauge,
		QudaFieldLocation	location
	)

Definition at line 387 of file ks_force_quda.cu.

References errorQuda, QUDA_CUDA_FIELD_LOCATION, QUDA_RECONSTRUCT_10, QUDA_RECONSTRUCT_NO, quda::GaugeField::Reconstruct(), and quda::LatticeField::X().

Here is the call graph for this function:

computeKSLongLinkForceCore()

template<typename Float , typename Result , typename Oprod , typename Gauge >

__host__ __device__ void quda::computeKSLongLinkForceCore	(	KSLongLinkArg< Result, Oprod, Gauge > &	arg,
		int	idx
	)

Definition at line 247 of file ks_force_quda.cu.

computeKSLongLinkForceCPU()

template<typename Float , typename Result , typename Oprod , typename Gauge >

void quda::computeKSLongLinkForceCPU

(

KSLongLinkArg< Result, Oprod, Gauge > &

arg

)

Definition at line 323 of file ks_force_quda.cu.

References arg(), and idx.

Here is the call graph for this function:

computeKSLongLinkForceKernel()

template<typename Float , typename Result , typename Oprod , typename Gauge >

__global__ void quda::computeKSLongLinkForceKernel

(

KSLongLinkArg< Result, Oprod, Gauge >

arg

)

Definition at line 311 of file ks_force_quda.cu.

References arg(), blockDim, and idx.

Here is the call graph for this function:

computeLinkInverse()

template<class Cmplx >

__device__ __host__ void quda::computeLinkInverse ( Matrix< Cmplx, 3 > *  uinv, const Matrix< Cmplx, 3 > &  u  )

inline

Definition at line 913 of file quda_matrix.h.

References getDeterminant().

Here is the call graph for this function:

computeMatrixInverse()

template<class T >

__device__ __host__ void quda::computeMatrixInverse ( const Matrix< T, 3 > &  u, Matrix< T, 3 > *  uinv  )

inline

Definition at line 501 of file quda_matrix.h.

References getDeterminant().

Referenced by checkUnitary(), checkUnitaryPrint(), computeOvrImpSTOUTStep(), and polarSu3().

Here is the call graph for this function:

Here is the caller graph for this function:

computeMomAction()

double quda::computeMomAction

(

const GaugeField &

mom

)

Compute and return global the momentum action 1/2 mom^2.

Parameters

mom	Momentum field

Returns

Momentum action contribution

Definition at line 113 of file momentum.cu.

References errorQuda, quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, and QUDA_SINGLE_PRECISION.

Referenced by momActionQuda().

Here is the call graph for this function:

Here is the caller graph for this function:

computeNeighborSum()

template<typename Float , int Nc, typename Vector , typename Arg >

__device__ __host__ void quda::computeNeighborSum ( Vector &  out, Arg &  arg, int  x_cb, int  parity  )

inline

Computes out = sum_mu U_mu(x)in(x+d) + U^(x-d)in(x-d)

Parameters

[out]	out	The out result field
[in]	U	The gauge field
[in]	in	The input field
[in]	x_cb	The checkerboarded site index
[in]	parity	The site parity

Definition at line 52 of file color_spinor_wuppertal.cu.

References arg(), conj(), coord, getCoords(), in, linkIndexM1(), linkIndexP1(), out, and parity.

Here is the call graph for this function:

computeOvrImpSTOUTStep()

template<typename Float , typename GaugeOr , typename GaugeDs >

__global__ void quda::computeOvrImpSTOUTStep

(

GaugeOvrImpSTOUTArg< Float, GaugeOr, GaugeDs >

arg

)

Definition at line 598 of file gauge_stout.cu.

References arg(), blockDim, computeMatrixInverse(), conj(), ErrorSU3(), exponentiate_iQ(), getCoords(), getTrace(), idx, linkIndexShift(), parity, printf(), setIdentity(), X, and x.

Referenced by quda::GaugeOvrImpSTOUT< Float, GaugeOr, GaugeDs >::apply().

Here is the call graph for this function:

Here is the caller graph for this function:

computeQCharge()

double quda::computeQCharge	(	GaugeField &	Fmunu,
		QudaFieldLocation	location
	)

Compute the topological charge

Parameters

Fmunu	The Fmunu tensor, usually calculated from a smeared configuration
location	The location of where to do the computation, currently supports only the GPU

Definition at line 143 of file qcharge_quda.cu.

References errorQuda, quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, and QUDA_SINGLE_PRECISION.

Referenced by qChargeCuda().

Here is the call graph for this function:

Here is the caller graph for this function:

ComputeRitz()

template<libtype which_lib>

void quda::ComputeRitz

(

EigCGArgs &

args

)

Definition at line 133 of file inv_eigcg_quda.cpp.

References errorQuda.

ComputeRitz< libtype::eigen_lib >()

template<>

void quda::ComputeRitz< libtype::eigen_lib >

(

EigCGArgs &

args

)

Definition at line 136 of file inv_eigcg_quda.cpp.

References args.

ComputeRitz< libtype::magma_lib >()

template<>

void quda::ComputeRitz< libtype::magma_lib >

(

EigCGArgs &

args

)

Definition at line 164 of file inv_eigcg_quda.cpp.

References args, errorQuda, magma_Xheev(), and memcpy().

Here is the call graph for this function:

computeStaggeredOprod() [1/2]

void quda::computeStaggeredOprod	(	GaugeField *	out[],
		ColorSpinorField &	in,
		const double	coeff[],
		int	nFace
	)

Compute the outer-product field between the staggered quark field's one and (for HISQ and ASQTAD) three hop sites. E.g.,.

out[0][d](x) = (in(x+1_d) x conj(in(x))) out[1][d](x) = (in(x+3_d) x conj(in(x)))

where 1_d and 3_d represent a relative shift of magnitude 1 and 3 in dimension d, respectively

Note out[1] is only computed if nFace=3

Parameters

[out]	out	Array of nFace outer-product matrix fields
[in]	in	Input quark field
[in]	coeff	Coefficient
[in]	nFace	Number of faces (1 or 3)

Definition at line 451 of file staggered_oprod.cu.

References dw_dslash_4D_cuda_gen::coeff(), errorQuda, quda::ColorSpinorField::Even(), in, quda::ColorSpinorField::Odd(), and out.

Referenced by computeHISQForceQuda(), and computeStaggeredForceQuda().

Here is the call graph for this function:

Here is the caller graph for this function:

computeStaggeredOprod() [2/2]

void quda::computeStaggeredOprod	(	GaugeField &	outA,
		GaugeField &	outB,
		ColorSpinorField &	inEven,
		ColorSpinorField &	inOdd,
		const unsigned int	parity,
		const double	coeff[2],
		int	nFace
	)

Definition at line 408 of file staggered_oprod.cu.

References quda::cudaColorSpinorField::allocateGhostBuffer(), dw_dslash_4D_cuda_gen::coeff(), errorQuda, quda::GaugeField::Order(), parity, quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, QUDA_FLOAT2_GAUGE_ORDER, and QUDA_SINGLE_PRECISION.

Here is the call graph for this function:

computeStapleRectangle()

template<typename Float , typename GaugeOr , typename GaugeDs , typename Float2 >

__host__ __device__ void quda::computeStapleRectangle	(	GaugeOvrImpSTOUTArg< Float, GaugeOr, GaugeDs > &	arg,
		int	idx,
		int	parity,
		int	dir,
		Matrix< Float2, 3 > &	staple,
		Matrix< Float2, 3 > &	rectangle
	)

Definition at line 362 of file gauge_stout.cu.

References arg(), conj(), getCoords(), idx, linkIndexShift(), mu, parity, setZero(), X, and x.

Here is the call graph for this function:

computeTMAV()

template<typename Float , int fineSpin, int fineColor, int coarseColor, typename Arg >

__device__ __host__ void quda::computeTMAV ( Arg &  arg, int  parity, int  x_cb, int  v  )

inline

Calculates the matrix A V^{s,c'}(x) = A^{c}(x) * V^{s,c}(x) for twisted-mass fermions Where: s = fine spin, c' = coarse color, c = fine color

Definition at line 209 of file coarse_op.cuh.

References arg(), c, parity, and s.

Here is the call graph for this function:

ComputeTMAVCPU()

template<typename Float , int fineSpin, int fineColor, int coarseColor, typename Arg >

void quda::ComputeTMAVCPU

(

Arg &

arg

)

Definition at line 229 of file coarse_op.cuh.

References arg(), and parity.

Here is the call graph for this function:

ComputeTMAVGPU()

template<typename Float , int fineSpin, int fineColor, int coarseColor, typename Arg >

__global__ void quda::ComputeTMAVGPU

(

Arg

arg

)

Definition at line 239 of file coarse_op.cuh.

References arg(), blockDim, and parity.

Here is the call graph for this function:

computeTMCAV()

template<typename Float , int fineSpin, int fineColor, int coarseColor, typename Arg >

__device__ __host__ void quda::computeTMCAV ( Arg &  arg, int  parity, int  x_cb  )

inline

Calculates the matrix A V^{s,c'}(x) = A^{c}(x) * V^{s,c}(x) for twisted-clover fermions Where: s = fine spin, c' = coarse color, c = fine color

Definition at line 391 of file coarse_op.cuh.

References arg(), c, mu, parity, and s.

Here is the call graph for this function:

ComputeTMCAVCPU()

template<typename Float , int fineSpin, int fineColor, int coarseColor, typename Arg >

void quda::ComputeTMCAVCPU

(

Arg &

arg

)

Definition at line 465 of file coarse_op.cuh.

References arg(), and parity.

Here is the call graph for this function:

ComputeTMCAVGPU()

template<typename Float , int fineSpin, int fineColor, int coarseColor, typename Arg >

__global__ void quda::ComputeTMCAVGPU

(

Arg

arg

)

Definition at line 474 of file coarse_op.cuh.

References arg(), blockDim, and parity.

Here is the call graph for this function:

computeUV()

template<bool from_coarse, typename Float , int dim, QudaDirection dir, int fineSpin, int fineColor, int coarseSpin, int coarseColor, typename Arg >

__device__ __host__ void quda::computeUV ( Arg &  arg, int  parity, int  x_cb, int  ic_c  )

inline

Calculates the matrix UV^{s,c'}_mu(x) = U^{c}_mu(x) * V^{s,c}_mu(x+mu) Where: mu = dir, s = fine spin, c' = coarse color, c = fine color

Definition at line 62 of file coarse_op.cuh.

References arg(), c, coord, dim, getCoords(), linkIndexP1(), parity, QUDA_FORWARDS, and s.

Here is the call graph for this function:

ComputeUVCPU()

template<bool from_coarse, typename Float , int dim, QudaDirection dir, int fineSpin, int fineColor, int coarseSpin, int coarseColor, typename Arg >

void quda::ComputeUVCPU

(

Arg &

arg

)

Definition at line 132 of file coarse_op.cuh.

References arg(), and parity.

Here is the call graph for this function:

ComputeUVGPU()

template<bool from_coarse, typename Float , int dim, QudaDirection dir, int fineSpin, int fineColor, int coarseSpin, int coarseColor, typename Arg >

__global__ void quda::ComputeUVGPU

(

Arg

arg

)

Definition at line 142 of file coarse_op.cuh.

References arg(), blockDim, and parity.

Here is the call graph for this function:

computeVUV()

template<bool from_coarse, typename Float , int dim, QudaDirection dir, int fineSpin, int fineColor, int coarseSpin, int coarseColor, typename Arg >

__device__ __host__ void quda::computeVUV	(	Arg &	arg,
		int	parity,
		int	x_cb,
		int	c_row
	)

Definition at line 570 of file coarse_op.cuh.

References arg(), coord, d, dim, getCoords(), parity, QUDA_BACKWARDS, and QUDA_MAX_DIM.

Here is the call graph for this function:

ComputeVUVCPU()

template<bool from_coarse, typename Float , int dim, QudaDirection dir, int fineSpin, int fineColor, int coarseSpin, int coarseColor, typename Arg >

void quda::ComputeVUVCPU

(

Arg

arg

)

Definition at line 614 of file coarse_op.cuh.

References arg(), and parity.

Here is the call graph for this function:

ComputeVUVGPU()

template<bool from_coarse, typename Float , int dim, QudaDirection dir, int fineSpin, int fineColor, int coarseSpin, int coarseColor, typename Arg >

__global__ void quda::ComputeVUVGPU

(

Arg

arg

)

Definition at line 624 of file coarse_op.cuh.

References arg(), blockDim, and parity.

Here is the call graph for this function:

computeWupperalStep()

template<typename Float , int Ns, int Nc, typename Arg >

__device__ __host__ void quda::computeWupperalStep ( Arg &  arg, int  x_cb, int  parity  )

inline

Definition at line 103 of file color_spinor_wuppertal.cu.

References arg(), in, out, and parity.

Here is the call graph for this function:

computeYhat()

template<typename Float , int n, typename Arg >

__device__ __host__ void quda::computeYhat	(	Arg &	arg,
		int	d,
		int	x_cb,
		int	parity,
		int	i
	)

Definition at line 1349 of file coarse_op.cuh.

References arg(), conj(), coord, d, getCoords(), fused_exterior_ndeg_tm_dslash_cuda_gen::i, linkIndexM1(), n, and parity.

Here is the call graph for this function:

computeYreverse()

template<typename Float , int nSpin, int nColor, typename Arg >

__device__ __host__ void quda::computeYreverse	(	Arg &	arg,
		int	parity,
		int	x_cb
	)

Compute the forward links from backwards links by flipping the sign of the spin projector

Definition at line 639 of file coarse_op.cuh.

References arg(), d, nColor, parity, and deg_tm_dslash_cuda_gen::sign().

Here is the call graph for this function:

ComputeYReverseCPU()

template<typename Float , int nSpin, int nColor, typename Arg >

void quda::ComputeYReverseCPU

(

Arg &

arg

)

Definition at line 661 of file coarse_op.cuh.

References arg(), and parity.

Here is the call graph for this function:

ComputeYReverseGPU()

template<typename Float , int nSpin, int nColor, typename Arg >

__global__ void quda::ComputeYReverseGPU

(

Arg

arg

)

Definition at line 670 of file coarse_op.cuh.

References arg(), blockDim, and parity.

Here is the call graph for this function:

conj() [1/3]

template<typename ValueType >

__host__ __device__ ValueType quda::conj ( ValueType  x )

inline

Definition at line 115 of file complex_quda.h.

References x.

Referenced by applyLaplace(), quda::blas::cDotProduct(), computeCoarseClover(), computeCoarseLocal(), computeNeighborSum(), computeOvrImpSTOUTStep(), computeStapleRectangle(), computeYhat(), conj(), ErrorSU3(), quda::GMResDR::FlexArnoldiProcedure(), quda::gauge::Reconstruct< 13, Float >::getPhase(), quda::gauge::Reconstruct< 9, Float >::getPhase(), quda::blas::hDotProduct(), quda::blas::hDotProduct_Anorm(), quda::Deflation::increment(), isUnitary(), makeAntiHerm(), multiplyVUV(), quda::BiCGstab::operator()(), quda::SimpleBiCGstab::operator()(), quda::MPBiCGstab::operator()(), outerProd(), polarSu3(), solve(), quda::CG::solve(), quda::gauge::Reconstruct< 12, Float >::Unpack(), quda::gauge::Reconstruct< 13, Float >::Unpack(), and quda::gauge::Reconstruct< 8, Float >::Unpack().

Here is the caller graph for this function:

conj() [2/3]

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::conj ( const complex< ValueType > &  z )

inline

Returns the complex conjugate of z.

Definition at line 858 of file complex_quda.h.

References z.

conj() [3/3]

template<class T , int N>

__device__ __host__ Matrix<T,N> quda::conj ( const Matrix< T, N > &  other )

inline

Definition at line 486 of file quda_matrix.h.

References conj(), and fused_exterior_ndeg_tm_dslash_cuda_gen::i.

Here is the call graph for this function:

constant()

template<class T >

void quda::constant	(	T &	t,
		int	k,
		int	s,
		int	c
	)

Set all space-time real elements at spin s and color c of the field equal to k

Definition at line 37 of file color_spinor_util.cu.

References c, parity, s, and t.

Referenced by genericSource().

Here is the caller graph for this function:

contractCuda() [1/2]

void quda::contractCuda	(	const cudaColorSpinorField &	x,
		const cudaColorSpinorField &	y,
		void *	result,
		const QudaContractType	contract_type,
		const QudaParity	parity,
		TimeProfile &	profile
	)

Contracts the x and y spinors (x is daggered) and stores the result in the array result. One must specify the contract type (time-sliced or volumed contract, and whether we should include a gamma5 in the middle), as well as the time-slice (see overloaded version of the same function) in case we don't want a volume contraction. The function works only with parity spinors, and the parity must be specified.

Definition at line 202 of file contract.cu.

References checkCudaError, contract(), errorQuda, Nstream, parity, QUDA_CONTRACT_TSLICE, QUDA_CONTRACT_TSLICE_MINUS, QUDA_CONTRACT_TSLICE_PLUS, QUDA_DOUBLE_PRECISION, QUDA_HALF_PRECISION, QUDA_PROFILE_COMPUTE, QUDA_PROFILE_EPILOGUE, QUDA_PROFILE_INIT, QUDA_PROFILE_TOTAL, QUDA_SINGLE_PRECISION, qudaStreamSynchronize(), streams, x, and y.

Referenced by contract().

Here is the call graph for this function:

Here is the caller graph for this function:

contractCuda() [2/2]

void quda::contractCuda	(	const cudaColorSpinorField &	x,
		const cudaColorSpinorField &	y,
		void *	result,
		const QudaContractType	contract_type,
		const int	nTSlice,
		const QudaParity	parity,
		TimeProfile &	profile
	)

Contracts the x and y spinors (x is daggered) and stores the result in the array result. One must specify the contract type (time-sliced or volumed contract, and whether we should include a gamma5 in the middle), as well as the time-slice in case we don't want a volume contraction. The function works only with parity spinors, and the parity must be specified.

Definition at line 248 of file contract.cu.

References checkCudaError, contract(), errorQuda, Nstream, parity, QUDA_CONTRACT_TSLICE, QUDA_CONTRACT_TSLICE_MINUS, QUDA_CONTRACT_TSLICE_PLUS, QUDA_DOUBLE_PRECISION, QUDA_HALF_PRECISION, QUDA_PROFILE_COMPUTE, QUDA_PROFILE_EPILOGUE, QUDA_PROFILE_INIT, QUDA_PROFILE_TOTAL, QUDA_SINGLE_PRECISION, qudaStreamSynchronize(), streams, x, and y.

Here is the call graph for this function:

copy() [1/9]

template<typename T1 , typename T2 >

__host__ __device__ void quda::copy ( T1 &  a, const T2 &  b  )

inline

Definition at line 114 of file register_traits.h.

References a, and b.

Referenced by computeCloverForceQuda(), genericCopyColorSpinor(), quda::colorspinor::FloatNOrder< Float, Ns, Nc, N, huge_alloc >::load(), quda::gauge::FloatNOrder< Float, length, N, reconLenParam, stag_phase, huge_alloc >::load(), quda::colorspinor::FloatNOrder< Float, Ns, Nc, N, huge_alloc >::loadGhost(), quda::gauge::FloatNOrder< Float, length, N, reconLenParam, stag_phase, huge_alloc >::loadGhost(), quda::gauge::FloatNOrder< Float, length, N, reconLenParam, stag_phase, huge_alloc >::loadGhostEx(), new_load_half(), new_save_half(), old_load_half(), old_save_half(), quda::PreconCG::operator()(), qudaMemcpy_(), quda::colorspinor::FloatNOrder< Float, Ns, Nc, N, huge_alloc >::save(), quda::gauge::FloatNOrder< Float, length, N, reconLenParam, stag_phase, huge_alloc >::save(), quda::colorspinor::FloatNOrder< Float, Ns, Nc, N, huge_alloc >::saveGhost(), quda::gauge::FloatNOrder< Float, length, N, reconLenParam, stag_phase, huge_alloc >::saveGhost(), and quda::gauge::FloatNOrder< Float, length, N, reconLenParam, stag_phase, huge_alloc >::saveGhostEx().

Here is the caller graph for this function:

copy() [2/9]

template<>

__host__ __device__ void quda::copy ( double &  a, const int2 &  b  )

inline

Definition at line 116 of file register_traits.h.

References a, b, and errorQuda.

copy() [3/9]

template<>

__host__ __device__ void quda::copy ( double2 &  a, const int4 &  b  )

inline

Definition at line 124 of file register_traits.h.

References a, b, and errorQuda.

copy() [4/9]

template<>

__host__ __device__ void quda::copy ( float &  a, const short &  b  )

inline

Definition at line 155 of file register_traits.h.

References a, b, and s2f().

Here is the call graph for this function:

copy() [5/9]

template<>

__host__ __device__ void quda::copy ( short &  a, const float &  b  )

inline

Definition at line 156 of file register_traits.h.

References a, b, f2i(), and MAX_SHORT.

Here is the call graph for this function:

copy() [6/9]

template<>

__host__ __device__ void quda::copy ( float2 &  a, const short2 &  b  )

inline

Definition at line 158 of file register_traits.h.

References a, b, and s2f().

Here is the call graph for this function:

copy() [7/9]

template<>

__host__ __device__ void quda::copy ( short2 &  a, const float2 &  b  )

inline

Definition at line 162 of file register_traits.h.

References a, b, f2i(), and MAX_SHORT.

Here is the call graph for this function:

copy() [8/9]

template<>

__host__ __device__ void quda::copy ( float4 &  a, const short4 &  b  )

inline

Definition at line 166 of file register_traits.h.

References a, b, and s2f().

Here is the call graph for this function:

copy() [9/9]

template<>

__host__ __device__ void quda::copy ( short4 &  a, const float4 &  b  )

inline

Definition at line 170 of file register_traits.h.

References a, b, f2i(), and MAX_SHORT.

Here is the call graph for this function:

copyArrayToLink() [1/2]

void quda::copyArrayToLink ( Matrix< float2, 3 > *  link, float *  array  )

inline

Definition at line 951 of file quda_matrix.h.

References array, fused_exterior_ndeg_tm_dslash_cuda_gen::i, x, and y.

copyArrayToLink() [2/2]

template<class Cmplx , class Real >

void quda::copyArrayToLink ( Matrix< Cmplx, 3 > *  link, Real *  array  )

inline

Definition at line 964 of file quda_matrix.h.

References array, fused_exterior_ndeg_tm_dslash_cuda_gen::i, x, and y.

copyColorSpinor()

template<typename FloatOut , typename FloatIn , int Ns, int Nc, typename Arg , typename Basis >

void quda::copyColorSpinor	(	Arg &	arg,
		const Basis &	basis
	)

CPU function to reorder spinor fields.

Definition at line 123 of file copy_color_spinor.cuh.

References arg(), in, out, parity, and x.

Here is the call graph for this function:

copyColorSpinorKernel()

template<typename FloatOut , typename FloatIn , int Ns, int Nc, typename Arg , typename Basis >

__global__ void quda::copyColorSpinorKernel	(	Arg	arg,
		Basis	basis
	)

CUDA kernel to reorder spinor fields. Adopts a similar form as the CPU version, using the same inlined functions.

Definition at line 139 of file copy_color_spinor.cuh.

References arg(), blockDim, in, out, parity, and x.

Here is the call graph for this function:

copyColumn()

template<class T , int N>

__device__ __host__ void quda::copyColumn ( const Matrix< T, N > &  m, int  c, Array< T, N > *  a  )

inline

Definition at line 683 of file quda_matrix.h.

References c, and fused_exterior_ndeg_tm_dslash_cuda_gen::i.

Referenced by getRealBidiagMatrix().

Here is the caller graph for this function:

copyExtendedColorSpinor() [1/2]

template<int Ns, typename dstFloat , typename srcFloat >

void quda::copyExtendedColorSpinor	(	ColorSpinorField &	dst,
		const ColorSpinorField &	src,
		const int	parity,
		const QudaFieldLocation	location,
		dstFloat *	Dst,
		srcFloat *	Src,
		float *	dstNorm,
		float *	srcNorm
	)

Definition at line 368 of file extended_color_spinor_utilities.cu.

References quda::ColorSpinorField::Bytes(), errorQuda, quda::ColorSpinorField::FieldOrder(), quda::ColorSpinorField::Ncolor(), quda::ColorSpinorField::Ndim(), quda::ColorSpinorField::Norm(), quda::ColorSpinorField::NormBytes(), parity, QUDA_EVEN_ODD_SITE_ORDER, QUDA_FULL_SITE_SUBSET, QUDA_LEXICOGRAPHIC_SITE_ORDER, QUDA_ODD_EVEN_SITE_ORDER, QUDA_QDPJIT_FIELD_ORDER, quda::ColorSpinorField::SiteOrder(), quda::ColorSpinorField::SiteSubset(), src, and quda::ColorSpinorField::V().

Here is the call graph for this function:

CopyExtendedColorSpinor()

template<typename dstFloat , typename srcFloat >

void quda::CopyExtendedColorSpinor	(	ColorSpinorField &	dst,
		const ColorSpinorField &	src,
		const int	parity,
		const QudaFieldLocation	location,
		dstFloat *	Dst,
		srcFloat *	Src,
		float *	dstNorm = 0,
		float *	srcNorm = 0
	)

Definition at line 436 of file extended_color_spinor_utilities.cu.

References errorQuda, quda::ColorSpinorField::Nspin(), parity, and src.

Referenced by copyExtendedColorSpinor().

Here is the call graph for this function:

Here is the caller graph for this function:

copyExtendedColorSpinor() [2/2]

void quda::copyExtendedColorSpinor	(	ColorSpinorField &	dst,
		const ColorSpinorField &	src,
		QudaFieldLocation	location,
		const int	parity,
		void *	Dst,
		void *	Src,
		void *	dstNorm,
		void *	srcNorm
	)

Definition at line 462 of file extended_color_spinor_utilities.cu.

References CopyExtendedColorSpinor(), errorQuda, parity, quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, QUDA_HALF_PRECISION, QUDA_SINGLE_PRECISION, and src.

Referenced by quda::XSD::operator()().

Here is the call graph for this function:

Here is the caller graph for this function:

copyExtendedGauge()

void quda::copyExtendedGauge	(	GaugeField &	out,
		const GaugeField &	in,
		QudaFieldLocation	location,
		void *	Out = 0,
		void *	In = 0
	)

This function is used for copying the gauge field into an extended gauge field. Defined in copy_extended_gauge.cu.

Parameters

out	The extended output field to which we are copying
in	The input field from which we are copying
location	The location of where we are doing the copying (CPU or CUDA)
Out	The output buffer (optional)
In	The input buffer (optional)

Definition at line 321 of file copy_gauge_extended.cu.

References copyGaugeEx(), d, errorQuda, in, out, QUDA_DOUBLE_PRECISION, and QUDA_SINGLE_PRECISION.

Referenced by computeGaugeFixingOVRQuda(), computeHISQForceQuda(), quda::cudaGaugeField::copy(), quda::cpuGaugeField::copy(), createExtendedGauge(), hisq_force_init(), main(), performWuppertalnStep(), quda::cudaGaugeField::saveCPUField(), and saveGaugeQuda().

Here is the call graph for this function:

Here is the caller graph for this function:

copyGauge() [1/5]

template<typename FloatOut , typename FloatIn , int length, typename InOrder >

void quda::copyGauge	(	const InOrder &	inOrder,
		const GaugeField &	out,
		const GaugeField &	in,
		QudaFieldLocation	location,
		FloatOut *	Out,
		FloatOut **	outGhost,
		int	type
	)

Definition at line 7 of file copy_gauge_inc.cu.

References errorQuda, fused_exterior_ndeg_tm_dslash_cuda_gen::i, in, quda::ColorSpinorField::isNative(), quda::ColorSpinorField::Ndim(), out, QUDA_ASQTAD_FAT_LINKS, QUDA_BQCD_GAUGE_ORDER, QUDA_CPS_WILSON_GAUGE_ORDER, QUDA_MAX_DIM, QUDA_MILC_GAUGE_ORDER, QUDA_MILC_SITE_GAUGE_ORDER, QUDA_QDP_GAUGE_ORDER, QUDA_QDPJIT_GAUGE_ORDER, QUDA_RECONSTRUCT_12, QUDA_RECONSTRUCT_13, QUDA_RECONSTRUCT_8, QUDA_RECONSTRUCT_9, QUDA_RECONSTRUCT_NO, QUDA_TIFR_GAUGE_ORDER, QUDA_TIFR_PADDED_GAUGE_ORDER, quda::LatticeField::SurfaceCB(), and quda::ColorSpinorField::Volume().

Here is the call graph for this function:

copyGauge() [2/5]

template<typename FloatOut , typename FloatIn , int length, typename OutOrder , typename InOrder >

void quda::copyGauge

(

CopyGaugeArg< OutOrder, InOrder >

arg

)

Generic CPU gauge reordering and packing

Definition at line 32 of file copy_gauge_helper.cuh.

References arg(), d, fused_exterior_ndeg_tm_dslash_cuda_gen::i, in, length, quda::gauge::Ncolor(), out, parity, and x.

Referenced by copyGenericGaugeDoubleOut(), copyGenericGaugeHalfOut(), and copyGenericGaugeSingleOut().

Here is the call graph for this function:

Here is the caller graph for this function:

copyGauge() [3/5]

template<typename FloatOut , typename FloatIn , int length>

void quda::copyGauge	(	GaugeField &	out,
		const GaugeField &	in,
		QudaFieldLocation	location,
		FloatOut *	Out,
		FloatIn *	In,
		FloatOut **	outGhost,
		FloatIn **	inGhost,
		int	type
	)

Definition at line 140 of file copy_gauge_inc.cu.

References errorQuda, in, quda::ColorSpinorField::isNative(), out, QUDA_ASQTAD_FAT_LINKS, QUDA_BQCD_GAUGE_ORDER, QUDA_CPS_WILSON_GAUGE_ORDER, QUDA_MILC_GAUGE_ORDER, QUDA_MILC_SITE_GAUGE_ORDER, QUDA_QDP_GAUGE_ORDER, QUDA_QDPJIT_GAUGE_ORDER, QUDA_RECONSTRUCT_12, QUDA_RECONSTRUCT_13, QUDA_RECONSTRUCT_8, QUDA_RECONSTRUCT_9, QUDA_RECONSTRUCT_NO, QUDA_TIFR_GAUGE_ORDER, and QUDA_TIFR_PADDED_GAUGE_ORDER.

Here is the call graph for this function:

copyGauge() [4/5]

template<typename FloatOut , typename FloatIn , int length, typename OutOrder , typename InOrder >

void quda::copyGauge	(	OutOrder &&	outOrder,
		const InOrder &	inOrder,
		int	volume,
		const int *	faceVolumeCB,
		int	nDim,
		int	geometry,
		const GaugeField &	out,
		const GaugeField &	in,
		QudaFieldLocation	location,
		int	type
	)

Definition at line 253 of file copy_gauge_helper.cuh.

References quda::CopyGauge< FloatOut, FloatIn, length, OutOrder, InOrder, isGhost >::apply(), arg(), errorQuda, in, out, QUDA_COARSE_GEOMETRY, QUDA_CPU_FIELD_LOCATION, QUDA_CUDA_FIELD_LOCATION, and QUDA_VECTOR_GEOMETRY.

Here is the call graph for this function:

copyGauge() [5/5]

template<typename FloatOut , typename FloatIn >

void quda::copyGauge	(	GaugeField &	out,
		const GaugeField &	in,
		QudaFieldLocation	location,
		FloatOut *	Out,
		FloatIn *	In,
		FloatOut **	outGhost,
		FloatIn **	inGhost,
		int	type
	)

Definition at line 271 of file copy_gauge_inc.cu.

References arg(), checkMomOrder(), d, errorQuda, in, quda::ColorSpinorField::Ncolor(), quda::ColorSpinorField::Ndim(), out, QUDA_ASQTAD_MOM_LINKS, QUDA_FLOAT2_GAUGE_ORDER, QUDA_MAX_DIM, QUDA_MILC_GAUGE_ORDER, QUDA_MILC_SITE_GAUGE_ORDER, QUDA_TIFR_GAUGE_ORDER, QUDA_TIFR_PADDED_GAUGE_ORDER, QUDA_VECTOR_GEOMETRY, quda::LatticeField::SurfaceCB(), and quda::ColorSpinorField::Volume().

Here is the call graph for this function:

copyGaugeEx() [1/6]

template<typename FloatOut , typename FloatIn , int length, typename OutOrder , typename InOrder , bool regularToextended>

__device__ __host__ void quda::copyGaugeEx	(	CopyGaugeExArg< OutOrder, InOrder > &	arg,
		int	X,
		int	parity
	)

Copy a regular/extended gauge field into an extended/regular gauge field

Definition at line 48 of file copy_gauge_extended.cu.

References arg(), d, fused_exterior_ndeg_tm_dslash_cuda_gen::i, in, length, out, parity, R, X, x, x0h, za, and zb.

Referenced by copyExtendedGauge().

Here is the call graph for this function:

Here is the caller graph for this function:

copyGaugeEx() [2/6]

template<typename FloatOut , typename FloatIn , int length, typename OutOrder , typename InOrder , bool regularToextended>

void quda::copyGaugeEx

(

CopyGaugeExArg< OutOrder, InOrder >

arg

)

Definition at line 92 of file copy_gauge_extended.cu.

References arg(), parity, and X.

Here is the call graph for this function:

copyGaugeEx() [3/6]

template<typename FloatOut , typename FloatIn , int length, typename OutOrder , typename InOrder >

void quda::copyGaugeEx	(	OutOrder	outOrder,
		const InOrder	inOrder,
		const int *	E,
		const int *	X,
		const int *	faceVolumeCB,
		const GaugeField &	meta,
		QudaFieldLocation	location
	)

Definition at line 157 of file copy_gauge_extended.cu.

References arg(), checkCudaError, E, quda::GaugeField::Geometry(), quda::LatticeField::Ndim(), QUDA_CUDA_FIELD_LOCATION, and X.

Here is the call graph for this function:

copyGaugeEx() [4/6]

template<typename FloatOut , typename FloatIn , int length, typename InOrder >

void quda::copyGaugeEx	(	const InOrder &	inOrder,
		const int *	X,
		GaugeField &	out,
		QudaFieldLocation	location,
		FloatOut *	Out
	)

Definition at line 168 of file copy_gauge_extended.cu.

References errorQuda, fused_exterior_ndeg_tm_dslash_cuda_gen::i, out, QUDA_ASQTAD_FAT_LINKS, QUDA_MAX_DIM, QUDA_MILC_GAUGE_ORDER, QUDA_QDP_GAUGE_ORDER, QUDA_RECONSTRUCT_12, QUDA_RECONSTRUCT_13, QUDA_RECONSTRUCT_8, QUDA_RECONSTRUCT_9, QUDA_RECONSTRUCT_NO, QUDA_TIFR_GAUGE_ORDER, and X.

copyGaugeEx() [5/6]

template<typename FloatOut , typename FloatIn , int length>

void quda::copyGaugeEx	(	GaugeField &	out,
		const GaugeField &	in,
		QudaFieldLocation	location,
		FloatOut *	Out,
		FloatIn *	In
	)

Definition at line 239 of file copy_gauge_extended.cu.

References errorQuda, in, out, QUDA_ASQTAD_FAT_LINKS, QUDA_MILC_GAUGE_ORDER, QUDA_QDP_GAUGE_ORDER, QUDA_RECONSTRUCT_12, QUDA_RECONSTRUCT_13, QUDA_RECONSTRUCT_8, QUDA_RECONSTRUCT_9, QUDA_RECONSTRUCT_NO, and QUDA_TIFR_GAUGE_ORDER.

copyGaugeEx() [6/6]

template<typename FloatOut , typename FloatIn >

void quda::copyGaugeEx	(	GaugeField &	out,
		const GaugeField &	in,
		QudaFieldLocation	location,
		FloatOut *	Out,
		FloatIn *	In
	)

Definition at line 302 of file copy_gauge_extended.cu.

References errorQuda, in, out, and QUDA_ASQTAD_MOM_LINKS.

copyGaugeExKernel()

template<typename FloatOut , typename FloatIn , int length, typename OutOrder , typename InOrder , bool regularToextended>

__global__ void quda::copyGaugeExKernel

(

CopyGaugeExArg< OutOrder, InOrder >

arg

)

Definition at line 101 of file copy_gauge_extended.cu.

References arg(), blockDim, parity, and X.

Here is the call graph for this function:

copyGaugeKernel()

template<typename FloatOut , typename FloatIn , int length, typename OutOrder , typename InOrder >

__global__ void quda::copyGaugeKernel

(

CopyGaugeArg< OutOrder, InOrder >

arg

)

Generic CUDA gauge reordering and packing. Adopts a similar form as the CPU version, using the same inlined functions.

Definition at line 96 of file copy_gauge_helper.cuh.

References arg(), blockDim, d, fused_exterior_ndeg_tm_dslash_cuda_gen::i, in, length, quda::gauge::Ncolor(), out, parity, and x.

Here is the call graph for this function:

copyGaugeMG() [1/3]

template<typename FloatOut , typename FloatIn , int length, typename InOrder >

void quda::copyGaugeMG	(	const InOrder &	inOrder,
		GaugeField &	out,
		const GaugeField &	in,
		QudaFieldLocation	location,
		FloatOut *	Out,
		FloatOut **	outGhost,
		int	type
	)

Definition at line 10 of file copy_gauge_mg.cu.

References errorQuda, fused_exterior_ndeg_tm_dslash_cuda_gen::i, in, quda::ColorSpinorField::isNative(), quda::ColorSpinorField::Ndim(), out, QUDA_MAX_DIM, QUDA_MILC_GAUGE_ORDER, QUDA_QDP_GAUGE_ORDER, QUDA_RECONSTRUCT_NO, quda::LatticeField::SurfaceCB(), and quda::ColorSpinorField::Volume().

Referenced by copyGenericGaugeMG().

Here is the call graph for this function:

Here is the caller graph for this function:

copyGaugeMG() [2/3]

template<typename FloatOut , typename FloatIn , int length>

void quda::copyGaugeMG	(	GaugeField &	out,
		const GaugeField &	in,
		QudaFieldLocation	location,
		FloatOut *	Out,
		FloatIn *	In,
		FloatOut **	outGhost,
		FloatIn **	inGhost,
		int	type
	)

Definition at line 67 of file copy_gauge_mg.cu.

References errorQuda, in, quda::ColorSpinorField::isNative(), out, QUDA_MILC_GAUGE_ORDER, QUDA_QDP_GAUGE_ORDER, and QUDA_RECONSTRUCT_NO.

Here is the call graph for this function:

copyGaugeMG() [3/3]

template<typename FloatOut , typename FloatIn >

void quda::copyGaugeMG	(	GaugeField &	out,
		const GaugeField &	in,
		QudaFieldLocation	location,
		FloatOut *	Out,
		FloatIn *	In,
		FloatOut **	outGhost,
		FloatIn **	inGhost,
		int	type
	)

Definition at line 114 of file copy_gauge_mg.cu.

References errorQuda, in, quda::ColorSpinorField::Ncolor(), and out.

Here is the call graph for this function:

copyGenericClover()

void quda::copyGenericClover	(	CloverField &	out,
		const CloverField &	in,
		bool	inverse,
		QudaFieldLocation	location,
		void *	Out = 0,
		void *	In = 0,
		void *	outNorm = 0,
		void *	inNorm = 0
	)

This generic function is used for copying the clover field where in the input and output can be in any order and location.

Parameters

out	The output field to which we are copying
in	The input field from which we are copying
inverse	Whether we are copying the inverse term or not
location	The location of where we are doing the copying (CPU or CUDA)
Out	The output buffer (optional)
In	The input buffer (optional)
outNorm	The output norm buffer (optional)
inNorm	The input norm buffer (optional)

Definition at line 175 of file copy_clover.cu.

References errorQuda, in, out, QUDA_DOUBLE_PRECISION, QUDA_HALF_PRECISION, and QUDA_SINGLE_PRECISION.

Referenced by quda::cudaCloverField::copy(), and quda::cudaCloverField::saveCPUField().

Here is the caller graph for this function:

copyGenericColorSpinor() [1/3]

template<int Ns, int Nc, typename dstFloat , typename srcFloat >

void quda::copyGenericColorSpinor	(	ColorSpinorField &	dst,
		const ColorSpinorField &	src,
		QudaFieldLocation	location,
		dstFloat *	Dst,
		srcFloat *	Src
	)

Definition at line 130 of file copy_color_spinor_mg.cuh.

References quda::ColorSpinorField::Bytes(), errorQuda, quda::ColorSpinorField::FieldOrder(), quda::ColorSpinorField::Ndim(), QUDA_EVEN_ODD_SITE_ORDER, QUDA_FULL_SITE_SUBSET, QUDA_LEXICOGRAPHIC_SITE_ORDER, QUDA_ODD_EVEN_SITE_ORDER, QUDA_QDPJIT_FIELD_ORDER, quda::ColorSpinorField::SiteOrder(), quda::ColorSpinorField::SiteSubset(), src, quda::ColorSpinorField::V(), and quda::ColorSpinorField::Volume().

Here is the call graph for this function:

CopyGenericColorSpinor() [1/2]

template<int Nc, typename dstFloat , typename srcFloat >

void quda::CopyGenericColorSpinor	(	ColorSpinorField &	dst,
		const ColorSpinorField &	src,
		QudaFieldLocation	location,
		dstFloat *	Dst,
		srcFloat *	Src
	)

Definition at line 184 of file copy_color_spinor_mg.cuh.

References errorQuda, quda::ColorSpinorField::Nspin(), and src.

Here is the call graph for this function:

copyGenericColorSpinor() [2/3]

template<int Ns, int Nc, typename dstFloat , typename srcFloat >

void quda::copyGenericColorSpinor	(	ColorSpinorField &	dst,
		const ColorSpinorField &	src,
		QudaFieldLocation	location,
		dstFloat *	Dst,
		srcFloat *	Src,
		float *	dstNorm,
		float *	srcNorm
	)

Definition at line 376 of file copy_color_spinor.cuh.

References errorQuda, quda::ColorSpinorField::FieldOrder(), quda::ColorSpinorField::Ndim(), QUDA_EVEN_ODD_SITE_ORDER, QUDA_FULL_SITE_SUBSET, QUDA_LEXICOGRAPHIC_SITE_ORDER, QUDA_ODD_EVEN_SITE_ORDER, QUDA_QDPJIT_FIELD_ORDER, quda::ColorSpinorField::SiteOrder(), quda::ColorSpinorField::SiteSubset(), src, and quda::ColorSpinorField::Volume().

Here is the call graph for this function:

CopyGenericColorSpinor() [2/2]

template<int Nc, typename dstFloat , typename srcFloat >

void quda::CopyGenericColorSpinor	(	ColorSpinorField &	dst,
		const ColorSpinorField &	src,
		QudaFieldLocation	location,
		dstFloat *	Dst,
		srcFloat *	Src,
		float *	dstNorm = 0,
		float *	srcNorm = 0
	)

Definition at line 411 of file copy_color_spinor.cuh.

References errorQuda, quda::ColorSpinorField::Nspin(), and src.

Here is the call graph for this function:

copyGenericColorSpinor() [3/3]

void quda::copyGenericColorSpinor	(	ColorSpinorField &	dst,
		const ColorSpinorField &	src,
		QudaFieldLocation	location,
		void *	Dst = 0,
		void *	Src = 0,
		void *	dstNorm = 0,
		void *	srcNorm = 0
	)

Definition at line 23 of file copy_color_spinor.cu.

References copyGenericColorSpinorDD(), copyGenericColorSpinorDH(), copyGenericColorSpinorDS(), copyGenericColorSpinorHD(), copyGenericColorSpinorHH(), copyGenericColorSpinorHS(), copyGenericColorSpinorMGDD(), copyGenericColorSpinorMGDS(), copyGenericColorSpinorMGSD(), copyGenericColorSpinorMGSS(), copyGenericColorSpinorSD(), copyGenericColorSpinorSH(), copyGenericColorSpinorSS(), errorQuda, quda::ColorSpinorField::Ncolor(), quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, QUDA_HALF_PRECISION, QUDA_SINGLE_PRECISION, quda::ColorSpinorField::SiteSubset(), and src.

Referenced by quda::cpuColorSpinorField::copy(), quda::cudaColorSpinorField::copySpinorField(), quda::cudaColorSpinorField::loadSpinorField(), and quda::cudaColorSpinorField::saveSpinorField().

Here is the call graph for this function:

Here is the caller graph for this function:

copyGenericColorSpinorDD()

void quda::copyGenericColorSpinorDD	(	ColorSpinorField &	dst,
		const ColorSpinorField &	src,
		QudaFieldLocation	location,
		void *	Dst,
		void *	Src,
		void *	a = 0,
		void *	b = 0
	)

Definition at line 5 of file copy_color_spinor_dd.cu.

References src.

Referenced by copyGenericColorSpinor().

Here is the caller graph for this function:

copyGenericColorSpinorDH()

void quda::copyGenericColorSpinorDH	(	ColorSpinorField &	dst,
		const ColorSpinorField &	src,
		QudaFieldLocation	location,
		void *	Dst,
		void *	Src,
		void *	a = 0,
		void *	b = 0
	)

Definition at line 5 of file copy_color_spinor_dh.cu.

References src.

Referenced by copyGenericColorSpinor().

Here is the caller graph for this function:

copyGenericColorSpinorDS()

void quda::copyGenericColorSpinorDS	(	ColorSpinorField &	dst,
		const ColorSpinorField &	src,
		QudaFieldLocation	location,
		void *	Dst,
		void *	Src,
		void *	a = 0,
		void *	b = 0
	)

Definition at line 5 of file copy_color_spinor_ds.cu.

References src.

Referenced by copyGenericColorSpinor().

Here is the caller graph for this function:

copyGenericColorSpinorHD()

void quda::copyGenericColorSpinorHD	(	ColorSpinorField &	dst,
		const ColorSpinorField &	src,
		QudaFieldLocation	location,
		void *	Dst,
		void *	Src,
		void *	a = 0,
		void *	b = 0
	)

Definition at line 5 of file copy_color_spinor_hd.cu.

References src.

Referenced by copyGenericColorSpinor().

Here is the caller graph for this function:

copyGenericColorSpinorHH()

void quda::copyGenericColorSpinorHH	(	ColorSpinorField &	dst,
		const ColorSpinorField &	src,
		QudaFieldLocation	location,
		void *	Dst,
		void *	Src,
		void *	a = 0,
		void *	b = 0
	)

Definition at line 5 of file copy_color_spinor_hh.cu.

References src.

Referenced by copyGenericColorSpinor().

Here is the caller graph for this function:

copyGenericColorSpinorHS()

void quda::copyGenericColorSpinorHS	(	ColorSpinorField &	dst,
		const ColorSpinorField &	src,
		QudaFieldLocation	location,
		void *	Dst,
		void *	Src,
		void *	a = 0,
		void *	b = 0
	)

Definition at line 5 of file copy_color_spinor_hs.cu.

References src.

Referenced by copyGenericColorSpinor().

Here is the caller graph for this function:

copyGenericColorSpinorMGDD()

void quda::copyGenericColorSpinorMGDD	(	ColorSpinorField &	dst,
		const ColorSpinorField &	src,
		QudaFieldLocation	location,
		void *	Dst,
		void *	Src,
		void *	a = 0,
		void *	b = 0
	)

Definition at line 5 of file copy_color_spinor_mg_dd.cu.

References errorQuda, and INSTANTIATE_COLOR.

Referenced by copyGenericColorSpinor().

Here is the caller graph for this function:

copyGenericColorSpinorMGDS()

void quda::copyGenericColorSpinorMGDS	(	ColorSpinorField &	dst,
		const ColorSpinorField &	src,
		QudaFieldLocation	location,
		void *	Dst,
		void *	Src,
		void *	a = 0,
		void *	b = 0
	)

Definition at line 5 of file copy_color_spinor_mg_ds.cu.

References errorQuda, and INSTANTIATE_COLOR.

Referenced by copyGenericColorSpinor().

Here is the caller graph for this function:

copyGenericColorSpinorMGSD()

void quda::copyGenericColorSpinorMGSD	(	ColorSpinorField &	dst,
		const ColorSpinorField &	src,
		QudaFieldLocation	location,
		void *	Dst,
		void *	Src,
		void *	a = 0,
		void *	b = 0
	)

Definition at line 5 of file copy_color_spinor_mg_sd.cu.

References errorQuda, and INSTANTIATE_COLOR.

Referenced by copyGenericColorSpinor().

Here is the caller graph for this function:

copyGenericColorSpinorMGSS()

void quda::copyGenericColorSpinorMGSS	(	ColorSpinorField &	dst,
		const ColorSpinorField &	src,
		QudaFieldLocation	location,
		void *	Dst,
		void *	Src,
		void *	a = 0,
		void *	b = 0
	)

Definition at line 5 of file copy_color_spinor_mg_ss.cu.

References errorQuda, and INSTANTIATE_COLOR.

Referenced by copyGenericColorSpinor().

Here is the caller graph for this function:

copyGenericColorSpinorSD()

void quda::copyGenericColorSpinorSD	(	ColorSpinorField &	dst,
		const ColorSpinorField &	src,
		QudaFieldLocation	location,
		void *	Dst,
		void *	Src,
		void *	a = 0,
		void *	b = 0
	)

Definition at line 5 of file copy_color_spinor_sd.cu.

References src.

Referenced by copyGenericColorSpinor().

Here is the caller graph for this function:

copyGenericColorSpinorSH()

void quda::copyGenericColorSpinorSH	(	ColorSpinorField &	dst,
		const ColorSpinorField &	src,
		QudaFieldLocation	location,
		void *	Dst,
		void *	Src,
		void *	a = 0,
		void *	b = 0
	)

Definition at line 5 of file copy_color_spinor_sh.cu.

References src.

Referenced by copyGenericColorSpinor().

Here is the caller graph for this function:

copyGenericColorSpinorSS()

void quda::copyGenericColorSpinorSS	(	ColorSpinorField &	dst,
		const ColorSpinorField &	src,
		QudaFieldLocation	location,
		void *	Dst,
		void *	Src,
		void *	a = 0,
		void *	b = 0
	)

Definition at line 5 of file copy_color_spinor_ss.cu.

References src.

Referenced by copyGenericColorSpinor().

Here is the caller graph for this function:

copyGenericGauge()

void quda::copyGenericGauge	(	GaugeField &	out,
		const GaugeField &	in,
		QudaFieldLocation	location,
		void *	Out = 0,
		void *	In = 0,
		void **	ghostOut = 0,
		void **	ghostIn = 0,
		int	type = 0
	)

This function is used for extracting the gauge ghost zone from a gauge field array. Defined in copy_gauge.cu.

Parameters

out	The output field to which we are copying
in	The input field from which we are copying
location	The location of where we are doing the copying (CPU or CUDA)
Out	The output buffer (optional)
In	The input buffer (optional)
ghostOut	The output ghost buffer (optional)
ghostIn	The input ghost buffer (optional)
type	The type of copy we doing (0 body and ghost else ghost only)

Definition at line 38 of file copy_gauge.cu.

References copyGenericGaugeDoubleOut(), copyGenericGaugeHalfOut(), copyGenericGaugeMG(), copyGenericGaugeSingleOut(), errorQuda, quda::LatticeField::GhostExchange(), in, quda::ColorSpinorField::Ncolor(), out, quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, QUDA_GHOST_EXCHANGE_PAD, QUDA_HALF_PRECISION, and QUDA_SINGLE_PRECISION.

Referenced by quda::cudaGaugeField::copy(), quda::cpuGaugeField::copy(), quda::cudaGaugeField::exchangeGhost(), quda::cudaGaugeField::injectGhost(), and quda::cudaGaugeField::saveCPUField().

Here is the call graph for this function:

Here is the caller graph for this function:

copyGenericGaugeDoubleOut()

void quda::copyGenericGaugeDoubleOut	(	GaugeField &	out,
		const GaugeField &	in,
		QudaFieldLocation	location,
		void *	Out,
		void *	In,
		void **	ghostOut,
		void **	ghostIn,
		int	type
	)

Definition at line 5 of file copy_gauge_double.cu.

References copyGauge(), errorQuda, in, out, quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, QUDA_HALF_PRECISION, and QUDA_SINGLE_PRECISION.

Referenced by copyGenericGauge().

Here is the call graph for this function:

Here is the caller graph for this function:

copyGenericGaugeHalfOut()

void quda::copyGenericGaugeHalfOut	(	GaugeField &	out,
		const GaugeField &	in,
		QudaFieldLocation	location,
		void *	Out,
		void *	In,
		void **	ghostOut,
		void **	ghostIn,
		int	type
	)

Definition at line 5 of file copy_gauge_half.cu.

References copyGauge(), errorQuda, in, out, quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, QUDA_HALF_PRECISION, and QUDA_SINGLE_PRECISION.

Referenced by copyGenericGauge().

Here is the call graph for this function:

Here is the caller graph for this function:

copyGenericGaugeMG()

void quda::copyGenericGaugeMG	(	GaugeField &	out,
		const GaugeField &	in,
		QudaFieldLocation	location,
		void *	Out,
		void *	In,
		void **	ghostOut,
		void **	ghostIn,
		int	type
	)

Definition at line 153 of file copy_gauge_mg.cu.

References copyGaugeMG(), errorQuda, in, out, quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, and QUDA_SINGLE_PRECISION.

Referenced by copyGenericGauge().

Here is the call graph for this function:

Here is the caller graph for this function:

copyGenericGaugeSingleOut()

void quda::copyGenericGaugeSingleOut	(	GaugeField &	out,
		const GaugeField &	in,
		QudaFieldLocation	location,
		void *	Out,
		void *	In,
		void **	ghostOut,
		void **	ghostIn,
		int	type
	)

Definition at line 5 of file copy_gauge_single.cu.

References copyGauge(), errorQuda, in, out, quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, QUDA_HALF_PRECISION, and QUDA_SINGLE_PRECISION.

Referenced by copyGenericGauge().

Here is the call graph for this function:

Here is the caller graph for this function:

copyGhost()

template<typename FloatOut , typename FloatIn , int length, typename OutOrder , typename InOrder >

void quda::copyGhost

(

CopyGaugeArg< OutOrder, InOrder >

arg

)

Generic CPU gauge ghost reordering and packing

Definition at line 124 of file copy_gauge_helper.cuh.

References arg(), d, fused_exterior_ndeg_tm_dslash_cuda_gen::i, in, length, quda::gauge::Ncolor(), out, parity, and x.

Here is the call graph for this function:

copyGhostKernel()

template<typename FloatOut , typename FloatIn , int length, typename OutOrder , typename InOrder >

__global__ void quda::copyGhostKernel

(

CopyGaugeArg< OutOrder, InOrder >

arg

)

Generic CUDA kernel for copying the ghost zone. Adopts a similar form as the CPU version, using the same inlined functions.

Definition at line 154 of file copy_gauge_helper.cuh.

References arg(), blockDim, d, fused_exterior_ndeg_tm_dslash_cuda_gen::i, in, length, quda::gauge::Ncolor(), out, parity, and x.

Here is the call graph for this function:

copyInterior() [1/2]

template<typename FloatOut , typename FloatIn , int Ns, int Nc, typename OutOrder , typename InOrder , typename Basis , bool extend>

__device__ __host__ void quda::copyInterior	(	CopySpinorExArg< OutOrder, InOrder, Basis > &	arg,
		int	X
	)

Definition at line 175 of file extended_color_spinor_utilities.cu.

References arg(), d, in, out, R, X, x, x0h, za, and zb.

Here is the call graph for this function:

copyInterior() [2/2]

template<typename FloatOut , typename FloatIn , int Ns, int Nc, typename OutOrder , typename InOrder , typename Basis , bool extend>

void quda::copyInterior

(

CopySpinorExArg< OutOrder, InOrder, Basis > &

arg

)

Definition at line 225 of file extended_color_spinor_utilities.cu.

References arg().

Here is the call graph for this function:

copyInteriorKernel()

template<typename FloatOut , typename FloatIn , int Ns, int Nc, typename OutOrder , typename InOrder , typename Basis , bool extend>

__global__ void quda::copyInteriorKernel

(

CopySpinorExArg< OutOrder, InOrder, Basis >

arg

)

Definition at line 211 of file extended_color_spinor_utilities.cu.

References arg(), blockDim, and gridDim.

Here is the call graph for this function:

copyLinkToArray() [1/2]

void quda::copyLinkToArray ( float *  array, const Matrix< float2, 3 > &  link  )

inline

Definition at line 978 of file quda_matrix.h.

References array, and fused_exterior_ndeg_tm_dslash_cuda_gen::i.

copyLinkToArray() [2/2]

template<class Cmplx , class Real >

void quda::copyLinkToArray ( Real *  array, const Matrix< Cmplx, 3 > &  link  )

inline

Definition at line 992 of file quda_matrix.h.

References array, and fused_exterior_ndeg_tm_dslash_cuda_gen::i.

copyMom()

template<typename FloatOut , typename FloatIn , int length, typename Out , typename In , typename Arg >

void quda::copyMom	(	Arg &	arg,
		const GaugeField &	out,
		const GaugeField &	in,
		QudaFieldLocation	location
	)

Definition at line 257 of file copy_gauge_inc.cu.

References quda::CopyGauge< FloatOut, FloatIn, length, OutOrder, InOrder, isGhost >::apply(), arg(), errorQuda, in, out, QUDA_CPU_FIELD_LOCATION, and QUDA_CUDA_FIELD_LOCATION.

Here is the call graph for this function:

copySpinorEx() [1/2]

template<typename FloatOut , typename FloatIn , int Ns, int Nc, typename OutOrder , typename InOrder , typename Basis >

void quda::copySpinorEx	(	OutOrder	outOrder,
		const InOrder	inOrder,
		const Basis	basis,
		const int *	E,
		const int *	X,
		const int	parity,
		const bool	extend,
		const ColorSpinorField &	meta,
		QudaFieldLocation	location
	)

Definition at line 279 of file extended_color_spinor_utilities.cu.

References quda::CopySpinorEx< FloatOut, FloatIn, Ns, Nc, OutOrder, InOrder, Basis, extend >::apply(), arg(), checkCudaError, E, parity, QUDA_CUDA_FIELD_LOCATION, and X.

Here is the call graph for this function:

copySpinorEx() [2/2]

template<typename FloatOut , typename FloatIn , int Ns, int Nc, typename OutOrder , typename InOrder >

void quda::copySpinorEx	(	OutOrder	outOrder,
		InOrder	inOrder,
		const QudaGammaBasis	outBasis,
		const QudaGammaBasis	inBasis,
		const int *	E,
		const int *	X,
		const int	parity,
		const bool	extend,
		const ColorSpinorField &	meta,
		QudaFieldLocation	location
	)

Definition at line 294 of file extended_color_spinor_utilities.cu.

References E, errorQuda, parity, QUDA_DEGRAND_ROSSI_GAMMA_BASIS, QUDA_UKQCD_GAMMA_BASIS, and X.

cos() [1/3]

template<typename ValueType >

__host__ __device__ ValueType quda::cos ( ValueType  x )

inline

Definition at line 35 of file complex_quda.h.

References cos(), and x.

Referenced by quda::Trig< isHalf, T >::Cos(), cos(), cosh(), exponentiate_iQ(), genGauss(), new_load_half(), polar(), polarSu3(), sin(), quda::Trig< isHalf, T >::SinCos(), sinh(), and tan().

Here is the call graph for this function:

Here is the caller graph for this function:

cos() [2/3]

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::cos ( const complex< ValueType > &  z )

inline

Definition at line 921 of file complex_quda.h.

References cos(), cosh(), sin(), sinh(), and z.

Here is the call graph for this function:

cos() [3/3]

template<>

__host__ __device__ complex<float> quda::cos ( const complex< float > &  z )

inline

Definition at line 929 of file complex_quda.h.

References cosf(), coshf(), sinf(), sinhf(), and z.

Referenced by cos().

Here is the call graph for this function:

Here is the caller graph for this function:

cosh() [1/3]

template<typename ValueType >

__host__ __device__ ValueType quda::cosh ( ValueType  x )

inline

Definition at line 70 of file complex_quda.h.

References cosh(), and x.

Referenced by cos(), cosh(), sin(), and sinh().

Here is the call graph for this function:

Here is the caller graph for this function:

cosh() [2/3]

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::cosh ( const complex< ValueType > &  z )

inline

Definition at line 937 of file complex_quda.h.

References cos(), cosh(), sin(), sinh(), and z.

Here is the call graph for this function:

cosh() [3/3]

template<>

__host__ __device__ complex<float> quda::cosh ( const complex< float > &  z )

inline

Definition at line 945 of file complex_quda.h.

References cosf(), coshf(), sinf(), sinhf(), and z.

Referenced by cosh().

Here is the call graph for this function:

Here is the caller graph for this function:

covDev()

void quda::covDev	(	cudaColorSpinorField *	out,
		cudaGaugeField &	gauge,
		const cudaColorSpinorField *	in,
		const int	parity,
		const int	mu,
		TimeProfile &	profile
	)

Referenced by quda::GaugeCovDev::operator=().

Here is the caller graph for this function:

create_gauge_buffer()

void * quda::create_gauge_buffer	(	size_t	bytes,
		QudaGaugeFieldOrder	order,
		QudaFieldGeometry	geometry
	)

Definition at line 548 of file cuda_gauge_field.cu.

References quda::blas::bytes, d, pool_device_malloc, and QUDA_QDP_GAUGE_ORDER.

Referenced by quda::cudaGaugeField::copy(), quda::cpuGaugeField::copy(), and quda::cudaGaugeField::saveCPUField().

Here is the caller graph for this function:

create_ghost_buffer()

void ** quda::create_ghost_buffer	(	size_t	bytes[],
		QudaGaugeFieldOrder	order,
		QudaFieldGeometry	geometry
	)

Definition at line 559 of file cuda_gauge_field.cu.

References quda::blas::bytes, d, and pool_device_malloc.

Referenced by quda::cudaGaugeField::copy(), quda::cpuGaugeField::copy(), and quda::cudaGaugeField::saveCPUField().

Here is the caller graph for this function:

createDirac()

void quda::createDirac	(	Dirac *&	d,
		Dirac *&	dSloppy,
		Dirac *&	dPre,
		QudaInvertParam &	param,
		const bool	pc_solve
	)

Definition at line 1513 of file interface_quda.cpp.

References quda::Dirac::create(), d, param, QUDA_INC_EIGCG_INVERTER, setDiracParam(), setDiracPreParam(), and setDiracSloppyParam().

Referenced by invertMultiShiftQuda(), invertMultiSrcQuda(), and invertQuda().

Here is the call graph for this function:

Here is the caller graph for this function:

createDslashEvents()

void quda::createDslashEvents

(

)

Definition at line 86 of file dslash_quda.cu.

References quda::dslash::aux_worker, checkCudaError, dslash::commsEnd_d, dslash::commsEnd_h, quda::dslash::dslashStart, quda::dslash::gatherEnd, quda::dslash::gatherStart, fused_exterior_ndeg_tm_dslash_cuda_gen::i, mapped_malloc, Nstream, quda::dslash::packEnd, quda::dslash::scatterEnd, and quda::dslash::scatterStart.

Referenced by initQudaMemory().

Here is the caller graph for this function:

d2i()

__device__ __host__ int quda::d2i ( double  d )

inline

Definition at line 147 of file register_traits.h.

References d.

deserializeTuneCache()

static void quda::deserializeTuneCache ( std::istream &  in )

static

Deserialize tunecache from an istream, useful for reading a file or receiving from other nodes.

Definition at line 116 of file tune.cpp.

References a, quda::TuneKey::aux, quda::TuneKey::aux_n, errorQuda, getline(), in, n, quda::TuneKey::name, quda::TuneKey::name_n, param, snprintf(), tunecache, quda::TuneKey::volume, and quda::TuneKey::volume_n.

Referenced by broadcastTuneCache(), and loadTuneCache().

Here is the call graph for this function:

Here is the caller graph for this function:

destroyDslashEvents()

void quda::destroyDslashEvents

(

)

Definition at line 118 of file dslash_quda.cu.

References checkCudaError, dslash::commsEnd_h, quda::dslash::dslashStart, quda::dslash::gatherEnd, quda::dslash::gatherStart, host_free, fused_exterior_ndeg_tm_dslash_cuda_gen::i, Nstream, quda::dslash::packEnd, quda::dslash::scatterEnd, and quda::dslash::scatterStart.

Referenced by endQuda().

Here is the caller graph for this function:

device_allocated_peak()

long quda::device_allocated_peak

(

)

Returns

peak device memory allocated

Definition at line 57 of file malloc.cpp.

References DEVICE, and max_total_bytes.

device_free_()

void quda::device_free_	(	const char *	func,
		const char *	file,
		int	line,
		void *	ptr
	)

Free device memory allocated with device_malloc(). This function should only be called via the device_free() macro, defined in malloc_quda.h

Definition at line 292 of file malloc.cpp.

References alloc, count, DEVICE, err, errorQuda, func, printfQuda, ptr, and track_free().

Referenced by quda::pool::device_free_(), and quda::pool::device_malloc_().

Here is the call graph for this function:

Here is the caller graph for this function:

device_malloc_()

void * quda::device_malloc_	(	const char *	func,
		const char *	file,
		int	line,
		size_t	size
	)

Perform a standard cudaMalloc() with error-checking. This function should only be called via the device_malloc() macro, defined in malloc_quda.h

Definition at line 167 of file malloc.cpp.

References a, DEVICE, err, errorQuda, func, printfQuda, ptr, size, and track_malloc().

Referenced by quda::pool::device_malloc_().

Here is the call graph for this function:

Here is the caller graph for this function:

device_pinned_free_()

void quda::device_pinned_free_	(	const char *	func,
		const char *	file,
		int	line,
		void *	ptr
	)

Free device memory allocated with device_pinned malloc(). This function should only be called via the device_pinned_free() macro, defined in malloc_quda.h

Definition at line 316 of file malloc.cpp.

References alloc, count, DEVICE, err, errorQuda, func, printfQuda, ptr, and track_free().

Here is the call graph for this function:

device_pinned_malloc_()

void * quda::device_pinned_malloc_	(	const char *	func,
		const char *	file,
		int	line,
		size_t	size
	)

Perform a cuMemAlloc with error-checking. This function is to guarantee a unique memory allocation on the device, since cudaMalloc can be redirected (as is the case with QDPJIT). This should only be called via the device_pinned_malloc() macro, defined in malloc_quda.h.

Definition at line 194 of file malloc.cpp.

References a, DEVICE, err, errorQuda, func, printfQuda, ptr, size, and track_malloc().

Here is the call graph for this function:

disableProfileCount()

void quda::disableProfileCount

(

)

Definition at line 107 of file tune.cpp.

References profile_count.

Referenced by quda::DslashCoarsePolicyTune::DslashCoarsePolicyTune(), anonymous_namespace{dslash_policy.cuh}::DslashPolicyTune::DslashPolicyTune(), and quda::blas::TileSizeTune< ReducerDiagonal, writeDiagonal, ReducerOffDiagonal, writeOffDiagonal >::TileSizeTune().

Here is the caller graph for this function:

domainWallDslashCuda() [1/2]

void quda::domainWallDslashCuda	(	cudaColorSpinorField *	out,
		const cudaGaugeField &	gauge,
		const cudaColorSpinorField *	in,
		const int	parity,
		const int	dagger,
		const cudaColorSpinorField *	x,
		const double &	m_f,
		const double &	k,
		const int *	commDim,
		TimeProfile &	profile
	)

Definition at line 203 of file dslash_domain_wall.cu.

References deg_tm_dslash_cuda_gen::dagger, deg_tm_dslash_cuda_gen::dslash, errorQuda, fused_exterior_ndeg_tm_dslash_cuda_gen::i, in, out, parity, QUDA_DOUBLE_PRECISION, QUDA_HALF_PRECISION, QUDA_MAX_DIM, QUDA_SINGLE_PRECISION, and x.

Referenced by quda::DiracDomainWall::Dslash(), quda::DiracDomainWall4DPC::Dslash4(), quda::DiracDomainWall4DPC::Dslash4Xpay(), quda::DiracDomainWall4DPC::Dslash5(), quda::DiracDomainWall4DPC::Dslash5inv(), quda::DiracDomainWall4DPC::Dslash5invXpay(), quda::DiracDomainWall4DPC::Dslash5Xpay(), and quda::DiracDomainWall::DslashXpay().

Here is the caller graph for this function:

domainWallDslashCuda() [2/2]

void quda::domainWallDslashCuda	(	cudaColorSpinorField *	out,
		const cudaGaugeField &	gauge,
		const cudaColorSpinorField *	in,
		const int	parity,
		const int	dagger,
		const cudaColorSpinorField *	x,
		const double &	m_f,
		const double &	a,
		const double &	b,
		const int *	commDim,
		const int	DS_type,
		TimeProfile &	profile
	)

Definition at line 252 of file dslash_domain_wall_4d.cu.

References a, b, deg_tm_dslash_cuda_gen::dagger, deg_tm_dslash_cuda_gen::dslash, errorQuda, fused_exterior_ndeg_tm_dslash_cuda_gen::i, in, out, parity, QUDA_DOUBLE_PRECISION, QUDA_HALF_PRECISION, QUDA_MAX_DIM, QUDA_SINGLE_PRECISION, and x.

enableProfileCount()

void quda::enableProfileCount

(

)

Definition at line 108 of file tune.cpp.

References profile_count.

Referenced by quda::DslashCoarsePolicyTune::DslashCoarsePolicyTune(), anonymous_namespace{dslash_policy.cuh}::DslashPolicyTune::DslashPolicyTune(), and quda::blas::TileSizeTune< ReducerDiagonal, writeDiagonal, ReducerOffDiagonal, writeOffDiagonal >::TileSizeTune().

Here is the caller graph for this function:

ErrorSU3()

template<class Cmplx >

__device__ __host__ double quda::ErrorSU3

(

const Matrix< Cmplx, 3 > &

matrix

)

Definition at line 1083 of file quda_matrix.h.

References conj(), fused_exterior_ndeg_tm_dslash_cuda_gen::i, and norm().

Referenced by computeOvrImpSTOUTStep().

Here is the call graph for this function:

Here is the caller graph for this function:

exchangeExtendedGhost()

void quda::exchangeExtendedGhost	(	cudaColorSpinorField *	spinor,
		int	R[],
		int	parity,
		cudaStream_t *	stream_p
	)

Definition at line 25 of file extended_color_spinor_utilities.cu.

References commDim(), deg_tm_dslash_cuda_gen::dagger, dim, dslash::gatherEnd, fused_exterior_ndeg_tm_dslash_cuda_gen::i, parity, qudaDeviceSynchronize(), qudaEventRecord(), R, spinor, and streams.

Referenced by gaussGaugeQuda(), and quda::XSD::operator()().

Here is the call graph for this function:

Here is the caller graph for this function:

exp() [1/3]

template<typename ValueType >

__host__ __device__ ValueType quda::exp ( ValueType  x )

inline

Definition at line 85 of file complex_quda.h.

References exp(), and x.

Referenced by exp(), pow(), tanh(), and test().

Here is the call graph for this function:

Here is the caller graph for this function:

exp() [2/3]

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::exp ( const complex< ValueType > &  z )

inline

Definition at line 954 of file complex_quda.h.

References exp(), polar(), and z.

Here is the call graph for this function:

exp() [3/3]

template<>

__host__ __device__ complex<float> quda::exp ( const complex< float > &  z )

inline

Definition at line 960 of file complex_quda.h.

References expf(), polar(), and z.

Referenced by exp().

Here is the call graph for this function:

Here is the caller graph for this function:

exponentiate_iQ()

template<class T >

__device__ __host__ void quda::exponentiate_iQ ( const Matrix< T, 3 > &  Q, Matrix< T, 3 > *  exp_iQ  )

inline

Definition at line 1110 of file quda_matrix.h.

References acos(), cos(), getDeterminant(), getTrace(), parity, pow(), setIdentity(), setZero(), sin(), sqrt(), and x.

Referenced by computeOvrImpSTOUTStep().

Here is the call graph for this function:

Here is the caller graph for this function:

extendedCopyColorSpinor() [1/2]

template<typename FloatOut , typename FloatIn , int Ns, int Nc, typename InOrder >

void quda::extendedCopyColorSpinor	(	InOrder &	inOrder,
		ColorSpinorField &	out,
		QudaGammaBasis	inBasis,
		const int *	E,
		const int *	X,
		const int	parity,
		const bool	extend,
		QudaFieldLocation	location,
		FloatOut *	Out,
		float *	outNorm
	)

Definition at line 321 of file extended_color_spinor_utilities.cu.

References E, errorQuda, out, parity, and X.

extendedCopyColorSpinor() [2/2]

template<typename FloatOut , typename FloatIn , int Ns, int Nc>

void quda::extendedCopyColorSpinor	(	ColorSpinorField &	out,
		const ColorSpinorField &	in,
		const int	parity,
		const QudaFieldLocation	location,
		FloatOut *	Out,
		FloatIn *	In,
		float *	outNorm,
		float *	inNorm
	)

Definition at line 337 of file extended_color_spinor_utilities.cu.

References d, E, errorQuda, in, out, parity, and X.

extractExtendedGaugeGhost()

void quda::extractExtendedGaugeGhost	(	const GaugeField &	u,
		int	dim,
		const int *	R,
		void **	ghost,
		bool	extract
	)

This function is used for extracting the gauge ghost zone from a gauge field array. Defined in extract_gauge_ghost.cu.

Parameters

u	The gauge field from which we want to extract/pack the ghost zone
dim	The dimension in which we are packing/unpacking
ghost	The array where we want to pack/unpack the ghost zone into/from
extract	Whether we are extracting into ghost or injecting from ghost

Definition at line 422 of file extract_gauge_ghost_extended.cu.

References dim, errorQuda, extractGhostEx(), quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, QUDA_HALF_PRECISION, QUDA_SINGLE_PRECISION, and R.

Referenced by quda::cudaGaugeField::exchangeExtendedGhost(), and quda::cpuGaugeField::exchangeExtendedGhost().

Here is the call graph for this function:

Here is the caller graph for this function:

extractGaugeGhost()

void quda::extractGaugeGhost	(	const GaugeField &	u,
		void **	ghost,
		bool	extract = true,
		int	offset = 0
	)

This function is used for extracting the gauge ghost zone from a gauge field array. Defined in extract_gauge_ghost.cu.

Parameters

u	The gauge field from which we want to extract the ghost zone
ghost	The array where we want to pack the ghost zone into
extract	Where we are extracting into ghost or injecting from ghost
offset	By default we exchange the nDim site-vector of links in the first nDim dimensions; offset allows us to instead exchange the links in nDim+offset dimensions. This is used to faciliate sending bi-directional links which is needed for the coarse links.

Definition at line 103 of file extract_gauge_ghost.cu.

References errorQuda, extractGaugeGhostMG(), extractGhost(), quda::GaugeField::Ncolor(), offset, quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, QUDA_HALF_PRECISION, and QUDA_SINGLE_PRECISION.

Referenced by quda::cudaGaugeField::exchangeGhost(), quda::cpuGaugeField::exchangeGhost(), quda::cudaGaugeField::injectGhost(), and quda::cpuGaugeField::injectGhost().

Here is the call graph for this function:

Here is the caller graph for this function:

extractGaugeGhostMG()

void quda::extractGaugeGhostMG	(	const GaugeField &	u,
		void **	ghost,
		bool	extract,
		int	offset
	)

Definition at line 74 of file extract_gauge_ghost_mg.cu.

References errorQuda, extractGhostMG(), offset, quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, and QUDA_SINGLE_PRECISION.

Referenced by extractGaugeGhost().

Here is the call graph for this function:

Here is the caller graph for this function:

extractGhost() [1/3]

template<typename Float >

void quda::extractGhost	(	const GaugeField &	u,
		Float **	Ghost,
		bool	extract,
		int	offset
	)

This is the template driver for extractGhost

Definition at line 10 of file extract_gauge_ghost.cu.

References errorQuda, quda::GaugeField::isNative(), length, quda::GaugeField::LinkType(), offset, quda::GaugeField::Order(), QUDA_ASQTAD_FAT_LINKS, QUDA_BQCD_GAUGE_ORDER, QUDA_CPS_WILSON_GAUGE_ORDER, QUDA_CPU_FIELD_LOCATION, QUDA_CUDA_FIELD_LOCATION, QUDA_MILC_GAUGE_ORDER, QUDA_QDP_GAUGE_ORDER, QUDA_QDPJIT_GAUGE_ORDER, QUDA_RECONSTRUCT_12, QUDA_RECONSTRUCT_13, QUDA_RECONSTRUCT_8, QUDA_RECONSTRUCT_9, QUDA_RECONSTRUCT_NO, QUDA_TIFR_GAUGE_ORDER, QUDA_TIFR_PADDED_GAUGE_ORDER, and quda::GaugeField::Reconstruct().

Referenced by extractGaugeGhost().

Here is the call graph for this function:

Here is the caller graph for this function:

extractGhost() [2/3]

template<typename Float , int length, int nDim, typename Order , bool extract>

void quda::extractGhost

(

ExtractGhostArg< Order, nDim >

arg

)

Generic CPU gauge ghost extraction and packing NB This routines is specialized to four dimensions

Definition at line 39 of file extract_gauge_ghost_helper.cuh.

References a, arg(), b, c, d, dim, fused_exterior_ndeg_tm_dslash_cuda_gen::i, length, quda::gauge::Ncolor(), and parity.

Here is the call graph for this function:

extractGhost() [3/3]

template<typename Float , int length, typename Order >

void quda::extractGhost	(	Order	order,
		const GaugeField &	u,
		QudaFieldLocation	location,
		bool	extract,
		int	offset
	)

Generic gauge ghost extraction and packing (or the converse) NB This routines is specialized to four dimensions

Definition at line 229 of file extract_gauge_ghost_helper.cuh.

References arg(), commDim(), dim, extractor(), f, quda::GaugeField::Nface(), offset, X, and quda::LatticeField::X().

Here is the call graph for this function:

extractGhostEx() [1/3]

template<typename Float , int length, int nDim, int dim, typename Order , bool extract>

void quda::extractGhostEx

(

ExtractGhostExArg< Order, nDim, dim >

arg

)

Generic CPU gauge ghost extraction and packing NB This routines is specialized to four dimensions

Definition at line 96 of file extract_gauge_ghost_extended.cu.

References a, arg(), b, c, d, dim, and parity.

Referenced by extractExtendedGaugeGhost().

Here is the call graph for this function:

Here is the caller graph for this function:

extractGhostEx() [2/3]

template<typename Float , int length, typename Order >

void quda::extractGhostEx	(	Order	order,
		const int	dim,
		const int *	surfaceCB,
		const int *	E,
		const int *	R,
		bool	extract,
		const GaugeField &	u,
		QudaFieldLocation	location
	)

Generic CPU gauge ghost extraction and packing NB This routines is specialized to four dimensions

Parameters

E	the extended gauge dimensions
R	array holding the radius of the extended region
extract	Whether we are extracting or injecting the ghost zone

Definition at line 256 of file extract_gauge_ghost_extended.cu.

References arg(), C0, C1, checkCudaError, commDim(), d, dim, E, errorQuda, extractor(), R, and X.

Here is the call graph for this function:

extractGhostEx() [3/3]

template<typename Float >

void quda::extractGhostEx	(	const GaugeField &	u,
		int	dim,
		const int *	R,
		Float **	Ghost,
		bool	extract
	)

This is the template driver for extractGhost

Definition at line 328 of file extract_gauge_ghost_extended.cu.

References dim, errorQuda, quda::GaugeField::isNative(), length, quda::GaugeField::LinkType(), quda::GaugeField::Order(), QUDA_ASQTAD_FAT_LINKS, QUDA_BQCD_GAUGE_ORDER, QUDA_CPS_WILSON_GAUGE_ORDER, QUDA_CPU_FIELD_LOCATION, QUDA_CUDA_FIELD_LOCATION, QUDA_MILC_GAUGE_ORDER, QUDA_QDP_GAUGE_ORDER, QUDA_QDPJIT_GAUGE_ORDER, QUDA_RECONSTRUCT_12, QUDA_RECONSTRUCT_13, QUDA_RECONSTRUCT_8, QUDA_RECONSTRUCT_9, QUDA_RECONSTRUCT_NO, QUDA_TIFR_GAUGE_ORDER, R, quda::GaugeField::Reconstruct(), quda::LatticeField::SurfaceCB(), and quda::LatticeField::X().

Here is the call graph for this function:

extractGhostExKernel()

template<typename Float , int length, int nDim, int dim, typename Order , bool extract>

__global__ void quda::extractGhostExKernel

(

ExtractGhostExArg< Order, nDim, dim >

arg

)

Generic GPU gauge ghost extraction and packing NB This routines is specialized to four dimensions FIXME this implementation will have two-way warp divergence Generic CPU gauge ghost extraction and packing NB This routines is specialized to four dimensions

Definition at line 141 of file extract_gauge_ghost_extended.cu.

References a, arg(), b, blockDim, c, d, dim, parity, and X.

Here is the call graph for this function:

extractGhostKernel()

template<typename Float , int length, int nDim, typename Order , bool extract>

__global__ void quda::extractGhostKernel

(

ExtractGhostArg< Order, nDim >

arg

)

Generic GPU gauge ghost extraction and packing NB This routines is specialized to four dimensions FIXME this implementation will have two-way warp divergence

Definition at line 106 of file extract_gauge_ghost_helper.cuh.

References a, arg(), b, blockDim, c, d, dim, fused_exterior_ndeg_tm_dslash_cuda_gen::i, length, quda::gauge::Ncolor(), parity, and X.

Here is the call graph for this function:

extractGhostMG() [1/2]

template<typename Float , int Nc>

void quda::extractGhostMG	(	const GaugeField &	u,
		Float **	Ghost,
		bool	extract,
		int	offset
	)

This is the template driver for extractGhost

Definition at line 15 of file extract_gauge_ghost_mg.cu.

References errorQuda, quda::GaugeField::isNative(), length, offset, quda::GaugeField::Order(), QUDA_CPU_FIELD_LOCATION, QUDA_CUDA_FIELD_LOCATION, QUDA_QDP_GAUGE_ORDER, QUDA_RECONSTRUCT_NO, and quda::GaugeField::Reconstruct().

Referenced by extractGaugeGhostMG().

Here is the call graph for this function:

Here is the caller graph for this function:

extractGhostMG() [2/2]

template<typename Float >

void quda::extractGhostMG	(	const GaugeField &	u,
		Float **	Ghost,
		bool	extract,
		int	offset
	)

This is the template driver for extractGhost

Definition at line 53 of file extract_gauge_ghost_mg.cu.

References errorQuda, quda::GaugeField::LinkType(), quda::GaugeField::Ncolor(), offset, QUDA_COARSE_LINKS, QUDA_RECONSTRUCT_NO, and quda::GaugeField::Reconstruct().

Here is the call graph for this function:

extractor()

template<typename Float , int length, int dim, typename Arg >

__device__ __host__ void quda::extractor	(	Arg &	arg,
		int	dir,
		int	a,
		int	b,
		int	c,
		int	d,
		int	g,
		int	parity
	)

Definition at line 54 of file extract_gauge_ghost_extended.cu.

References a, arg(), b, c, d, dim, length, and parity.

Referenced by extractGhost(), and extractGhostEx().

Here is the call graph for this function:

Here is the caller graph for this function:

f2i()

__device__ __host__ int quda::f2i ( float  f )

inline

Definition at line 138 of file register_traits.h.

References f.

Referenced by copy().

Here is the caller graph for this function:

fatLongKSLink()

void quda::fatLongKSLink	(	cudaGaugeField *	fat,
		cudaGaugeField *	lng,
		const cudaGaugeField &	gauge,
		const double *	coeff
	)

Compute the fat and long links for an improved staggered (Kogut-Susskind) fermions.

Parameters

fat[out]	The computed fat link
lng[out]	The computed long link (only computed if lng!=0)
u[in]	The input gauge field
coeff[in]	Array of path coefficients

Definition at line 524 of file llfat_quda.cu.

References checkCudaError, dw_dslash_4D_cuda_gen::coeff(), quda::GaugeFieldParam::create, errorQuda, fabs(), gParam, MIN_COEFF, quda::LatticeFieldParam::precision, QUDA_NULL_FIELD_CREATE, QUDA_RECONSTRUCT_NO, qudaDeviceSynchronize(), quda::GaugeFieldParam::reconstruct, quda::GaugeField::Reconstruct(), quda::GaugeFieldParam::setPrecision(), and quda::LatticeField::X().

Referenced by computeKSLinkQuda().

Here is the call graph for this function:

Here is the caller graph for this function:

file_name()

constexpr const char* quda::file_name ( const char *  str )

inline

Definition at line 48 of file malloc_quda.h.

References r_slant(), str_end(), and str_slant().

Here is the call graph for this function:

fillEigCGInnerSolverParam()

static void quda::fillEigCGInnerSolverParam ( SolverParam &  inner, const SolverParam &  outer, bool  use_sloppy_partial_accumulator = true  )

static

Definition at line 210 of file inv_eigcg_quda.cpp.

References quda::SolverParam::delta, e, quda::SolverParam::gflops, quda::SolverParam::inv_type, quda::SolverParam::inv_type_precondition, quda::SolverParam::is_preconditioner, quda::SolverParam::iter, quda::SolverParam::maxiter, quda::SolverParam::maxiter_precondition, quda::SolverParam::precision, quda::SolverParam::precision_precondition, quda::SolverParam::precision_sloppy, quda::SolverParam::preserve_source, QUDA_EIGCG_INVERTER, QUDA_INVALID_INVERTER, QUDA_PRESERVE_SOURCE_NO, QUDA_PRESERVE_SOURCE_YES, quda::SolverParam::secs, quda::SolverParam::tol, quda::SolverParam::tol_precondition, and quda::SolverParam::use_sloppy_partial_accumulator.

Referenced by quda::IncEigCG::IncEigCG().

Here is the caller graph for this function:

fillFGMResDRInnerSolveParam()

void quda::fillFGMResDRInnerSolveParam	(	SolverParam &	inner,
		const SolverParam &	outer
	)

Definition at line 187 of file inv_gmresdr_quda.cpp.

References quda::SolverParam::delta, e, quda::SolverParam::gflops, quda::SolverParam::global_reduction, quda::SolverParam::inv_type, quda::SolverParam::inv_type_precondition, quda::SolverParam::is_preconditioner, quda::SolverParam::iter, quda::SolverParam::maxiter, quda::SolverParam::maxiter_precondition, quda::SolverParam::precision, quda::SolverParam::precision_precondition, quda::SolverParam::precision_sloppy, quda::SolverParam::preserve_source, QUDA_INVALID_INVERTER, QUDA_PRESERVE_SOURCE_NO, QUDA_PRESERVE_SOURCE_YES, quda::SolverParam::secs, quda::SolverParam::tol, quda::SolverParam::tol_precondition, and warningQuda.

Referenced by quda::GMResDR::GMResDR().

Here is the caller graph for this function:

fillInitCGSolverParam()

static void quda::fillInitCGSolverParam ( SolverParam &  inner, const SolverParam &  outer  )

static

Definition at line 233 of file inv_eigcg_quda.cpp.

References quda::SolverParam::delta, quda::SolverParam::gflops, quda::SolverParam::inv_type, quda::SolverParam::iter, quda::SolverParam::maxiter, quda::SolverParam::precision, quda::SolverParam::precision_precondition, quda::SolverParam::precision_sloppy, QUDA_CG_INVERTER, QUDA_USE_INIT_GUESS_YES, quda::SolverParam::secs, quda::SolverParam::tol, quda::SolverParam::tol_restart, quda::SolverParam::use_init_guess, and quda::SolverParam::use_sloppy_partial_accumulator.

Referenced by quda::IncEigCG::IncEigCG().

Here is the caller graph for this function:

fillInnerSolveParam()

void quda::fillInnerSolveParam	(	SolverParam &	inner,
		const SolverParam &	outer
	)

Definition at line 25 of file inv_gcr_quda.cpp.

References quda::SolverParam::delta, e, quda::SolverParam::gflops, quda::SolverParam::global_reduction, quda::SolverParam::inv_type, quda::SolverParam::inv_type_precondition, quda::SolverParam::is_preconditioner, quda::SolverParam::iter, quda::SolverParam::maxiter, quda::SolverParam::maxiter_precondition, quda::SolverParam::precision, quda::SolverParam::precision_precondition, quda::SolverParam::precision_sloppy, quda::SolverParam::preserve_source, QUDA_GCR_INVERTER, QUDA_INVALID_INVERTER, QUDA_PRESERVE_SOURCE_NO, QUDA_PRESERVE_SOURCE_YES, QUDA_USE_INIT_GUESS_NO, quda::SolverParam::secs, quda::SolverParam::tol, quda::SolverParam::tol_precondition, and quda::SolverParam::use_init_guess.

Referenced by quda::GCR::GCR(), and quda::BiCGstab::operator()().

Here is the caller graph for this function:

fillInnerSolverParam()

static void quda::fillInnerSolverParam ( SolverParam &  inner, const SolverParam &  outer  )

static

Definition at line 18 of file inv_pcg_quda.cpp.

References quda::SolverParam::delta, e, quda::SolverParam::gflops, quda::SolverParam::inv_type, quda::SolverParam::inv_type_precondition, quda::SolverParam::is_preconditioner, quda::SolverParam::iter, quda::SolverParam::maxiter, quda::SolverParam::maxiter_precondition, quda::SolverParam::precision, quda::SolverParam::precision_precondition, quda::SolverParam::precision_sloppy, quda::SolverParam::preserve_source, QUDA_INVALID_INVERTER, QUDA_PCG_INVERTER, QUDA_PRESERVE_SOURCE_NO, QUDA_PRESERVE_SOURCE_YES, quda::SolverParam::secs, quda::SolverParam::tol, and quda::SolverParam::tol_precondition.

Referenced by quda::PreconCG::PreconCG().

Here is the caller graph for this function:

FillV()

void quda::FillV	(	ColorSpinorField &	V,
		const std::vector< ColorSpinorField *> &	B,
		int	Nvec
	)

Helper method that takes a vector of ColorSpinorFields and packes them into a single matrix field.

Parameters

[out]	V	The resulting packed matrix field
[in]	B	Vector of ColorSpinorFields to be packed
[in]	Nvec	Vector length

Definition at line 172 of file transfer_util.cu.

References errorQuda, QUDA_DOUBLE_PRECISION, QUDA_SINGLE_PRECISION, and V.

Referenced by quda::Transfer::fillV().

Here is the caller graph for this function:

flushProfile()

void quda::flushProfile

(

)

Flush profile contents, setting all counts to zero.

Definition at line 462 of file tune.cpp.

References entry, param, and tunecache.

Referenced by newDeflationQuda(), and newMultigridQuda().

Here is the caller graph for this function:

free_gauge_buffer()

void quda::free_gauge_buffer	(	void *	buffer,
		QudaGaugeFieldOrder	order,
		QudaFieldGeometry	geometry
	)

Definition at line 571 of file cuda_gauge_field.cu.

References d, pool_device_free, and QUDA_QDP_GAUGE_ORDER.

Referenced by quda::cudaGaugeField::copy(), quda::cpuGaugeField::copy(), and quda::cudaGaugeField::saveCPUField().

Here is the caller graph for this function:

free_ghost_buffer()

void quda::free_ghost_buffer	(	void **	buffer,
		QudaGaugeFieldOrder	order,
		QudaFieldGeometry	geometry
	)

Definition at line 580 of file cuda_gauge_field.cu.

References d, and pool_device_free.

Referenced by quda::cudaGaugeField::copy(), quda::cpuGaugeField::copy(), and quda::cudaGaugeField::saveCPUField().

Here is the caller graph for this function:

gamma5()

void quda::gamma5	(	ColorSpinorField &	out,
		const ColorSpinorField &	in
	)

Applies a gamma5 matrix to a spinor (wrapper to ApplyGamma)

Parameters

[out]	out	Output field
[in]	in	Input field

Definition at line 427 of file dslash_quda.cu.

References ApplyGamma(), in, and out.

Referenced by computeCloverForceQuda().

Here is the call graph for this function:

Here is the caller graph for this function:

gammaCPU()

template<typename Float , int nColor, typename Arg >

void quda::gammaCPU

(

Arg

arg

)

Definition at line 195 of file dslash_quda.cu.

References arg(), in, and parity.

Here is the call graph for this function:

gammaGPU()

template<typename Float , int nColor, int d, typename Arg >

__global__ void quda::gammaGPU

(

Arg

arg

)

Definition at line 210 of file dslash_quda.cu.

References arg(), blockDim, d, in, and parity.

Here is the call graph for this function:

GaugeFixHit_AtomicAdd() [1/2]

template<int blockSize, typename Float , int gauge_dir, int NCOLORS>

__forceinline__ __device__ void quda::GaugeFixHit_AtomicAdd	(	Matrix< complex< Float >, NCOLORS > &	link,
		const Float	relax_boost,
		const int	tid
	)

Device function to perform gauge fixing with overrelxation. Uses 8 treads per lattice site, the reduction is performed by shared memory without using atomicadd. This implementation needs 8x more shared memory than the implementation using atomicadd

Definition at line 69 of file gauge_fix_ovr_hit_devf.cuh.

References __syncthreads(), deg_tm_dslash_cuda_gen::block(), blockSize, p, x, and y.

Here is the call graph for this function:

GaugeFixHit_AtomicAdd() [2/2]

template<int blockSize, typename Float , int gauge_dir, int NCOLORS>

__forceinline__ __device__ void quda::GaugeFixHit_AtomicAdd	(	Matrix< complex< Float >, NCOLORS > &	link,
		Matrix< complex< Float >, NCOLORS > &	link1,
		const Float	relax_boost,
		const int	tid
	)

Device function to perform gauge fixing with overrelxation. Uses 8 treads per lattice site, the reduction is performed by shared memory without using atomicadd. This implementation needs 8x more shared memory than the implementation using atomicadd

Definition at line 392 of file gauge_fix_ovr_hit_devf.cuh.

References __syncthreads(), deg_tm_dslash_cuda_gen::block(), blockSize, p, x, and y.

Here is the call graph for this function:

GaugeFixHit_NoAtomicAdd() [1/2]

template<int blockSize, typename Float , int gauge_dir, int NCOLORS>

__forceinline__ __device__ void quda::GaugeFixHit_NoAtomicAdd	(	Matrix< complex< Float >, NCOLORS > &	link,
		const Float	relax_boost,
		const int	tid
	)

Device function to perform gauge fixing with overrelxation. Uses 4 treads per lattice site, the reduction is performed by shared memory using atomicadd.

Definition at line 159 of file gauge_fix_ovr_hit_devf.cuh.

References __syncthreads(), fused_exterior_ndeg_tm_dslash_cuda_gen::a1, fused_exterior_ndeg_tm_dslash_cuda_gen::a2, deg_tm_dslash_cuda_gen::block(), blockSize, fused_exterior_ndeg_tm_dslash_cuda_gen::i, p, and x.

Here is the call graph for this function:

GaugeFixHit_NoAtomicAdd() [2/2]

template<int blockSize, typename Float , int gauge_dir, int NCOLORS>

__forceinline__ __device__ void quda::GaugeFixHit_NoAtomicAdd	(	Matrix< complex< Float >, NCOLORS > &	link,
		Matrix< complex< Float >, NCOLORS > &	link1,
		const Float	relax_boost,
		const int	tid
	)

Device function to perform gauge fixing with overrelxation. Uses 4 treads per lattice site, the reduction is performed by shared memory using atomicadd.

Definition at line 486 of file gauge_fix_ovr_hit_devf.cuh.

References __syncthreads(), fused_exterior_ndeg_tm_dslash_cuda_gen::a1, fused_exterior_ndeg_tm_dslash_cuda_gen::a2, deg_tm_dslash_cuda_gen::block(), blockSize, fused_exterior_ndeg_tm_dslash_cuda_gen::i, p, and x.

Here is the call graph for this function:

GaugeFixHit_NoAtomicAdd_LessSM() [1/2]

template<int blockSize, typename Float , int gauge_dir, int NCOLORS>

__forceinline__ __device__ void quda::GaugeFixHit_NoAtomicAdd_LessSM	(	Matrix< complex< Float >, NCOLORS > &	link,
		const Float	relax_boost,
		const int	tid
	)

Device function to perform gauge fixing with overrelxation. Uses 8 treads per lattice site, the reduction is performed by shared memory without using atomicadd. This implementation uses the same amount of shared memory as the atomicadd implementation with more thread block synchronization

Definition at line 254 of file gauge_fix_ovr_hit_devf.cuh.

References __syncthreads(), deg_tm_dslash_cuda_gen::block(), blockSize, p, and x.

Here is the call graph for this function:

GaugeFixHit_NoAtomicAdd_LessSM() [2/2]

template<int blockSize, typename Float , int gauge_dir, int NCOLORS>

__forceinline__ __device__ void quda::GaugeFixHit_NoAtomicAdd_LessSM	(	Matrix< complex< Float >, NCOLORS > &	link,
		Matrix< complex< Float >, NCOLORS > &	link1,
		const Float	relax_boost,
		const int	tid
	)

Device function to perform gauge fixing with overrelxation. Uses 4 treads per lattice site, the reduction is performed by shared memory without using atomicadd. This implementation uses the same amount of shared memory as the atomicadd implementation with more thread block synchronization

Definition at line 563 of file gauge_fix_ovr_hit_devf.cuh.

References __syncthreads(), deg_tm_dslash_cuda_gen::block(), blockSize, p, and x.

Here is the call graph for this function:

gaugefixingFFT()

void quda::gaugefixingFFT	(	cudaGaugeField &	data,
		const int	gauge_dir,
		const int	Nsteps,
		const int	verbose_interval,
		const double	alpha,
		const int	autotune,
		const double	tolerance,
		const int	stopWtheta
	)

Gauge fixing with Steepest descent method with FFTs with support for single GPU only.

Parameters

[in,out]	data,quda	gauge field
[in]	gauge_dir,3	for Coulomb gauge fixing, other for Landau gauge fixing
[in]	Nsteps,maximum	number of steps to perform gauge fixing
[in]	verbose_interval,print	gauge fixing info when iteration count is a multiple of this
[in]	alpha,gauge	fixing parameter of the method, most common value is 0.08
[in]	autotune,1	to autotune the method, i.e., if the Fg inverts its tendency we decrease the alpha value
[in]	tolerance,torelance	value to stop the method, if this value is zero then the method stops when iteration reachs the maximum number of steps defined by Nsteps
[in]	stopWtheta,0	for MILC criterium and 1 to use the theta value

Definition at line 1202 of file gauge_fix_fft.cu.

References comm_dim_partitioned(), errorQuda, float, quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, QUDA_HALF_PRECISION, and QUDA_SINGLE_PRECISION.

Referenced by computeGaugeFixingFFTQuda(), and TEST_F().

Here is the call graph for this function:

Here is the caller graph for this function:

gaugefixingOVR()

void quda::gaugefixingOVR	(	cudaGaugeField &	data,
		const int	gauge_dir,
		const int	Nsteps,
		const int	verbose_interval,
		const double	relax_boost,
		const double	tolerance,
		const int	reunit_interval,
		const int	stopWtheta
	)

Gauge fixing with overrelaxation with support for single and multi GPU.

Parameters

[in,out]	data,quda	gauge field
[in]	gauge_dir,3	for Coulomb gauge fixing, other for Landau gauge fixing
[in]	Nsteps,maximum	number of steps to perform gauge fixing
[in]	verbose_interval,print	gauge fixing info when iteration count is a multiple of this
[in]	relax_boost,gauge	fixing parameter of the overrelaxation method, most common value is 1.5 or 1.7.
[in]	tolerance,torelance	value to stop the method, if this value is zero then the method stops when iteration reachs the maximum number of steps defined by Nsteps
[in]	reunit_interval,reunitarize	gauge field when iteration count is a multiple of this
[in]	stopWtheta,0	for MILC criterium and 1 to use the theta value

Definition at line 1790 of file gauge_fix_ovr.cu.

References errorQuda, float, quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, QUDA_HALF_PRECISION, and QUDA_SINGLE_PRECISION.

Referenced by computeGaugeFixingOVRQuda(), and TEST_F().

Here is the call graph for this function:

Here is the caller graph for this function:

gaugeForce()

void quda::gaugeForce	(	GaugeField &	mom,
		const GaugeField &	u,
		double	coeff,
		int ***	input_path,
		int *	length,
		double *	path_coeff,
		int	num_paths,
		int	max_length
	)

Compute the gauge-force contribution to the momentum.

Parameters

[out]	mom	Momentum field
[in]	u	Gauge field (extended when running no multiple GPUs)
[in]	coeff	Step-size coefficient
[in]	input_path	Host-array holding all path contributions for the gauge action
[in]	length	Host array holding the length of all paths
[in]	path_coeff	Coefficient of each path
[in]	num_paths	Numer of paths
[in]	max_length	Maximum length of each path

Definition at line 339 of file gauge_force.cu.

References dw_dslash_4D_cuda_gen::coeff(), errorQuda, length, quda::LatticeField::Location(), quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, and QUDA_SINGLE_PRECISION.

Referenced by computeGaugeForceQuda().

Here is the call graph for this function:

Here is the caller graph for this function:

gaugeGauss()

void quda::gaugeGauss	(	GaugeField &	dataDs,
		RNG &	rngstate
	)

Generate Gaussian distributed GaugeField

Parameters

dataDs	The GaugeField
rngstate	random states

Definition at line 182 of file gauge_random.cu.

References errorQuda, quda::GaugeField::isNative(), quda::GaugeField::Order(), quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, QUDA_HALF_PRECISION, QUDA_SINGLE_PRECISION, and quda::GaugeField::Reconstruct().

Referenced by gaussGaugeQuda().

Here is the call graph for this function:

Here is the caller graph for this function:

gaussSpinor() [1/3]

template<typename FloatIn , int Ns, int Nc, typename InOrder >

void quda::gaussSpinor	(	InOrder &	inOrder,
		int	volume,
		RNG	rngstate
	)

CPU function to reorder spinor fields.

Definition at line 32 of file spinor_gauss.cu.

References c, s, quda::RNG::State(), and x.

Here is the call graph for this function:

gaussSpinor() [2/3]

template<typename FloatIn , int Ns, int Nc, typename InOrder >

void quda::gaussSpinor	(	InOrder &	inOrder,
		const ColorSpinorField &	meta,
		RNG &	rngstate
	)

Definition at line 103 of file spinor_gauss.cu.

References quda::GaussSpinor< FloatIn, Ns, Nc, InOrder >::apply().

Here is the call graph for this function:

gaussSpinor() [3/3]

template<typename FloatIn , int Ns, int Nc>

void quda::gaussSpinor	(	ColorSpinorField &	in,
		RNG &	rngstate
	)

Decide on the input order

Definition at line 110 of file spinor_gauss.cu.

References errorQuda, in, QUDA_FLOAT2_FIELD_ORDER, and QUDA_SPACE_SPIN_COLOR_FIELD_ORDER.

gaussSpinorKernel()

template<typename FloatIn , int Ns, int Nc, typename InOrder >

__global__ void quda::gaussSpinorKernel	(	InOrder	inOrder,
		int	volume,
		RNG	rngstate
	)

CUDA kernel to reorder spinor fields. Adopts a similar form as the CPU version, using the same inlined functions.

Definition at line 47 of file spinor_gauss.cu.

References blockDim, c, s, quda::RNG::State(), and x.

Here is the call graph for this function:

genericCompare()

int quda::genericCompare	(	const cpuColorSpinorField &	a,
		const cpuColorSpinorField &	b,
		int	tol
	)

Definition at line 204 of file color_spinor_util.cu.

References a, b, compareSpinor(), errorQuda, ret, and tol.

Referenced by quda::cpuColorSpinorField::Compare().

Here is the call graph for this function:

Here is the caller graph for this function:

genericCopyColorSpinor() [1/6]

template<typename FloatOut , typename FloatIn , int Ns, int Nc, typename OutOrder , typename InOrder >

void quda::genericCopyColorSpinor	(	OutOrder &	outOrder,
		const InOrder &	inOrder,
		const ColorSpinorField &	out,
		QudaFieldLocation	location
	)

Definition at line 84 of file copy_color_spinor_mg.cuh.

References copy(), and out.

Here is the call graph for this function:

genericCopyColorSpinor() [2/6]

template<typename FloatOut , typename FloatIn , int Ns, int Nc, typename InOrder >

void quda::genericCopyColorSpinor	(	InOrder &	inOrder,
		ColorSpinorField &	out,
		QudaFieldLocation	location,
		FloatOut *	Out
	)

Decide on the output order

Definition at line 92 of file copy_color_spinor_mg.cuh.

References errorQuda, out, QUDA_FLOAT2_FIELD_ORDER, and QUDA_SPACE_SPIN_COLOR_FIELD_ORDER.

genericCopyColorSpinor() [3/6]

template<typename FloatOut , typename FloatIn , int Ns, int Nc>

void quda::genericCopyColorSpinor	(	ColorSpinorField &	out,
		const ColorSpinorField &	in,
		QudaFieldLocation	location,
		FloatOut *	Out,
		FloatIn *	In
	)

Decide on the input order

Definition at line 111 of file copy_color_spinor_mg.cuh.

References errorQuda, in, out, QUDA_FLOAT2_FIELD_ORDER, and QUDA_SPACE_SPIN_COLOR_FIELD_ORDER.

genericCopyColorSpinor() [4/6]

template<typename FloatOut , typename FloatIn , int Ns, int Nc, typename Out , typename In >

void quda::genericCopyColorSpinor	(	Out &	outOrder,
		const In &	inOrder,
		const ColorSpinorField &	out,
		const ColorSpinorField &	in,
		QudaFieldLocation	location
	)

Decide whether we are changing basis or not

Definition at line 268 of file copy_color_spinor.cuh.

References arg(), copy(), in, and out.

Here is the call graph for this function:

genericCopyColorSpinor() [5/6]

template<typename FloatOut , typename FloatIn , int Ns, int Nc, typename InOrder >

void quda::genericCopyColorSpinor	(	InOrder &	inOrder,
		ColorSpinorField &	out,
		const ColorSpinorField &	in,
		QudaFieldLocation	location,
		FloatOut *	Out,
		float *	outNorm
	)

Decide on the output order

Definition at line 280 of file copy_color_spinor.cuh.

References errorQuda, in, out, QUDA_FLOAT2_FIELD_ORDER, QUDA_PADDED_SPACE_SPIN_COLOR_FIELD_ORDER, QUDA_QDPJIT_FIELD_ORDER, QUDA_SPACE_COLOR_SPIN_FIELD_ORDER, and QUDA_SPACE_SPIN_COLOR_FIELD_ORDER.

genericCopyColorSpinor() [6/6]

template<typename FloatOut , typename FloatIn , int Ns, int Nc>

void quda::genericCopyColorSpinor	(	ColorSpinorField &	out,
		const ColorSpinorField &	in,
		QudaFieldLocation	location,
		FloatOut *	Out,
		FloatIn *	In,
		float *	outNorm,
		float *	inNorm
	)

Decide on the input order

Definition at line 331 of file copy_color_spinor.cuh.

References errorQuda, in, out, QUDA_FLOAT2_FIELD_ORDER, QUDA_PADDED_SPACE_SPIN_COLOR_FIELD_ORDER, QUDA_QDPJIT_FIELD_ORDER, QUDA_SPACE_COLOR_SPIN_FIELD_ORDER, and QUDA_SPACE_SPIN_COLOR_FIELD_ORDER.

GenericPackGhost()

template<typename Float , int Ns, int Ms, int Nc, int Mc, int nDim, typename Arg >

void quda::GenericPackGhost

(

Arg &

arg

)

Definition at line 81 of file color_spinor_pack.cu.

References arg(), fused_exterior_ndeg_tm_dslash_cuda_gen::i, and parity.

Here is the call graph for this function:

genericPackGhost()

void quda::genericPackGhost ( void **  ghost, const ColorSpinorField &  a, QudaParity  parity, int  nFace, int  dagger, MemoryLocation *  destination = nullptr  )

inline

Generic ghost packing routine.

Parameters

[out]	ghost	Array of packed ghosts with array ordering [2*dim+dir]
[in]	a	Input field that is being packed
[in]	parity	Which parity are we packing
[in]	dagger	Is for a dagger operator (presently ignored)
	[in[	location Array specifiying the memory location of each resulting ghost [2*dim+dir]

Definition at line 163 of file color_spinor_pack.cu.

References a, quda::GenericPackGhostLauncher< Float, Ns, Ms, Nc, Mc, Arg >::apply(), arg(), deg_tm_dslash_cuda_gen::dagger, and parity.

Referenced by quda::cudaColorSpinorField::exchangeGhost(), and quda::cpuColorSpinorField::packGhost().

Here is the call graph for this function:

Here is the caller graph for this function:

GenericPackGhostKernel()

template<typename Float , int Ns, int Ms, int Nc, int Mc, int nDim, typename Arg >

__global__ void quda::GenericPackGhostKernel

(

Arg

arg

)

Definition at line 93 of file color_spinor_pack.cu.

References arg(), blockDim, and parity.

Here is the call graph for this function:

genericPrintVector()

void quda::genericPrintVector	(	cpuColorSpinorField &	a,
		unsigned int	x
	)

Definition at line 285 of file color_spinor_util.cu.

References a, errorQuda, print_vector(), and x.

Referenced by quda::cpuColorSpinorField::PrintVector().

Here is the call graph for this function:

Here is the caller graph for this function:

genericSource()

void quda::genericSource	(	cpuColorSpinorField &	a,
		QudaSourceType	sourceType,
		int	x,
		int	s,
		int	c
	)

Definition at line 76 of file color_spinor_util.cu.

References a, c, constant(), errorQuda, point(), QUDA_CONSTANT_SOURCE, QUDA_POINT_SOURCE, QUDA_RANDOM_SOURCE, QUDA_SINUSOIDAL_SOURCE, random(), s, sin(), and x.

Referenced by quda::cpuColorSpinorField::Source().

Here is the call graph for this function:

Here is the caller graph for this function:

genGauss()

template<typename InOrder , typename FloatIn >

__device__ __host__ void quda::genGauss	(	InOrder &	inOrder,
		cuRNGState &	localState,
		int	x,
		int	s,
		int	c
	)

Definition at line 23 of file spinor_gauss.cu.

References c, cos(), log(), s, sin(), sqrt(), and x.

Here is the call graph for this function:

GetBlockDim()

dim3 quda::GetBlockDim	(	size_t	threads,
		size_t	size
	)

Definition at line 18 of file random.cu.

References BLOCKSDIVUP, and size.

Referenced by launch_kernel_random().

Here is the caller graph for this function:

getCoords()

template<typename I >

static __device__ __host__ void quda::getCoords ( int  x[], int  cb_index, const I  X[], int  parity  )

inlinestatic

Compute the 4-d spatial index from the checkerboarded 1-d index at parity parity

Parameters

x	Computed spatial index
cb_index	1-d checkerboarded index
X	Full lattice dimensions
parity	Site parity

Definition at line 129 of file index_helper.cuh.

References parity, X, x, za, and zb.

Referenced by applyLaplace(), completeKSForceCore(), computeCoarseClover(), computeNeighborSum(), computeOvrImpSTOUTStep(), computeStapleRectangle(), computeUV(), computeVUV(), computeYhat(), quda::colorspinor::PaddedSpaceSpinorColorOrder< Float, Ns, Nc >::getPaddedIndex(), quda::gauge::TIFRPaddedOrder< Float, length >::getPaddedIndex(), kernel_random(), packGhost(), and sin().

Here is the caller graph for this function:

getCoords5()

template<typename I >

static __device__ __host__ void quda::getCoords5 ( int  x[5], int  cb_index, const I  X[5], int  parity, QudaDWFPCType  pc_type  )

inlinestatic

Compute the 4-d spatial index from the checkerboarded 1-d index at parity parity

Parameters

x	Computed spatial index
cb_index	1-d checkerboarded index
X	Full lattice dimensions
parity	Site parity

Definition at line 181 of file index_helper.cuh.

References parity, QUDA_5D_PC, X, x, za, and zb.

Referenced by packGhost().

Here is the caller graph for this function:

getCoordsExtended()

template<typename I , typename J >

static __device__ __host__ void quda::getCoordsExtended ( I  x[], int  cb_index, const J  X[], int  parity, const int  R[]  )

inlinestatic

Compute the 4-d spatial index from the checkerboarded 1-d index at parity parity

Parameters

x	Computed spatial index
cb_index	1-d checkerboarded index
X	Full lattice dimensions
parity	Site parity

Definition at line 154 of file index_helper.cuh.

References d, parity, R, X, x, za, and zb.

getDeterminant()

template<template< typename, int > class Mat, class T >

__device__ __host__ T quda::getDeterminant ( const Mat< T, 3 > &  a )

inline

Definition at line 312 of file quda_matrix.h.

References a.

Referenced by computeLinkInverse(), computeMatrixInverse(), exponentiate_iQ(), quda::gauge::Reconstruct< 13, Float >::getPhase(), quda::gauge::Reconstruct< 9, Float >::getPhase(), and polarSu3().

Here is the caller graph for this function:

getDslashLaunch()

bool quda::getDslashLaunch

(

)

getIndexFull()

template<typename I >

static __device__ __host__ int quda::getIndexFull ( int  cb_index, const I  X[4], int  parity  )

inlinestatic

Compute the 1-d global index from 1-d checkerboard index and parity. This should never be used to index into QUDA fields due to the potential of padding between even and odd regions.

Parameters

cb_index	1-d checkerboard index
X	lattice dimensions
parity	Site parity

Definition at line 211 of file index_helper.cuh.

References parity, X, za, and zb.

getKernelPackT()

bool quda::getKernelPackT

(

)

Returns

Whether the T dimension is kernel packed or not

Definition at line 61 of file dslash_quda.cu.

References kernelPackT.

Referenced by anonymous_namespace{dslash_policy.cuh}::DslashPolicyTune::apply(), anonymous_namespace{dslash_policy.cuh}::DslashPolicyTune::DslashPolicyTune(), quda::cudaColorSpinorField::exchangeGhost(), anonymous_namespace{dslash_policy.cuh}::issueGather(), anonymous_namespace{dslash_policy.cuh}::issuePack(), anonymous_namespace{dslash_policy.cuh}::DslashBasic::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashPthreads::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashFusedExterior::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashGDRRecv::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashFusedGDRRecv::operator()(), quda::cudaColorSpinorField::sendGhost(), quda::cudaColorSpinorField::sendStart(), and DslashCuda::setParam().

Here is the caller graph for this function:

getLinkDeterminant()

double2 quda::getLinkDeterminant

(

cudaGaugeField &

data

)

Calculate the Determinant.

Parameters

[in]

data

Gauge field

Returns

double2 complex Determinant value

Definition at line 193 of file pgauge_det_trace.cu.

References errorQuda, quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, and QUDA_SINGLE_PRECISION.

Referenced by GaugeAlgTest::TearDown(), and TEST_F().

Here is the call graph for this function:

Here is the caller graph for this function:

getLinkTrace()

double2 quda::getLinkTrace

(

cudaGaugeField &

data

)

Calculate the Trace.

Parameters

[in]

data

Gauge field

Returns

double2 complex trace value

Definition at line 214 of file pgauge_det_trace.cu.

References errorQuda, quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, and QUDA_SINGLE_PRECISION.

Referenced by GaugeAlgTest::TearDown().

Here is the call graph for this function:

Here is the caller graph for this function:

getRealTraceUVdagger()

template<class T >

__device__ __host__ double quda::getRealTraceUVdagger ( const Matrix< T, 3 > &  a, const Matrix< T, 3 > &  b  )

inline

Definition at line 1021 of file quda_matrix.h.

References a, b, double, and sum().

Here is the call graph for this function:

getSubTraceUnit()

template<class T >

__device__ __host__ Matrix<T,3> quda::getSubTraceUnit ( const Matrix< T, 3 > &  a )

inline

Definition at line 1005 of file quda_matrix.h.

References a.

getTrace()

template<class T >

__device__ __host__ T quda::getTrace ( const Matrix< T, 3 > &  a )

inline

Definition at line 305 of file quda_matrix.h.

References a.

Referenced by completeKSForceCore(), computeOvrImpSTOUTStep(), and exponentiate_iQ().

Here is the caller graph for this function:

getTuneCache()

const map & quda::getTuneCache

(

)

Definition at line 110 of file tune.cpp.

References tunecache.

Referenced by quda::DslashCoarsePolicyTune::DslashCoarsePolicyTune(), anonymous_namespace{dslash_policy.cuh}::DslashPolicyTune::DslashPolicyTune(), and quda::blas::TileSizeTune< ReducerDiagonal, writeDiagonal, ReducerOffDiagonal, writeOffDiagonal >::TileSizeTune().

Here is the caller graph for this function:

ghostFaceIndex()

template<int dir, typename I >

__device__ __host__ int quda::ghostFaceIndex ( const int  x[], const I  X[], int  dim, int  nFace  )

inline

Compute the checkerboarded index into the ghost field corresponding to full (local) site index x[]

Parameters

x	local site
X	local lattice dimensions
dim	dimension
depth	of ghost

Definition at line 230 of file index_helper.cuh.

References dim, index(), X, and x.

Here is the call graph for this function:

host_allocated_peak()

long quda::host_allocated_peak

(

)

Returns

peak host memory allocated

Definition at line 63 of file malloc.cpp.

References HOST, and max_total_bytes.

host_free_()

void quda::host_free_	(	const char *	func,
		const char *	file,
		int	line,
		void *	ptr
	)

Free host memory allocated with safe_malloc(), pinned_malloc(), or mapped_malloc(). This function should only be called via the host_free() macro, defined in malloc_quda.h

Definition at line 340 of file malloc.cpp.

References alloc, count, err, errorQuda, free(), func, HOST, MAPPED, PINNED, print_trace(), printfQuda, ptr, and track_free().

Referenced by quda::pool::pinned_free_().

Here is the call graph for this function:

Here is the caller graph for this function:

i32toa()

void quda::i32toa ( char *  buffer, int32_t  value  )

inline

Definition at line 117 of file uint_to_char.h.

References u32toa(), and value.

Here is the call graph for this function:

i64toa()

void quda::i64toa ( char *  buffer, int64_t  value  )

inline

Definition at line 284 of file uint_to_char.h.

References u64toa(), and value.

Here is the call graph for this function:

improvedStaggeredDslashCuda()

void quda::improvedStaggeredDslashCuda	(	cudaColorSpinorField *	out,
		const cudaGaugeField &	fatGauge,
		const cudaGaugeField &	longGauge,
		const cudaColorSpinorField *	in,
		const int	parity,
		const int	dagger,
		const cudaColorSpinorField *	x,
		const double &	k,
		const int *	commDim,
		TimeProfile &	profile
	)

Definition at line 272 of file dslash_improved_staggered.cu.

References deg_tm_dslash_cuda_gen::dagger, deg_tm_dslash_cuda_gen::dslash, errorQuda, fused_exterior_ndeg_tm_dslash_cuda_gen::i, in, out, parity, QUDA_DOUBLE_PRECISION, QUDA_HALF_PRECISION, QUDA_MAX_DIM, QUDA_SINGLE_PRECISION, and x.

Referenced by quda::DiracImprovedStaggered::Dslash(), and quda::DiracImprovedStaggered::DslashXpay().

Here is the caller graph for this function:

IndexBlock()

template<int NCOLORS>

static __host__ __device__ void quda::IndexBlock ( int  block, int &  p, int &  q  )

inlinestatic

Retrieve the SU(N) indices for the current block number

Parameters

[in]	block,current	block number, from 0 to (NCOLORS * (NCOLORS - 1) / 2)
[out]	p,row	index pointing to the SU(N) matrix
[out]	q,column	index pointing to the SU(N) matrix

Definition at line 36 of file gauge_fix_ovr_hit_devf.cuh.

References deg_tm_dslash_cuda_gen::block(), index(), and p.

Here is the call graph for this function:

InitGaugeField() [1/2]

void quda::InitGaugeField

(

cudaGaugeField &

data

)

Perform a cold start to the gauge field, identity SU(3) matrix, also fills the ghost links in multi-GPU case (no need to exchange data)

Parameters

[in,out]

data

Gauge field

Referenced by main(), and GaugeAlgTest::SetUp().

Here is the caller graph for this function:

InitGaugeField() [2/2]

void quda::InitGaugeField	(	cudaGaugeField &	data,
		RNG &	rngstate
	)

Perform a hot start to the gauge field, random SU(3) matrix, followed by reunitarization, also exchange borders links in multi-GPU case.

Parameters

[in,out]	data	Gauge field
[in,out]	rngstate	state of the CURAND random number generator

Definition at line 459 of file pgauge_init.cu.

References errorQuda, quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, and QUDA_SINGLE_PRECISION.

Here is the call graph for this function:

injector()

template<typename Float , int length, int dim, typename Arg >

__device__ __host__ void quda::injector	(	Arg &	arg,
		int	dir,
		int	a,
		int	b,
		int	c,
		int	d,
		int	g,
		int	parity
	)

Definition at line 73 of file extract_gauge_ghost_extended.cu.

References a, arg(), b, c, d, dim, length, and parity.

Here is the call graph for this function:

isUnitary() [1/2]

bool quda::isUnitary	(	const cpuGaugeField &	field,
		double	max_error
	)

Referenced by ProjectSU3kernel().

Here is the caller graph for this function:

isUnitary() [2/2]

template<class Cmplx >

__device__ __host__ bool quda::isUnitary	(	const Matrix< Cmplx, 3 > &	matrix,
		double	max_error
	)

Definition at line 1054 of file quda_matrix.h.

References conj(), fabs(), fused_exterior_ndeg_tm_dslash_cuda_gen::i, x, and y.

Here is the call graph for this function:

kernel_random() [1/2]

__global__ void quda::kernel_random	(	cuRNGState *	state,
		int	seed,
		int	rng_size,
		int	node_offset
	)

CUDA kernel to initialize CURAND RNG states.

Parameters

state	CURAND RNG state array
seed	initial seed for RNG
rng_size	size of the CURAND RNG state array
node_offset	this parameter is used to skip ahead the index in the sequence, usefull for multigpu.

Definition at line 45 of file random.cu.

References blockDim.

kernel_random() [2/2]

__global__ void quda::kernel_random	(	cuRNGState *	state,
		int	seed,
		int	rng_size,
		int	node_offset,
		rngArg	arg
	)

Definition at line 61 of file random.cu.

References arg(), blockDim, getCoords(), fused_exterior_ndeg_tm_dslash_cuda_gen::i, and x.

Here is the call graph for this function:

laplace()

template<typename Float , int nDim, int nColor, typename Arg >

__device__ __host__ void quda::laplace ( Arg &  arg, int  x_cb, int  parity  )

inline

Definition at line 113 of file laplace.cu.

References arg(), out, parity, and x.

Referenced by ApplyLaplace(), quda::GaugeLaplace::operator=(), and quda::GaugeLaplacePC::operator=().

Here is the call graph for this function:

Here is the caller graph for this function:

laplaceCPU()

template<typename Float , int nDim, int nColor, typename Arg >

void quda::laplaceCPU

(

Arg

arg

)

Definition at line 129 of file laplace.cu.

References arg(), for(), and parity.

Here is the call graph for this function:

laplaceGPU()

template<typename Float , int nDim, int nColor, typename Arg >

__global__ void quda::laplaceGPU

(

Arg

arg

)

Definition at line 145 of file laplace.cu.

References arg(), blockDim, if(), and parity.

Here is the call graph for this function:

launch_kernel_random()

void quda::launch_kernel_random	(	cuRNGState *	state,
		int	seed,
		int	rng_size,
		int	node_offset,
		int	X[4]
	)

Call CUDA kernel to initialize CURAND RNG states.

Parameters

state	CURAND RNG state array
seed	initial seed for RNG
rng_size	size of the CURAND RNG state array
node_offset	this parameter is used to skip ahead the index in the sequence, usefull for multigpu.

Definition at line 85 of file random.cu.

References arg(), comm_coord(), comm_dim(), GetBlockDim(), fused_exterior_ndeg_tm_dslash_cuda_gen::i, qudaDeviceSynchronize(), and X.

Referenced by quda::RNG::Init().

Here is the call graph for this function:

Here is the caller graph for this function:

linkIndex() [1/2]

template<typename I >

static __device__ __host__ int quda::linkIndex ( const int  x[], const I  X[4]  )

inlinestatic

Compute the checkerboard 1-d index from the 4-d coordinate x[]

Returns

1-d checkerboard index

Parameters

x	4-d lattice index
X	Full lattice dimensions

Definition at line 46 of file index_helper.cuh.

References idx, X, and x.

Referenced by quda::colorspinor::PaddedSpaceSpinorColorOrder< Float, Ns, Nc >::getPaddedIndex(), and quda::gauge::TIFRPaddedOrder< Float, length >::getPaddedIndex().

Here is the caller graph for this function:

linkIndex() [2/2]

template<typename I >

static __device__ __host__ int quda::linkIndex ( int  y[], const int  x[], const I  X[4]  )

inlinestatic

Compute the checkerboard 1-d index from the 4-d coordinate x[]

Returns

1-d checkerboard index

Parameters

y	copy of 4-d lattice index
x	4-d lattice index
X	Full lattice dimensions

Definition at line 60 of file index_helper.cuh.

References idx, X, x, and y.

linkIndexM1()

template<typename I >

static __device__ __host__ int quda::linkIndexM1 ( const int  x[], const I  X[4], const int  mu  )

inlinestatic

Compute the checkerboard 1-d index from the 4-d coordinate x[] -1 in the mu direction

Returns

1-d checkerboard index

Parameters

x	4-d lattice index
X	Full lattice dimensions
mu	direction in which to subtract 1

Definition at line 75 of file index_helper.cuh.

References fused_exterior_ndeg_tm_dslash_cuda_gen::i, idx, mu, X, x, and y.

Referenced by applyLaplace(), computeNeighborSum(), and computeYhat().

Here is the caller graph for this function:

linkIndexP1()

template<typename I >

static __device__ __host__ int quda::linkIndexP1 ( const int  x[], const I  X[4], const int  mu  )

inlinestatic

Compute the checkerboard 1-d index from the 4-d coordinate x[] +1 in the mu direction

Returns

1-d checkerboard index

Parameters

x	4-d lattice index
X	Full lattice dimensions
mu	direction in which to add 1

Definition at line 111 of file index_helper.cuh.

References fused_exterior_ndeg_tm_dslash_cuda_gen::i, idx, mu, X, x, and y.

Referenced by applyLaplace(), computeNeighborSum(), and computeUV().

Here is the caller graph for this function:

linkIndexShift() [1/2]

template<typename I , typename J , typename K >

static __device__ __host__ int quda::linkIndexShift ( const I  x[], const J  dx[], const K  X[4]  )

inlinestatic

Compute the checkerboard 1-d index from the 4-d coordinate x[] + dx[]

Returns

1-d checkerboard index

Parameters

x	4-d lattice index
dx	4-d shift index
X	Full lattice dimensions

Definition at line 13 of file index_helper.cuh.

References fused_exterior_ndeg_tm_dslash_cuda_gen::i, idx, X, x, and y.

Referenced by completeKSForceCore(), computeOvrImpSTOUTStep(), and computeStapleRectangle().

Here is the caller graph for this function:

linkIndexShift() [2/2]

template<typename I , typename J , typename K >

static __device__ __host__ int quda::linkIndexShift ( I  y[], const I  x[], const J  dx[], const K  X[4]  )

inlinestatic

Compute the checkerboard 1-d index from the 4-d coordinate x[] + dx[]

Returns

1-d checkerboard index

Parameters

y	new 4-d lattice index
x	original 4-d lattice index
dx	4-d shift index
X	Full lattice dimensions

Definition at line 31 of file index_helper.cuh.

References fused_exterior_ndeg_tm_dslash_cuda_gen::i, idx, X, x, and y.

linkNormalIndexP1()

template<typename I >

static __device__ __host__ int quda::linkNormalIndexP1 ( const int  x[], const I  X[4], const int  mu  )

inlinestatic

Compute the full 1-d index from the 4-d coordinate x[] +1 in the mu direction

Returns

1-d checkerboard index

Parameters

x	4-d lattice index
X	Full lattice dimensions
mu	direction in which to add 1

Definition at line 93 of file index_helper.cuh.

References fused_exterior_ndeg_tm_dslash_cuda_gen::i, idx, mu, X, x, and y.

load_global_float4()

__device__ void quda::load_global_float4 ( float4 &  a, const float4 *  addr  )

inline

Definition at line 32 of file inline_ptx.h.

References __PTR, a, w, x, y, and z.

load_streaming_double2()

__device__ void quda::load_streaming_double2 ( double2 &  a, const double2 *  addr  )

inline

Definition at line 18 of file inline_ptx.h.

References __PTR, a, x, and y.

load_streaming_float4()

__device__ void quda::load_streaming_float4 ( float4 &  a, const float4 *  addr  )

inline

Definition at line 25 of file inline_ptx.h.

References __PTR, a, w, x, y, and z.

loadLinkVariableFromArray() [1/2]

template<class T , class U >

__device__ void quda::loadLinkVariableFromArray ( const T *const  array, const int  dir, const int  idx, const int  stride, Matrix< U, 3 > *  link  )

inline

Definition at line 747 of file quda_matrix.h.

References array, quda::Matrix< T, N >::data, fused_exterior_ndeg_tm_dslash_cuda_gen::i, and idx.

loadLinkVariableFromArray() [2/2]

__device__ void quda::loadLinkVariableFromArray ( const float2 *const  array, const int  dir, const int  idx, const int  stride, Matrix< complex< double >, 3 > *  link  )

inline

Definition at line 769 of file quda_matrix.h.

References array, fused_exterior_ndeg_tm_dslash_cuda_gen::i, and idx.

loadMatrixFromArray()

template<class T , class U , int N>

__device__ void quda::loadMatrixFromArray ( const T *const  array, const int  idx, const int  stride, Matrix< U, N > *  mat  )

inline

Definition at line 759 of file quda_matrix.h.

References array, fused_exterior_ndeg_tm_dslash_cuda_gen::i, idx, and mat().

Here is the call graph for this function:

loadMomentumFromArray()

template<class T >

__device__ void quda::loadMomentumFromArray ( const T *const  array, const int  dir, const int  idx, const int  stride, Matrix< T, 3 > *  mom  )

inline

Definition at line 845 of file quda_matrix.h.

References array, quda::Matrix< T, N >::data, and idx.

loadTuneCache()

void quda::loadTuneCache

(

)

Definition at line 302 of file tune.cpp.

References broadcastTuneCache(), comm_rank(), deserializeTuneCache(), errorQuda, getenv(), getline(), getTuning(), getVerbosity(), gitversion, initial_cache_size, printfQuda, quda_hash, QUDA_SUMMARIZE, QUDA_TUNE_NO, quda_version, resource_path, tunecache, and warningQuda.

Referenced by initQudaMemory().

Here is the call graph for this function:

Here is the caller graph for this function:

Location_() [1/2]

QudaFieldLocation quda::Location_ ( const char *  func, const char *  file, int  line, const LatticeField &  a, const LatticeField &  b  )

inline

Helper function for determining if the location of the fields is the same.

Parameters

[in]	a	Input field
[in]	b	Input field

Returns

If location is unique return the location

Definition at line 539 of file lattice_field.h.

References a, b, errorQuda, func, and QUDA_INVALID_FIELD_LOCATION.

Referenced by Location_().

Here is the caller graph for this function:

Location_() [2/2]

template<typename... Args>

QudaFieldLocation quda::Location_ ( const char *  func, const char *  file, int  line, const LatticeField &  a, const LatticeField &  b, const Args &...  args  )

inline

Helper function for determining if the location of the fields is the same.

Parameters

[in]	a	Input field
[in]	b	Input field
[in]	args	List of additional fields to check location on

Returns

If location is unique return the location

Definition at line 556 of file lattice_field.h.

References a, args, b, func, and Location_().

Here is the call graph for this function:

log() [1/3]

template<typename ValueType >

__host__ __device__ ValueType quda::log ( ValueType  x )

inline

Definition at line 90 of file complex_quda.h.

References log(), and x.

Referenced by acosh(), asinh(), atanh(), genGauss(), log(), log10(), and pow().

Here is the call graph for this function:

Here is the caller graph for this function:

log() [2/3]

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::log ( const complex< ValueType > &  z )

inline

Definition at line 966 of file complex_quda.h.

References abs(), arg(), log(), and z.

Here is the call graph for this function:

log() [3/3]

template<>

__host__ __device__ complex<float> quda::log ( const complex< float > &  z )

inline

Definition at line 972 of file complex_quda.h.

References abs(), arg(), logf(), and z.

Referenced by log().

Here is the call graph for this function:

Here is the caller graph for this function:

log10() [1/2]

template<typename ValueType >

__host__ __device__ ValueType quda::log10 ( ValueType  x )

inline

Definition at line 95 of file complex_quda.h.

References log10(), and x.

Here is the call graph for this function:

log10() [2/2]

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::log10 ( const complex< ValueType > &  z )

inline

Definition at line 979 of file complex_quda.h.

References log(), and z.

Referenced by log10().

Here is the call graph for this function:

Here is the caller graph for this function:

make_Complex() [1/2]

complex<double> quda::make_Complex ( const double2 &  a )

inline

Definition at line 278 of file float_vector.h.

References a.

Referenced by genericBlas(), genericMultiBlas(), and genericReduce().

Here is the caller graph for this function:

make_Complex() [2/2]

complex<float> quda::make_Complex ( const float2 &  a )

inline

Definition at line 279 of file float_vector.h.

References a.

make_Float2() [1/9]

template<typename Float2 , typename Complex >

Float2 quda::make_Float2 ( const Complex &  a )

inline

Definition at line 257 of file float_vector.h.

make_Float2() [2/9]

template<>

double2 quda::make_Float2 ( const complex< double > &  a )

inline

Definition at line 260 of file float_vector.h.

References a.

make_Float2() [3/9]

template<>

double2 quda::make_Float2 ( const complex< float > &  a )

inline

Definition at line 262 of file float_vector.h.

References a.

make_Float2() [4/9]

template<>

float2 quda::make_Float2 ( const complex< double > &  a )

inline

Definition at line 264 of file float_vector.h.

References a.

make_Float2() [5/9]

template<>

float2 quda::make_Float2 ( const complex< float > &  a )

inline

Definition at line 266 of file float_vector.h.

References a.

make_Float2() [6/9]

template<>

double2 quda::make_Float2 ( const std::complex< double > &  a )

inline

Definition at line 269 of file float_vector.h.

References a.

make_Float2() [7/9]

template<>

double2 quda::make_Float2 ( const std::complex< float > &  a )

inline

Definition at line 271 of file float_vector.h.

References a.

make_Float2() [8/9]

template<>

float2 quda::make_Float2 ( const std::complex< double > &  a )

inline

Definition at line 273 of file float_vector.h.

References a.

make_Float2() [9/9]

template<>

float2 quda::make_Float2 ( const std::complex< float > &  a )

inline

Definition at line 275 of file float_vector.h.

References a.

make_FloatN() [1/4]

__forceinline__ __host__ __device__ float2 quda::make_FloatN

(

const double2 &

a

)

Definition at line 222 of file float_vector.h.

References a.

make_FloatN() [2/4]

__forceinline__ __host__ __device__ float4 quda::make_FloatN

(

const double4 &

a

)

Definition at line 226 of file float_vector.h.

References a.

make_FloatN() [3/4]

__forceinline__ __host__ __device__ double2 quda::make_FloatN

(

const float2 &

a

)

Definition at line 230 of file float_vector.h.

References a.

make_FloatN() [4/4]

__forceinline__ __host__ __device__ double4 quda::make_FloatN

(

const float4 &

a

)

Definition at line 234 of file float_vector.h.

References a.

make_shortN() [1/4]

__forceinline__ __host__ __device__ short4 quda::make_shortN

(

const float4 &

a

)

Definition at line 238 of file float_vector.h.

References a.

make_shortN() [2/4]

__forceinline__ __host__ __device__ short2 quda::make_shortN

(

const float2 &

a

)

Definition at line 242 of file float_vector.h.

References a.

make_shortN() [3/4]

__forceinline__ __host__ __device__ short4 quda::make_shortN

(

const double4 &

a

)

Definition at line 246 of file float_vector.h.

References a.

make_shortN() [4/4]

__forceinline__ __host__ __device__ short2 quda::make_shortN

(

const double2 &

a

)

Definition at line 250 of file float_vector.h.

References a.

makeAntiHerm()

template<typename Complex , int N>

__device__ __host__ void quda::makeAntiHerm ( Matrix< Complex, N > &  m )

inline

Definition at line 636 of file quda_matrix.h.

References conj(), and fused_exterior_ndeg_tm_dslash_cuda_gen::i.

Here is the call graph for this function:

mapped_allocated_peak()

long quda::mapped_allocated_peak

(

)

Returns

peak mapped memory allocated

Definition at line 61 of file malloc.cpp.

References MAPPED, and max_total_bytes.

mapped_malloc_()

void * quda::mapped_malloc_	(	const char *	func,
		const char *	file,
		int	line,
		size_t	size
	)

Allocate page-locked ("pinned") host memory, and map it into the GPU address space. This function should only be called via the mapped_malloc() macro, defined in malloc_quda.h

Definition at line 269 of file malloc.cpp.

References a, aligned_malloc(), err, errorQuda, func, MAPPED, memset(), printfQuda, ptr, size, and track_malloc().

Here is the call graph for this function:

massRescale()

void quda::massRescale	(	cudaColorSpinorField &	b,
		QudaInvertParam &	param
	)

Definition at line 1532 of file interface_quda.cpp.

References quda::blas::ax(), b, errorQuda, getVerbosity(), fused_exterior_ndeg_tm_dslash_cuda_gen::i, if(), kappa, kappa5, quda::blas::norm2(), param, pow(), printfQuda, QUDA_ASQTAD_DSLASH, QUDA_ASYMMETRIC_MASS_NORMALIZATION, QUDA_DEBUG_VERBOSE, QUDA_DOMAIN_WALL_4D_DSLASH, QUDA_DOMAIN_WALL_DSLASH, QUDA_KAPPA_NORMALIZATION, QUDA_MASS_NORMALIZATION, QUDA_MAT_SOLUTION, QUDA_MATDAG_MAT_SOLUTION, QUDA_MATPC_SOLUTION, QUDA_MATPCDAG_MATPC_SOLUTION, QUDA_MOBIUS_DWF_DSLASH, QUDA_STAGGERED_DSLASH, and unscaled_shifts.

Referenced by invertMultiShiftQuda(), invertMultiSrcQuda(), and invertQuda().

Here is the call graph for this function:

Here is the caller graph for this function:

max_fabs() [1/4]

__forceinline__ __host__ __device__ float quda::max_fabs

(

const float4 &

c

)

Definition at line 198 of file float_vector.h.

References a, b, c, fabsf(), and fmaxf().

Referenced by store_norm().

Here is the call graph for this function:

Here is the caller graph for this function:

max_fabs() [2/4]

__forceinline__ __host__ __device__ float quda::max_fabs

(

const float2 &

b

)

Definition at line 204 of file float_vector.h.

References b, fabsf(), and fmaxf().

Here is the call graph for this function:

max_fabs() [3/4]

__forceinline__ __host__ __device__ double quda::max_fabs

(

const double4 &

c

)

Definition at line 208 of file float_vector.h.

References a, b, c, fabs(), and fmax().

Here is the call graph for this function:

max_fabs() [4/4]

__forceinline__ __host__ __device__ double quda::max_fabs

(

const double2 &

b

)

Definition at line 214 of file float_vector.h.

References b, fabs(), and fmax().

Here is the call graph for this function:

maxGauge() [1/2]

template<typename Float , int Nc, typename Order >

double quda::maxGauge	(	const Order	order,
		int	volume,
		int	nDim
	)

Generic CPU function find the gauge maximum

Definition at line 11 of file max_gauge.cu.

References abs(), d, fused_exterior_ndeg_tm_dslash_cuda_gen::i, parity, and x.

Here is the call graph for this function:

maxGauge() [2/2]

double quda::maxGauge

(

const GaugeField &

u

)

This function is used to calculate the maximum absolute value of a gauge field array. Defined in max_gauge.cu.

Parameters

[in]

u

The gauge field from which we want to compute the max

Definition at line 31 of file max_gauge.cu.

References errorQuda, quda::GaugeField::Gauge_p(), quda::GaugeField::Ncolor(), quda::GaugeField::Order(), QUDA_BQCD_GAUGE_ORDER, QUDA_CPS_WILSON_GAUGE_ORDER, QUDA_MILC_GAUGE_ORDER, QUDA_QDP_GAUGE_ORDER, QUDA_TIFR_GAUGE_ORDER, and reduceMaxDouble().

Referenced by quda::cpuGaugeField::cpuGaugeField().

Here is the call graph for this function:

Here is the caller graph for this function:

MDWFDslashCuda()

void quda::MDWFDslashCuda	(	cudaColorSpinorField *	out,
		const cudaGaugeField &	gauge,
		const cudaColorSpinorField *	in,
		const int	parity,
		const int	dagger,
		const cudaColorSpinorField *	x,
		const double &	m_f,
		const double &	k,
		const double *	b5,
		const double *	c_5,
		const double &	m5,
		const int *	commDim,
		const int	DS_type,
		TimeProfile &	profile
	)

Definition at line 273 of file dslash_mobius.cu.

References deg_tm_dslash_cuda_gen::dagger, deg_tm_dslash_cuda_gen::dslash, errorQuda, fused_exterior_ndeg_tm_dslash_cuda_gen::i, in, m5, out, parity, QUDA_DOUBLE_PRECISION, QUDA_HALF_PRECISION, QUDA_MAX_DIM, QUDA_SINGLE_PRECISION, and x.

Referenced by quda::DiracMobius::Dslash4(), quda::DiracMobius::Dslash4pre(), quda::DiracMobius::Dslash4preXpay(), quda::DiracMobius::Dslash4Xpay(), quda::DiracMobius::Dslash5(), quda::DiracMobiusPC::Dslash5inv(), quda::DiracMobiusPC::Dslash5invXpay(), and quda::DiracMobius::Dslash5Xpay().

Here is the caller graph for this function:

Monte()

void quda::Monte	(	cudaGaugeField &	data,
		RNG &	rngstate,
		double	Beta,
		int	nhb,
		int	nover
	)

Perform heatbath and overrelaxation. Performs nhb heatbath steps followed by nover overrelaxation steps.

Parameters

[in,out]	data	Gauge field
[in,out]	rngstate	state of the CURAND random number generator
[in]	Beta	inverse of the gauge coupling, beta = 2 Nc / g_0^2
[in]	nhb	number of heatbath steps
[in]	nover	number of overrelaxation steps

Definition at line 857 of file pgauge_heatbath.cu.

References errorQuda, float, quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, and QUDA_SINGLE_PRECISION.

Referenced by main(), and GaugeAlgTest::SetUp().

Here is the call graph for this function:

Here is the caller graph for this function:

multiplyVUV()

template<bool from_coarse, typename Float , int dim, QudaDirection dir, int fineSpin, int fineColor, int coarseSpin, int coarseColor, typename Arg >

__device__ __host__ void quda::multiplyVUV ( complex< Float >  vuv[], Arg &  arg, int  parity, int  x_cb, int  ic_c  )

inline

Do a single (AV)^ * UV product, where for preconditioned clover, AV correspond to the clover inverse multiplied by the packed null space vectors, else AV is simply the packed null space vectors.

Parameters

[out]	vuv	Result array
[in,out]	arg	Arg storing the fields and parameters
[in]	Fine	grid parity we're working on
[in]	x_cb	Checkboarded x dimension

Definition at line 494 of file coarse_op.cuh.

References arg(), conj(), gamma(), fused_exterior_ndeg_tm_dslash_cuda_gen::i, parity, QUDA_BACKWARDS, and s.

Here is the call graph for this function:

ndegTwistedMassDslashCuda()

void quda::ndegTwistedMassDslashCuda	(	cudaColorSpinorField *	out,
		const cudaGaugeField &	gauge,
		const cudaColorSpinorField *	in,
		const int	parity,
		const int	dagger,
		const cudaColorSpinorField *	x,
		const QudaTwistDslashType	type,
		const double &	kappa,
		const double &	mu,
		const double &	epsilon,
		const double &	k,
		const int *	commDim,
		TimeProfile &	profile
	)

Definition at line 144 of file dslash_ndeg_twisted_mass.cu.

References deg_tm_dslash_cuda_gen::dagger, deg_tm_dslash_cuda_gen::dslash, errorQuda, fused_exterior_ndeg_tm_dslash_cuda_gen::i, in, kappa, mu, out, parity, QUDA_DOUBLE_PRECISION, QUDA_HALF_PRECISION, QUDA_SINGLE_PRECISION, and x.

Referenced by quda::DiracTwistedMass::NdegTwistedDslash(), and quda::DiracTwistedMass::NdegTwistedDslashXpay().

Here is the caller graph for this function:

neighborIndex()

template<IndexType idxType, typename Int >

__device__ __forceinline__ int quda::neighborIndex	(	const unsigned int &	cb_idx,
		const int(&)	shift[4],
		const bool(&)	partitioned[4],
		const unsigned int &	parity
	)

Definition at line 41 of file shift_quark_field.cu.

References coordsFromIndex(), full_idx, idx, parity, shift, t, x, y, and z.

Referenced by gaugeLink(), shiftColorSpinorFieldKernel(), and spinorNeighbor().

Here is the call graph for this function:

Here is the caller graph for this function:

norm()

template<typename ValueType >

__host__ __device__ ValueType quda::norm ( const complex< ValueType > &  z )

inline

Returns the magnitude of z squared.

Definition at line 896 of file complex_quda.h.

References z.

Referenced by construct_clover_field(), constructCloverField(), ErrorSU3(), quda::GMResDR::FlexArnoldiProcedure(), init(), main(), new_load_half(), new_save_half(), newCopyToFloat(), newCopyToHalf(), quda::gauge::FieldOrder< Float, nColor, nSpinCoarse, order, native_ghost >::norm2(), normalize(), old_load_half(), old_save_half(), oldCopyToFloat(), oldCopyToHalf(), quda::gauge::square< ReduceType, Float >::operator()(), quda::colorspinor::square< ReduceType, Float >::operator()(), quda::GMResDR::operator()(), operator/(), performWuppertalnStep(), quda::ArpackArgs< Float >::save(), store_norm(), and quda::gauge::Reconstruct< 8, Float >::Unpack().

Here is the caller graph for this function:

norm1() [1/2]

double quda::norm1	(	const CloverField &	u,
		bool	inverse = false
	)

This is a debugging function, where we cast a clover field into a spinor field so we can compute its L1 norm.

Parameters

a	The clover field that we want the norm of

Returns

The L1 norm of the gauge field

Definition at line 455 of file clover_field.cpp.

References a, b, colorSpinorParam(), quda::ColorSpinorField::Create(), and quda::blas::norm1().

Here is the call graph for this function:

norm1() [2/2]

double quda::norm1

(

const GaugeField &

u

)

This is a debugging function, where we cast a gauge field into a spinor field so we can compute its L1 norm.

Parameters

u	The gauge field that we want the norm of

Returns

The L1 norm of the gauge field

Definition at line 314 of file gauge_field.cpp.

References a, b, colorSpinorParam(), quda::ColorSpinorField::Create(), and quda::blas::norm1().

Here is the call graph for this function:

norm2() [1/2]

double quda::norm2	(	const CloverField &	a,
		bool	inverse = false
	)

This is a debugging function, where we cast a clover field into a spinor field so we can compute its L2 norm.

Parameters

a	The clover field that we want the norm of

Returns

The L2 norm squared of the gauge field

Definition at line 447 of file clover_field.cpp.

References a, b, colorSpinorParam(), quda::ColorSpinorField::Create(), and quda::blas::norm2().

Referenced by quda::GMResDR::FlexArnoldiProcedure(), quda::MG::generateNullVectors(), quda::Lanczos::operator()(), quda::Deflation::operator()(), quda::MG::operator()(), quda::PreconCG::operator()(), quda::SimpleBiCGstab::operator()(), quda::SD::operator()(), quda::IncEigCG::operator()(), quda::GMResDR::operator()(), quda::Deflation::reduce(), quda::Deflation::verify(), and quda::MG::verify().

Here is the call graph for this function:

Here is the caller graph for this function:

norm2() [2/2]

double quda::norm2

(

const GaugeField &

u

)

This is a debugging function, where we cast a gauge field into a spinor field so we can compute its L2 norm.

Parameters

u	The gauge field that we want the norm of

Returns

The L2 norm squared of the gauge field

Definition at line 306 of file gauge_field.cpp.

References a, b, colorSpinorParam(), quda::ColorSpinorField::Create(), and quda::blas::norm2().

Here is the call graph for this function:

operator!=() [1/3]

template<typename ValueType >

__host__ __device__ bool quda::operator!= ( const complex< ValueType > &  lhs, const complex< ValueType > &  rhs  )

inline

Definition at line 839 of file complex_quda.h.

Referenced by std::__1::__attribute(), and std::__1::__attribute__().

Here is the caller graph for this function:

operator!=() [2/3]

template<typename ValueType >

__host__ __device__ bool quda::operator!= ( const ValueType &  lhs, const complex< ValueType > &  rhs  )

inline

Definition at line 845 of file complex_quda.h.

operator!=() [3/3]

template<typename ValueType >

__host__ __device__ bool quda::operator!= ( const complex< ValueType > &  lhs, const ValueType &  rhs  )

inline

Definition at line 851 of file complex_quda.h.

operator*() [1/16]

__host__ __device__ float4 quda::operator* ( const float  a, const float4  x  )

inline

Definition at line 48 of file float_vector.h.

References a, x, and y.

operator*() [2/16]

__host__ __device__ float2 quda::operator* ( const float  a, const float2  x  )

inline

Definition at line 57 of file float_vector.h.

References a, x, and y.

operator*() [3/16]

__host__ __device__ double2 quda::operator* ( const double  a, const double2  x  )

inline

Definition at line 64 of file float_vector.h.

References a, x, and y.

operator*() [4/16]

__host__ __device__ double4 quda::operator* ( const double  a, const double4  x  )

inline

Definition at line 71 of file float_vector.h.

References a, x, and y.

operator*() [5/16]

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::operator* ( const complex< ValueType > &  lhs, const complex< ValueType > &  rhs  )

inline

Definition at line 727 of file complex_quda.h.

operator*() [6/16]

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::operator* ( const complex< ValueType > &  lhs, const ValueType &  rhs  )

inline

Definition at line 736 of file complex_quda.h.

operator*() [7/16]

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::operator* ( const ValueType &  lhs, const complex< ValueType > &  rhs  )

inline

Definition at line 743 of file complex_quda.h.

operator*() [8/16]

template<template< typename, int > class Mat, class T , int N, class S >

__device__ __host__ Mat<T,N> quda::operator* ( const S &  scalar, const Mat< T, N > &  a  )

inline

Definition at line 366 of file quda_matrix.h.

References a, and fused_exterior_ndeg_tm_dslash_cuda_gen::i.

operator*() [9/16]

template<template< typename, int > class Mat, class T , int N, class S >

__device__ __host__ Mat<T,N> quda::operator* ( const Mat< T, N > &  a, const S &  scalar  )

inline

Definition at line 374 of file quda_matrix.h.

References a.

operator*() [10/16]

template<template< typename, int > class Mat, class T , int N>

__device__ __host__ Mat<T,N> quda::operator* ( const Mat< T, N > &  a, const Mat< T, N > &  b  )

inline

Generic implementation of matrix multiplication.

Definition at line 397 of file quda_matrix.h.

References a, b, and fused_exterior_ndeg_tm_dslash_cuda_gen::i.

operator*() [11/16]

template<template< typename > class complex, typename T , int N>

__device__ __host__ Matrix<complex<T>,N> quda::operator* ( const Matrix< complex< T >, N > &  a, const Matrix< complex< T >, N > &  b  )

inline

Specialization of complex matrix multiplication that will issue optimal fma instructions.

Definition at line 418 of file quda_matrix.h.

References a, b, and fused_exterior_ndeg_tm_dslash_cuda_gen::i.

operator*() [12/16]

template<class T , class U , int N>

__device__ __host__ Matrix<typename PromoteTypeId<T,U>::Type,N> quda::operator* ( const Matrix< T, N > &  a, const Matrix< U, N > &  b  )

inline

Definition at line 453 of file quda_matrix.h.

References a, b, and fused_exterior_ndeg_tm_dslash_cuda_gen::i.

operator*() [13/16]

template<class T >

__device__ __host__ Matrix<T,2> quda::operator* ( const Matrix< T, 2 > &  a, const Matrix< T, 2 > &  b  )

inline

Definition at line 473 of file quda_matrix.h.

References a, and b.

operator*() [14/16]

template<typename Float , int Nc, int Ns, typename S >

__device__ __host__ ColorSpinor<Float,Nc,Ns> quda::operator* ( const S &  a, const ColorSpinor< Float, Nc, Ns > &  x  )

inline

Compute the scalar-vector product y = a * x.

Parameters

[in]	a	Input scalar
[in]	x	Input vector

Returns

The vector a * x

Definition at line 929 of file color_spinor.h.

References a, fused_exterior_ndeg_tm_dslash_cuda_gen::i, s, x, and y.

operator*() [15/16]

template<typename Float , int Nc, int Ns>

__device__ __host__ ColorSpinor<Float,Nc,Ns> quda::operator* ( const Matrix< complex< Float >, Nc > &  A, const ColorSpinor< Float, Nc, Ns > &  x  )

inline

Compute the matrix-vector product y = A * x.

Parameters

[in]	A	Input matrix
[in]	x	Input vector

Returns

The vector A * x

Definition at line 951 of file color_spinor.h.

References fused_exterior_ndeg_tm_dslash_cuda_gen::i, s, x, and y.

operator*() [16/16]

template<typename Float , int Nc, int Ns>

__device__ __host__ ColorSpinor<Float,Nc,Ns> quda::operator* ( const HMatrix< Float, Nc *Ns > &  A, const ColorSpinor< Float, Nc, Ns > &  x  )

inline

Compute the matrix-vector product y = A * x.

Parameters

[in]	A	Input Hermitian matrix with dimensions NcxNs x NcxNs
[in]	x	Input vector

Returns

The vector A * x

Definition at line 986 of file color_spinor.h.

References fused_exterior_ndeg_tm_dslash_cuda_gen::i, x, and y.

operator*=() [1/7]

__host__ __device__ float2 quda::operator*= ( float2 &  x, const float  a  )

inline

Definition at line 151 of file float_vector.h.

References a, and x.

operator*=() [2/7]

__host__ __device__ double2 quda::operator*= ( double2 &  x, const float  a  )

inline

Definition at line 157 of file float_vector.h.

References a, and x.

operator*=() [3/7]

__host__ __device__ float4 quda::operator*= ( float4 &  a, const float &  b  )

inline

Definition at line 163 of file float_vector.h.

References a, and b.

operator*=() [4/7]

__host__ __device__ double2 quda::operator*= ( double2 &  a, const double &  b  )

inline

Definition at line 171 of file float_vector.h.

References a, and b.

operator*=() [5/7]

__host__ __device__ double4 quda::operator*= ( double4 &  a, const double &  b  )

inline

Definition at line 177 of file float_vector.h.

References a, and b.

operator*=() [6/7]

template<template< typename, int > class Mat, class T , int N, class S >

__device__ __host__ Mat<T,N> quda::operator*= ( Mat< T, N > &  a, const S &  scalar  )

inline

Definition at line 379 of file quda_matrix.h.

References a.

operator*=() [7/7]

template<class T , int N>

__device__ __host__ Matrix<T,N> quda::operator*= ( Matrix< T, N > &  a, const Matrix< T, N > &  b  )

inline

Definition at line 442 of file quda_matrix.h.

References a, b, and c.

operator+() [1/13]

__host__ __device__ double2 quda::operator+ ( const double2 &  x, const double2 &  y  )

inline

Definition at line 24 of file float_vector.h.

References x, and y.

operator+() [2/13]

__host__ __device__ double3 quda::operator+ ( const double3 &  x, const double3 &  y  )

inline

Definition at line 40 of file float_vector.h.

References x, and y.

operator+() [3/13]

__host__ __device__ double4 quda::operator+ ( const double4 &  x, const double4 &  y  )

inline

Definition at line 44 of file float_vector.h.

References x, and y.

operator+() [4/13]

__host__ __device__ float2 quda::operator+ ( const float2  x, const float2  y  )

inline

Definition at line 80 of file float_vector.h.

References x, y, and z.

operator+() [5/13]

__host__ __device__ float4 quda::operator+ ( const float4  x, const float4  y  )

inline

Definition at line 87 of file float_vector.h.

References x, y, and z.

operator+() [6/13]

template<typename scalar , int n>

__device__ __host__ vector_type<scalar,n> quda::operator+ ( const vector_type< scalar, n > &  a, const vector_type< scalar, n > &  b  )

inline

Definition at line 88 of file cub_helper.cuh.

References a, b, c, fused_exterior_ndeg_tm_dslash_cuda_gen::i, and n.

operator+() [7/13]

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::operator+ ( const complex< ValueType > &  lhs, const complex< ValueType > &  rhs  )

inline

Definition at line 679 of file complex_quda.h.

operator+() [8/13]

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::operator+ ( const complex< ValueType > &  lhs, const ValueType &  rhs  )

inline

Definition at line 695 of file complex_quda.h.

operator+() [9/13]

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::operator+ ( const ValueType &  lhs, const complex< ValueType > &  rhs  )

inline

Definition at line 701 of file complex_quda.h.

operator+() [10/13]

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::operator+ ( const complex< ValueType > &  rhs )

inline

Definition at line 800 of file complex_quda.h.

operator+() [11/13]

template<template< typename, int > class Mat, class T , int N>

__device__ __host__ Mat<T,N> quda::operator+ ( const Mat< T, N > &  a, const Mat< T, N > &  b  )

inline

Definition at line 323 of file quda_matrix.h.

References a, b, and fused_exterior_ndeg_tm_dslash_cuda_gen::i.

operator+() [12/13]

template<typename ValueType >

__host__ __device__ complex<ValueType> quda::operator+ ( const volatile complex< ValueType > &  lhs, const volatile complex< ValueType > &  rhs  )

inline

Definition at line 687 of file complex_quda.h.

operator+() [13/13]

template<typename Float , int Nc, int Ns>

__device__ __host__ ColorSpinor<Float,Nc,Ns> quda::operator+ ( const ColorSpinor< Float, Nc, Ns > &  x, const ColorSpinor< Float, Nc, Ns > &  y  )

inline

ColorSpinor addition operator.

Parameters

[in]	x	Input vector
[in]	y	Input vector

Returns

The vector x + y

Definition at line 885 of file color_spinor.h.

References fused_exterior_ndeg_tm_dslash_cuda_gen::i, s, x, y, and z.

operator+=() [1/8]

template<typename real , typename Link >

__device__ void quda::operator+= ( real *  y, const Link &  x  )

inline

Definition at line 86 of file clover_deriv_quda.cu.

References deg_tm_dslash_cuda_gen::block(), blockDim, for(), fused_exterior_ndeg_tm_dslash_cuda_gen::i, x, and y.

Here is the call graph for this function:

operator+=() [2/8]

__host__ __device__ float4 quda::operator+= ( float4 &  x, const float4  y  )

inline

Definition at line 96 of file float_vector.h.

References x, and y.

operator+=() [3/8]

__host__ __device__ float2 quda::operator+= ( float2 &  x, const float2  y  )

inline

Definition at line 104 of file float_vector.h.

References x, and y.

operator+=() [4/8]

__host__ __device__ double2 quda::operator+= ( double2 &  x, const double2  y  )

inline

Definition at line 110 of file float_vector.h.

References x, and y.

operator+=() [5/8]

__host__ __device__ double3 quda::operator+= ( double3 &  x, const double3  y  )

inline

Definition at line 116 of file float_vector.h.

References x, and y.

operator+=() [6/8]

__host__ __device__ double4 quda::operator+= ( double4 &  x, const double4  y  )

inline

Definition at line 123 of file float_vector.h.

References x, and y.

operator+=() [7/8]

template<template< typename, int > class Mat, class T , int N>

__device__ __host__ Mat<T,N> quda::operator+= ( Mat< T, N > &  a, const Mat< T, N > &  b  )

inline

Definition at line 333 of file quda_matrix.h.

References a, b, and fused_exterior_ndeg_tm_dslash_cuda_gen::i.

operator+=() [8/8]

template<template< typename, int > class Mat, class T , int N>

__device__ __host__ Mat<T,N> quda::operator+= ( Mat< T, N > &  a, const T &  b  )

inline

Definition at line 341 of file quda_matrix.h.

References a, b, and fused_exterior_ndeg_tm_dslash_cuda_gen::i.

operator-() [1/12]

__host__ __device__ double2 quda::operator- ( const double2 &  x, const double2 &  y  )

inline

Definition at line 28 of file float_vector.h.

References x, and y.

operator-() [2/12]

__host__ __device__ float2 quda::operator- ( const float2 &  x, const float2 &  y  )

inline

Definition at line 32 of file float_vector.h.

References x, and y.

operator-() [3/12]

__host__ __device__ float4 quda::operator- ( const float4 &  x, const float4 &  y  )

inline

Definition at line 36 of file float_vector.h.

References x, and y.

operator-() [4/12]

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::operator- ( const complex< ValueType > &  lhs, const complex< ValueType > &  rhs  )

inline

Definition at line 708 of file complex_quda.h.

operator-() [5/12]

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::operator- ( const complex< ValueType > &  lhs, const ValueType &  rhs  )

inline

Definition at line 714 of file complex_quda.h.

operator-() [6/12]

__host__ __device__ float2 quda::operator- ( const float2 &  x )

inline

Definition at line 185 of file float_vector.h.

References x.

operator-() [7/12]

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::operator- ( const ValueType &  lhs, const complex< ValueType > &  rhs  )

inline

Definition at line 720 of file complex_quda.h.

operator-() [8/12]

__host__ __device__ double2 quda::operator- ( const double2 &  x )

inline

Definition at line 189 of file float_vector.h.

References x.

operator-() [9/12]

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::operator- ( const complex< ValueType > &  rhs )

inline

Definition at line 805 of file complex_quda.h.

operator-() [10/12]

template<template< typename, int > class Mat, class T , int N>

__device__ __host__ Mat<T,N> quda::operator- ( const Mat< T, N > &  a, const Mat< T, N > &  b  )

inline

Definition at line 357 of file quda_matrix.h.

References a, b, and fused_exterior_ndeg_tm_dslash_cuda_gen::i.

operator-() [11/12]

template<template< typename, int > class Mat, class T , int N>

__device__ __host__ Mat<T,N> quda::operator- ( const Mat< T, N > &  a )

inline

Definition at line 385 of file quda_matrix.h.

References a, and fused_exterior_ndeg_tm_dslash_cuda_gen::i.

operator-() [12/12]

template<typename Float , int Nc, int Ns>

__device__ __host__ ColorSpinor<Float,Nc,Ns> quda::operator- ( const ColorSpinor< Float, Nc, Ns > &  x, const ColorSpinor< Float, Nc, Ns > &  y  )

inline

ColorSpinor subtraction operator.

Parameters

[in]	x	Input vector
[in]	y	Input vector

Returns

The vector x + y

Definition at line 907 of file color_spinor.h.

References fused_exterior_ndeg_tm_dslash_cuda_gen::i, s, x, y, and z.

operator-=() [1/5]

template<typename real , typename Link >

__device__ void quda::operator-= ( real *  y, const Link &  x  )

inline

Definition at line 97 of file clover_deriv_quda.cu.

References deg_tm_dslash_cuda_gen::block(), blockDim, for(), fused_exterior_ndeg_tm_dslash_cuda_gen::i, x, and y.

Here is the call graph for this function:

operator-=() [2/5]

__host__ __device__ float4 quda::operator-= ( float4 &  x, const float4  y  )

inline

Definition at line 131 of file float_vector.h.

References x, and y.

operator-=() [3/5]

__host__ __device__ float2 quda::operator-= ( float2 &  x, const float2  y  )

inline

Definition at line 139 of file float_vector.h.

References x, and y.

operator-=() [4/5]

__host__ __device__ double2 quda::operator-= ( double2 &  x, const double2  y  )

inline

Definition at line 145 of file float_vector.h.

References x, and y.

operator-=() [5/5]

template<template< typename, int > class Mat, class T , int N>

__device__ __host__ Mat<T,N> quda::operator-= ( Mat< T, N > &  a, const Mat< T, N > &  b  )

inline

Definition at line 349 of file quda_matrix.h.

References a, b, and fused_exterior_ndeg_tm_dslash_cuda_gen::i.

operator/() [1/7]

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::operator/ ( const complex< ValueType > &  lhs, const complex< ValueType > &  rhs  )

inline

Definition at line 751 of file complex_quda.h.

References norm().

Here is the call graph for this function:

operator/() [2/7]

template<>

__host__ __device__ complex< float > quda::operator/ ( const complex< float > &  lhs, const complex< float > &  rhs  )

inline

Definition at line 760 of file complex_quda.h.

operator/() [3/7]

template<>

__host__ __device__ complex< double > quda::operator/ ( const complex< double > &  lhs, const complex< double > &  rhs  )

inline

Definition at line 766 of file complex_quda.h.

operator/() [4/7]

template<typename ValueType >

__host__ __device__ complex<ValueType> quda::operator/ ( const complex< ValueType > &  lhs, const ValueType &  rhs  )

inline

Definition at line 772 of file complex_quda.h.

operator/() [5/7]

template<typename ValueType >

__host__ __device__ complex<ValueType> quda::operator/ ( const ValueType &  lhs, const complex< ValueType > &  rhs  )

inline

Definition at line 779 of file complex_quda.h.

References norm().

Here is the call graph for this function:

operator/() [6/7]

template<>

__host__ __device__ complex<float> quda::operator/ ( const float &  lhs, const complex< float > &  rhs  )

inline

Definition at line 787 of file complex_quda.h.

operator/() [7/7]

template<>

__host__ __device__ complex<double> quda::operator/ ( const double &  lhs, const complex< double > &  rhs  )

inline

Definition at line 792 of file complex_quda.h.

operator<<() [1/8]

std::ostream & quda::operator<<	(	std::ostream &	output,
		const CloverFieldParam &	param
	)

Definition at line 404 of file clover_field.cpp.

References param.

operator<<() [2/8]

std::ostream & quda::operator<<	(	std::ostream &	output,
		const LatticeFieldParam &	param
	)

Definition at line 566 of file lattice_field.cpp.

References fused_exterior_ndeg_tm_dslash_cuda_gen::i, and param.

operator<<() [3/8]

std::ostream & quda::operator<<	(	std::ostream &	output,
		const GaugeFieldParam &	param
	)

Definition at line 254 of file gauge_field.cpp.

References QudaGaugeParam_s::anisotropy, param, QUDA_RECONSTRUCT_NO, QudaGaugeParam_s::reconstruct, QudaGaugeParam_s::scale, and QudaGaugeParam_s::t_boundary.

operator<<() [4/8]

template<typename ValueType , class charT , class traits >

std::basic_ostream< charT, traits > & quda::operator<<	(	std::basic_ostream< charT, traits > &	os,
		const complex< ValueType > &	z
	)

Definition at line 295 of file complex_quda.h.

References z.

operator<<() [5/8]

template<class T , int N>

std::ostream& quda::operator<<	(	std::ostream &	os,
		const Matrix< T, N > &	m
	)

Definition at line 723 of file quda_matrix.h.

References fused_exterior_ndeg_tm_dslash_cuda_gen::i.

operator<<() [6/8]

template<class T , int N>

std::ostream& quda::operator<<	(	std::ostream &	os,
		const Array< T, N > &	a
	)

Definition at line 737 of file quda_matrix.h.

References a, and fused_exterior_ndeg_tm_dslash_cuda_gen::i.

operator<<() [7/8]

std::ostream& quda::operator<<	(	std::ostream &	out,
		const ColorSpinorField &	a
	)

Definition at line 833 of file color_spinor_field.cpp.

References a, d, and out.

operator<<() [8/8]

std::ostream& quda::operator<<	(	std::ostream &	out,
		const cudaColorSpinorField &	a
	)

Definition at line 1446 of file cuda_color_spinor_field.cu.

References a, and out.

operator==() [1/3]

template<typename ValueType >

__host__ __device__ bool quda::operator== ( const complex< ValueType > &  lhs, const complex< ValueType > &  rhs  )

inline

Definition at line 812 of file complex_quda.h.

Referenced by std::__1::__attribute(), and std::__1::__attribute__().

Here is the caller graph for this function:

operator==() [2/3]

template<typename ValueType >

__host__ __device__ bool quda::operator== ( const ValueType &  lhs, const complex< ValueType > &  rhs  )

inline

Definition at line 821 of file complex_quda.h.

operator==() [3/3]

template<typename ValueType >

__host__ __device__ bool quda::operator== ( const complex< ValueType > &  lhs, const ValueType &  rhs  )

inline

Definition at line 829 of file complex_quda.h.

operator>>()

template<typename ValueType , typename charT , class traits >

std::basic_istream< charT, traits > & quda::operator>>	(	std::basic_istream< charT, traits > &	is,
		complex< ValueType > &	z
	)

Definition at line 303 of file complex_quda.h.

References z.

orthoDir()

void quda::orthoDir	(	Complex **	beta,
		std::vector< ColorSpinorField *>	Ap,
		int	k,
		int	pipeline
	)

Definition at line 83 of file inv_gcr_quda.cpp.

References quda::blas::caxpy(), quda::blas::caxpyDotzy(), quda::blas::cDotProduct(), computeBeta(), errorQuda, fused_exterior_ndeg_tm_dslash_cuda_gen::i, pipeline, and updateAp().

Referenced by quda::GCR::operator()().

Here is the call graph for this function:

Here is the caller graph for this function:

outerProd() [1/2]

template<class T , int N>

__device__ __host__ void quda::outerProd ( const Array< T, N > &  a, const Array< T, N > &  b, Matrix< T, N > *  m  )

inline

Definition at line 695 of file quda_matrix.h.

References a, b, conj(), and fused_exterior_ndeg_tm_dslash_cuda_gen::i.

Referenced by constructHHMat().

Here is the call graph for this function:

Here is the caller graph for this function:

outerProd() [2/2]

template<class T , int N>

__device__ __host__ void quda::outerProd ( const T(&)  a[N], const T(&)  b[N], Matrix< T, N > *  m  )

inline

Definition at line 708 of file quda_matrix.h.

References a, b, conj(), and fused_exterior_ndeg_tm_dslash_cuda_gen::i.

Here is the call graph for this function:

outerProdSpinTrace()

template<typename Float , int Nc, int Ns>

__device__ __host__ Matrix<complex<Float>,Nc> quda::outerProdSpinTrace ( const ColorSpinor< Float, Nc, Ns > &  a, const ColorSpinor< Float, Nc, Ns > &  b  )

inline

Compute the outer product over color and take the spin trace out(j,i) = a(s,j) * conj (b(s,i))

Parameters

a	Left-hand side ColorSpinor
b	Right-hand side ColorSpinor

Returns

The spin traced matrix

Definition at line 849 of file color_spinor.h.

References a, b, fused_exterior_ndeg_tm_dslash_cuda_gen::i, out, and s.

OvrImpSTOUTStep() [1/3]

void quda::OvrImpSTOUTStep	(	GaugeField &	dataDs,
		const GaugeField &	dataOr,
		double	rho,
		double	epsilon
	)

Apply Over Improved STOUT smearing to the gauge field

Parameters

dataDs	Output smeared field
dataOr	Input gauge field
rho	smearing parameter
epsilon	smearing parameter

Definition at line 801 of file gauge_stout.cu.

References errorQuda, float, quda::GaugeField::isNative(), quda::GaugeField::Order(), quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, QUDA_HALF_PRECISION, QUDA_SINGLE_PRECISION, and quda::GaugeField::Reconstruct().

Referenced by OvrImpSTOUTStep(), and performOvrImpSTOUTnStep().

Here is the call graph for this function:

Here is the caller graph for this function:

OvrImpSTOUTStep() [2/3]

template<typename Float , typename GaugeOr , typename GaugeDs >

void quda::OvrImpSTOUTStep	(	GaugeOr	origin,
		GaugeDs	dest,
		const GaugeField &	dataOr,
		Float	rho,
		Float	epsilon
	)

Definition at line 740 of file gauge_stout.cu.

References arg(), DOUBLE_TOL, quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, qudaDeviceSynchronize(), and SINGLE_TOL.

Here is the call graph for this function:

OvrImpSTOUTStep() [3/3]

template<typename Float >

void quda::OvrImpSTOUTStep	(	GaugeField &	dataDs,
		const GaugeField &	dataOr,
		Float	rho,
		Float	epsilon
	)

Definition at line 749 of file gauge_stout.cu.

References errorQuda, OvrImpSTOUTStep(), QUDA_RECONSTRUCT_12, QUDA_RECONSTRUCT_8, QUDA_RECONSTRUCT_NO, and quda::GaugeField::Reconstruct().

Here is the call graph for this function:

packFace()

void quda::packFace	(	void *	ghost_buf[2 *QUDA_MAX_DIM],
		cudaColorSpinorField &	in,
		MemoryLocation	location,
		const int	nFace,
		const int	dagger,
		const int	parity,
		const int	dim,
		const int	face_num,
		const cudaStream_t &	stream,
		const double	a = 0.0,
		const double	b = 0.0
	)

Dslash face packing routine.

Parameters

[out]	ghost_buf	Array of packed halos, order is [2*dim+dir]
[in]	in	Input ColorSpinorField to be packed
[in]	location	Locations where the packed fields are (Device, Host and/or Remote)
[in]	nFace	Depth of halo
[in]	dagger	Whether this is for the dagger operator
[in]	parity	Field parity
[in]	dim	Which dimensions we are packing
[in]	face_num	Are we packing backwards (0), forwards (1) or both directions (2)
[in]	stream	Which stream are we executing in
[in]	a	Packing coefficient (twisted-mass only)
[in]	b	Packing coefficient (twisted-mass only)

Referenced by quda::cudaColorSpinorField::packGhost().

Here is the caller graph for this function:

packFaceExtended()

void quda::packFaceExtended	(	void *	ghost_buf[2 *QUDA_MAX_DIM],
		cudaColorSpinorField &	field,
		MemoryLocation	location,
		const int	nFace,
		const int	R[],
		const int	dagger,
		const int	parity,
		const int	dim,
		const int	face_num,
		const cudaStream_t &	stream,
		const bool	unpack = false
	)

Referenced by quda::cudaColorSpinorField::packGhostExtended(), and quda::cudaColorSpinorField::unpackGhostExtended().

Here is the caller graph for this function:

packGhost()

template<typename Float , int Ns, int Ms, int Nc, int Mc, int nDim, typename Arg >

__device__ __host__ void quda::packGhost ( Arg &  arg, int  cb_idx, int  parity, int  spinor_parity, int  spin_block, int  color_block  )

inline

Definition at line 47 of file color_spinor_pack.cu.

References arg(), c, dim, getCoords(), getCoords5(), parity, s, and x.

Here is the call graph for this function:

packSpinor()

template<typename FloatOut , typename FloatIn , int Ns, int Nc, typename OutOrder , typename InOrder >

void quda::packSpinor	(	OutOrder &	outOrder,
		const InOrder &	inOrder,
		int	volume
	)

CPU function to reorder spinor fields.

Definition at line 22 of file copy_color_spinor_mg.cuh.

References c, s, and x.

packSpinorKernel()

template<typename FloatOut , typename FloatIn , int Ns, int Nc, typename OutOrder , typename InOrder >

__global__ void quda::packSpinorKernel	(	OutOrder	outOrder,
		const InOrder	inOrder,
		int	volume
	)

CUDA kernel to reorder spinor fields. Adopts a similar form as the CPU version, using the same inlined functions.

Definition at line 34 of file copy_color_spinor_mg.cuh.

References blockDim, c, s, and x.

PGaugeExchange()

void quda::PGaugeExchange	(	cudaGaugeField &	data,
		const int	dir,
		const int	parity
	)

Perform heatbath and overrelaxation. Performs nhb heatbath steps followed by nover overrelaxation steps.

Parameters

[in,out]	data	Gauge field
[in,out]	rngstate	state of the CURAND random number generator
[in]	Beta	inverse of the gauge coupling, beta = 2 Nc / g_0^2
[in]	nhb	number of heatbath steps
[in]	nover	number of overrelaxation steps

Definition at line 345 of file pgauge_exchange.cu.

References comm_dim_partitioned(), errorQuda, parity, quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, QUDA_HALF_PRECISION, and QUDA_SINGLE_PRECISION.

Here is the call graph for this function:

PGaugeExchangeFree()

void quda::PGaugeExchangeFree

(

)

Release all allocated memory used to exchange data between nodes.

Referenced by main(), and GaugeAlgTest::TearDown().

Here is the caller graph for this function:

pinned_allocated_peak()

long quda::pinned_allocated_peak

(

)

Returns

peak pinned memory allocated

Definition at line 59 of file malloc.cpp.

References max_total_bytes, and PINNED.

pinned_malloc_()

void * quda::pinned_malloc_	(	const char *	func,
		const char *	file,
		int	line,
		size_t	size
	)

Allocate page-locked ("pinned") host memory. This function should only be called via the pinned_malloc() macro, defined in malloc_quda.h

Note that we do not rely on cudaHostAlloc(), since buffers allocated in this way have been observed to cause problems when shared with MPI via GPU Direct on some systems.

Definition at line 246 of file malloc.cpp.

References a, aligned_malloc(), err, errorQuda, func, memset(), PINNED, printfQuda, ptr, size, and track_malloc().

Referenced by quda::pool::pinned_malloc_().

Here is the call graph for this function:

Here is the caller graph for this function:

plaquette()

double3 quda::plaquette	(	const GaugeField &	U,
		QudaFieldLocation	location
	)

Compute the plaquette of the gauge field

Parameters

U	The gauge field upon which to compute the plaquette
location	The locaiton where to do the computation

Returns

double3 variable returning (plaquette, spatial plaquette, temporal plaquette) site averages normalized such that each plaquette is in the range [0,1]

Definition at line 138 of file gauge_plaq.cu.

References errorQuda, and INSTANTIATE_PRECISION.

Referenced by main(), performAPEnStep(), performOvrImpSTOUTnStep(), performSTOUTnStep(), plaqQuda(), GaugeAlgTest::SetUp(), and TEST_F().

Here is the caller graph for this function:

point()

template<class T >

void quda::point	(	T &	t,
		int	x,
		int	s,
		int	c
	)

Create a point source at spacetime point x, spin s and colour c

Definition at line 30 of file color_spinor_util.cu.

References c, s, t, and x.

Referenced by genericSource().

Here is the caller graph for this function:

polar() [1/3]

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::polar ( const ValueType &  m, const ValueType &  theta = 0  )

inline

Returns the complex with magnitude m and angle theta in radians.

Definition at line 902 of file complex_quda.h.

References cos(), and sin().

Referenced by construct_fat_long_gauge_field(), exp(), and sqrt().

Here is the call graph for this function:

Here is the caller graph for this function:

polar() [2/3]

template<>

__host__ __device__ complex<float> quda::polar ( const float &  magnitude, const float &  angle  )

inline

Definition at line 908 of file complex_quda.h.

References cosf(), and sinf().

Here is the call graph for this function:

polar() [3/3]

template<>

__host__ __device__ complex<double> quda::polar ( const double &  magnitude, const double &  angle  )

inline

Definition at line 914 of file complex_quda.h.

References cos(), and sin().

Here is the call graph for this function:

polarSu3()

template<typename Float >

__host__ __device__ void quda::polarSu3	(	Matrix< complex< Float >, 3 > &	in,
		Float	tol
	)

Project the input matrix on the SU(3) group. First unitarize the matrix and then project onto the special unitary group.

Parameters

in	The input matrix to which we're projecting
tol	Tolerance to which this check is applied

Definition at line 71 of file su3_project.cuh.

References atan2(), checkUnitary(), computeMatrixInverse(), conj(), cos(), getDeterminant(), in, mod(), out, pow(), sin(), and tol.

Here is the call graph for this function:

policyTuning()

bool quda::policyTuning

(

)

Definition at line 453 of file tune.cpp.

References policy_tuning.

Referenced by tuneLaunch().

Here is the caller graph for this function:

pow() [1/6]

template<typename ValueType , typename ExponentType >

__host__ __device__ ValueType quda::pow ( ValueType  x, ExponentType  e  )

inline

Definition at line 100 of file complex_quda.h.

References e, pow(), and x.

Referenced by compareSpinor(), dslashReference_5th_inv(), exponentiate_iQ(), insertNoise(), quda::RitzMat::operator()(), quda::CG::operator()(), quda::MultiShiftCG::operator()(), polarSu3(), TEST(), and TEST_P().

Here is the call graph for this function:

Here is the caller graph for this function:

pow() [2/6]

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::pow ( const complex< ValueType > &  z, const int &  n  )

inline

Definition at line 1012 of file complex_quda.h.

References exp(), log(), and z.

Here is the call graph for this function:

pow() [3/6]

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::pow ( const complex< ValueType > &  z, const ValueType &  x  )

inline

Definition at line 988 of file complex_quda.h.

References exp(), log(), and z.

Here is the call graph for this function:

pow() [4/6]

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::pow ( const complex< ValueType > &  z, const complex< ValueType > &  z2  )

inline

Definition at line 994 of file complex_quda.h.

References exp(), log(), and z.

Here is the call graph for this function:

pow() [5/6]

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::pow ( const ValueType &  x, const complex< ValueType > &  z  )

inline

Definition at line 1000 of file complex_quda.h.

References exp(), log(), and x.

Here is the call graph for this function:

pow() [6/6]

template<>

__host__ __device__ complex<float> quda::pow ( const float &  x, const complex< float > &  exponent  )

inline

Definition at line 1006 of file complex_quda.h.

References exp(), logf(), and x.

Referenced by pow().

Here is the call graph for this function:

Here is the caller graph for this function:

Precision_() [1/2]

QudaPrecision quda::Precision_ ( const char *  func, const char *  file, int  line, const LatticeField &  a, const LatticeField &  b  )

inline

Helper function for determining if the precision of the fields is the same.

Parameters

[in]	a	Input field
[in]	b	Input field

Returns

If precision is unique return the precision

Definition at line 569 of file lattice_field.h.

References a, b, errorQuda, func, and QUDA_INVALID_PRECISION.

Referenced by Precision_().

Here is the caller graph for this function:

Precision_() [2/2]

template<typename... Args>

QudaPrecision quda::Precision_ ( const char *  func, const char *  file, int  line, const LatticeField &  a, const LatticeField &  b, const Args &...  args  )

inline

Helper function for determining if the precision of the fields is the same.

Parameters

[in]	a	Input field
[in]	b	Input field
[in]	args	List of additional fields to check precision on

Returns

If precision is unique return the precision

Definition at line 586 of file lattice_field.h.

References a, args, b, func, and Precision_().

Here is the call graph for this function:

print()

void quda::print	(	const double	d[],
		int	n
	)

Definition at line 44 of file inv_mpcg_quda.cpp.

References d, fused_exterior_ndeg_tm_dslash_cuda_gen::i, and n.

print_alloc()

static void quda::print_alloc ( AllocType  type )

static

Definition at line 83 of file malloc.cpp.

References a, alloc, entry, printfQuda, and ptr.

Referenced by assertAllMemFree().

Here is the caller graph for this function:

print_alloc_header()

static void quda::print_alloc_header ( )

static

Definition at line 76 of file malloc.cpp.

References printfQuda.

Referenced by assertAllMemFree().

Here is the caller graph for this function:

print_trace()

static void quda::print_trace ( void  )

static

Definition at line 65 of file malloc.cpp.

References array, free(), fused_exterior_ndeg_tm_dslash_cuda_gen::i, printfQuda, and size.

Referenced by host_free_().

Here is the call graph for this function:

Here is the caller graph for this function:

print_vector()

template<class Order >

void quda::print_vector	(	const Order &	o,
		unsigned int	x
	)

Definition at line 267 of file color_spinor_util.cu.

References c, parity, s, and x.

Referenced by genericPrintVector().

Here is the caller graph for this function:

printAPIProfile()

void quda::printAPIProfile

(

)

Print out the timer profile for CUDA API calls.

Definition at line 303 of file quda_cuda_api.cpp.

References apiTimer, and quda::TimeProfile::Print().

Referenced by endQuda().

Here is the call graph for this function:

Here is the caller graph for this function:

printLaunchTimer()

void quda::printLaunchTimer

(

)

Definition at line 797 of file tune.cpp.

References launchTimer, and quda::TimeProfile::Print().

Referenced by endQuda().

Here is the call graph for this function:

Here is the caller graph for this function:

printLink()

template<class Cmplx >

__host__ __device__ void quda::printLink ( const Matrix< Cmplx, 3 > &  link )

inline

Definition at line 1039 of file quda_matrix.h.

References printf(), x, and y.

Here is the call graph for this function:

printPeakMemUsage()

void quda::printPeakMemUsage

(

)

Definition at line 371 of file malloc.cpp.

References DEVICE, max_total_bytes, max_total_host_bytes, max_total_pinned_bytes, and printfQuda.

Referenced by endQuda().

Here is the caller graph for this function:

projectSU3()

void quda::projectSU3	(	cudaGaugeField &	U,
		double	tol,
		int *	fails
	)

Project the input gauge field onto the SU(3) group. This is a destructive operation. The number of link failures is reported so appropriate action can be taken.

Parameters

U	Gauge field that we are projecting onto SU(3)
tol	Tolerance to which the iterative algorithm works
fails	Number of link failures (device pointer)

Definition at line 584 of file unitarize_links_quda.cu.

References quda::ProjectSU3< Float, G >::apply(), arg(), checkCudaError, errorQuda, QUDA_RECONSTRUCT_NO, qudaDeviceSynchronize(), quda::GaugeField::Reconstruct(), and tol.

Referenced by projectSU3Quda().

Here is the call graph for this function:

Here is the caller graph for this function:

ProjectSU3kernel()

template<typename Float , typename G >

__global__ void quda::ProjectSU3kernel

(

ProjectSU3Arg< Float, G >

arg

)

Definition at line 532 of file unitarize_links_quda.cu.

References arg(), blockDim, quda::Matrix< T, N >::data, idx, isUnitary(), mu, and parity.

Here is the call graph for this function:

Prolongate()

void quda::Prolongate	(	ColorSpinorField &	out,
		const ColorSpinorField &	in,
		const ColorSpinorField &	v,
		int	Nvec,
		const int *	fine_to_coarse,
		const int *	spin_map,
		int	parity = QUDA_INVALID_PARITY
	)

Apply the prolongation operator.

Parameters

[out]	out	Resulting fine grid field
[in]	in	Input field on coarse grid
[in]	v	Matrix field containing the null-space components
[in]	Nvec	Number of null-space components
[in]	fine_to_coarse	Fine-to-coarse lookup table (linear indices)
[in]	spin_map	Spin blocking lookup table
[in]	parity	of the output fine field (if single parity output field)

Definition at line 284 of file prolongator.cu.

References checkCudaError, checkLocation, checkPrecision, errorQuda, quda::ColorSpinorField::FieldOrder(), in, out, parity, QUDA_CUDA_FIELD_LOCATION, QUDA_DOUBLE_PRECISION, and QUDA_SINGLE_PRECISION.

Referenced by quda::Transfer::P().

Here is the call graph for this function:

Here is the caller graph for this function:

qudaDeviceSynchronize()

cudaError_t quda::qudaDeviceSynchronize

(

)

Wrapper around cudaDeviceSynchronize or cuDeviceSynchronize.

Definition at line 277 of file quda_cuda_api.cpp.

References cudaErrorUnknown, errorQuda, PROFILE, and QUDA_PROFILE_DEVICE_SYNCHRONIZE.

Referenced by quda::cublas::BatchInvertMatrix(), completeKSForce(), computeKSLongLinkForce(), computeStaggeredForceQuda(), quda::cudaGaugeField::exchangeExtendedGhost(), exchangeExtendedGhost(), quda::cudaGaugeField::exchangeGhost(), quda::cudaColorSpinorField::exchangeGhost(), fatLongKSLink(), quda::cudaGaugeField::injectGhost(), launch_kernel_random(), quda::cudaGaugeField::loadCPUField(), OvrImpSTOUTStep(), projectSU3(), quda::Transfer::R(), remove_staggered_phase_quda_(), and quda::cudaGaugeField::saveCPUField().

Here is the caller graph for this function:

qudaEventQuery()

cudaError_t quda::qudaEventQuery

(

cudaEvent_t &

event

)

Wrapper around cudaEventQuery or cuEventQuery.

Parameters

[in]

event

Event we are querying

Returns

Status of event query

Definition at line 190 of file quda_cuda_api.cpp.

References cudaErrorUnknown, errorQuda, event, PROFILE, and QUDA_PROFILE_EVENT_QUERY.

Referenced by multiReduceLaunch(), anonymous_namespace{dslash_policy.cuh}::DslashBasic::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashPthreads::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashFusedExterior::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashGDRRecv::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashFusedGDRRecv::operator()(), and reduceLaunch().

Here is the caller graph for this function:

qudaEventRecord()

cudaError_t quda::qudaEventRecord	(	cudaEvent_t &	event,
		cudaStream_t	stream = 0
	)

Wrapper around cudaEventRecord or cuEventRecord.

Parameters

[in,out]	event	Event we are recording
[in,out]	stream	Stream where to record the event

Definition at line 209 of file quda_cuda_api.cpp.

References cudaErrorUnknown, errorQuda, event, PROFILE, QUDA_PROFILE_EVENT_RECORD, and stream.

Referenced by exchangeExtendedGhost(), anonymous_namespace{dslash_policy.cuh}::issueGather(), anonymous_namespace{dslash_policy.cuh}::issuePack(), multiReduceLaunch(), anonymous_namespace{dslash_policy.cuh}::DslashBasic::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashPthreads::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashFusedExterior::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashGDRRecv::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashFusedGDRRecv::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashZeroCopyPack::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashFusedZeroCopyPack::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashZeroCopyPackGDRRecv::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashFusedZeroCopyPackGDRRecv::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashZeroCopy::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashFusedZeroCopy::operator()(), reduceLaunch(), quda::cudaGaugeField::sendStart(), quda::cudaColorSpinorField::sendStart(), and shiftColorSpinorField().

Here is the caller graph for this function:

qudaEventSynchronize()

cudaError_t quda::qudaEventSynchronize

(

cudaEvent_t &

event

)

Wrapper around cudaEventSynchronize or cuEventSynchronize.

Parameters

[in]

event

Event which we are synchronizing with respect to

Definition at line 260 of file quda_cuda_api.cpp.

References cudaErrorUnknown, errorQuda, event, PROFILE, and QUDA_PROFILE_EVENT_SYNCHRONIZE.

Referenced by quda::cudaGaugeField::commsComplete().

Here is the caller graph for this function:

qudaLaunchKernel()

cudaError_t quda::qudaLaunchKernel	(	const void *	func,
		dim3	gridDim,
		dim3	blockDim,
		void **	args,
		size_t	sharedMem,
		cudaStream_t	stream
	)

Wrapper around cudaLaunchKernel.

Parameters

[in]	func	Device function symbol
[in]	gridDim	Grid dimensions
[in]	blockDim	Block dimensions
[in]	args	Arguments
[in]	sharedMem	Shared memory requested per thread block
[in]	stream	Stream identifier

Definition at line 182 of file quda_cuda_api.cpp.

References activeTuning(), args, blockDim, errorQuda, func, gridDim, PROFILE, QUDA_PROFILE_LAUNCH_KERNEL, sharedMem, and stream.

Here is the call graph for this function:

qudaMemcpy2DAsync_()

void quda::qudaMemcpy2DAsync_	(	void *	dst,
		size_t	dpitch,
		const void *	src,
		size_t	spitch,
		size_t	width,
		size_t	hieght,
		cudaMemcpyKind	kind,
		const cudaStream_t &	stream,
		const char *	func,
		const char *	file,
		const char *	line
	)

Wrapper around cudaMemcpy2DAsync or driver API equivalent Potentially add auto-profiling support.

Parameters

[out]	dst	Destination pointer
[in]	dpitch	Destination pitch
[in]	src	Source pointer
[in]	spitch	Source pitch
[in]	width	Width in bytes
[in]	height	Number of rows
[in]	kind	Type of memory copy
[in]	stream	Stream to issue copy

Definition at line 151 of file quda_cuda_api.cpp.

References dpitch, errorQuda, height, kind, param, PROFILE, QUDA_PROFILE_MEMCPY2D_D2H_ASYNC, spitch, src, stream, and width.

qudaMemcpy_()

void quda::qudaMemcpy_	(	void *	dst,
		const void *	src,
		size_t	count,
		cudaMemcpyKind	kind,
		const char *	func,
		const char *	file,
		const char *	line
	)

Wrapper around cudaMemcpy used for auto-profiling. Do not call directly, rather call macro below which will grab the location of the call.

Parameters

[out]	dst	Destination pointer
[in]	src	Source pointer
[in]	count	Size of transfer
[in]	kind	Type of memory copy

Definition at line 113 of file quda_cuda_api.cpp.

References checkCudaError, copy(), count, func, getVerbosity(), kind, printfQuda, QUDA_DEBUG_VERBOSE, and src.

Here is the call graph for this function:

qudaMemcpyAsync_()

void quda::qudaMemcpyAsync_	(	void *	dst,
		const void *	src,
		size_t	count,
		cudaMemcpyKind	kind,
		const cudaStream_t &	stream,
		const char *	func,
		const char *	file,
		const char *	line
	)

Wrapper around cudaMemcpyAsync or driver API equivalent Potentially add auto-profiling support.

Parameters

[out]	dst	Destination pointer
[in]	src	Source pointer
[in]	count	Size of transfer
[in]	kind	Type of memory copy
[in]	stream	Stream to issue copy

Definition at line 128 of file quda_cuda_api.cpp.

References count, errorQuda, kind, PROFILE, QUDA_PROFILE_MEMCPY_D2D_ASYNC, QUDA_PROFILE_MEMCPY_D2H_ASYNC, QUDA_PROFILE_MEMCPY_H2D_ASYNC, src, and stream.

qudaStreamSynchronize()

cudaError_t quda::qudaStreamSynchronize

(

cudaStream_t &

stream

)

Wrapper around cudaStreamSynchronize or cuStreamSynchronize.

Parameters

[in]

stream

Stream which we are synchronizing with respect to

Definition at line 243 of file quda_cuda_api.cpp.

References cudaErrorUnknown, errorQuda, PROFILE, QUDA_PROFILE_STREAM_SYNCHRONIZE, and stream.

Referenced by contractCuda(), quda::cudaGaugeField::exchangeGhost(), quda::cudaGaugeField::injectGhost(), anonymous_namespace{dslash_policy.cuh}::DslashZeroCopyPack::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashFusedZeroCopyPack::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashZeroCopyPackGDRRecv::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashFusedZeroCopyPackGDRRecv::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashZeroCopy::operator()(), and anonymous_namespace{dslash_policy.cuh}::DslashFusedZeroCopy::operator()().

Here is the caller graph for this function:

qudaStreamWaitEvent()

cudaError_t quda::qudaStreamWaitEvent	(	cudaStream_t	stream,
		cudaEvent_t	event,
		unsigned int	flags
	)

Wrapper around cudaEventRecord or cuEventRecord.

Parameters

[in,out]	stream	Stream which we are instructing to wait
[in]	event	Event we are waiting on
[in]	flags	Flags to pass to function

Definition at line 226 of file quda_cuda_api.cpp.

References cudaErrorUnknown, errorQuda, event, flags, PROFILE, QUDA_PROFILE_STREAM_WAIT_EVENT, and stream.

Referenced by anonymous_namespace{dslash_policy.cuh}::commsComplete(), anonymous_namespace{dslash_policy.cuh}::completeDslash(), quda::cudaColorSpinorField::exchangeGhost(), anonymous_namespace{dslash_policy.cuh}::issueGather(), anonymous_namespace{dslash_policy.cuh}::DslashBasic::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashPthreads::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashFusedExterior::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashZeroCopyPack::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashFusedZeroCopyPack::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashZeroCopyPackGDRRecv::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashFusedZeroCopyPackGDRRecv::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashZeroCopy::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashFusedZeroCopy::operator()(), and shiftColorSpinorField().

Here is the caller graph for this function:

r_slant()

constexpr const char* quda::r_slant ( const char *  str )

inline

Definition at line 47 of file malloc_quda.h.

Referenced by file_name().

Here is the caller graph for this function:

random()

template<class T >

void quda::random

(

T &

t

)

Random number insertion over all field elements

Definition at line 13 of file color_spinor_util.cu.

References c, comm_drand(), parity, s, and t.

Referenced by genericSource().

Here is the call graph for this function:

Here is the caller graph for this function:

Random() [1/2]

template<class Real >

__device__ Real quda::Random ( cuRNGState &  state, Real  a, Real  b  )

inline

Return a random number between a and b.

Parameters

state	curand rng state
a	lower range
b	upper range

Returns

random number in range a,b

Definition at line 70 of file random_quda.h.

Random() [2/2]

template<class Real >

__device__ Real quda::Random ( cuRNGState &  state )

inline

Return a random number between 0 and 1.

Parameters

state

curand rng state

Returns

random number in range 0,1

Definition at line 91 of file random_quda.h.

Random< double >() [1/2]

template<>

__device__ double quda::Random< double > ( cuRNGState &  state, double  a, double  b  )

inline

Definition at line 81 of file random_quda.h.

References a, and b.

Random< double >() [2/2]

template<>

__device__ double quda::Random< double > ( cuRNGState &  state )

inline

Definition at line 102 of file random_quda.h.

Random< float >() [1/2]

template<>

__device__ float quda::Random< float > ( cuRNGState &  state, float  a, float  b  )

inline

Definition at line 76 of file random_quda.h.

References a, and b.

Random< float >() [2/2]

template<>

__device__ float quda::Random< float > ( cuRNGState &  state )

inline

Definition at line 97 of file random_quda.h.

reduce()

template<int block_size, typename T >

__device__ void quda::reduce ( ReduceArg< T >  arg, const T &  in, const int  idx = 0  )

inline

Definition at line 163 of file cub_helper.cuh.

References arg(), idx, and in.

Referenced by multiReduceCuda(), and reduceCuda().

Here is the call graph for this function:

Here is the caller graph for this function:

reduce2d()

template<int block_size_x, int block_size_y, typename T >

__device__ void quda::reduce2d ( ReduceArg< T >  arg, const T &  in, const int  idx = 0  )

inline

Definition at line 122 of file cub_helper.cuh.

References __syncthreads(), arg(), count, gridDim, fused_exterior_ndeg_tm_dslash_cuda_gen::i, idx, in, isLastBlockDone, sum(), value, and zero().

Here is the call graph for this function:

reduceRow()

template<int block_size_x, int block_size_y, typename T >

__device__ void quda::reduceRow ( ReduceArg< T >  arg, const T &  in  )

inline

Definition at line 233 of file cub_helper.cuh.

References __syncthreads(), arg(), count, quda::ColorSpinorField::exchange(), gridDim, fused_exterior_ndeg_tm_dslash_cuda_gen::i, in, isLastBlockDone, sum(), value, and y.

Here is the call graph for this function:

reliable()

int quda::reliable	(	double &	rNorm,
		double &	maxrx,
		double &	maxrr,
		const double &	r2,
		const double &	delta
	)

Definition at line 37 of file inv_bicgstab_quda.cpp.

References delta, sqrt(), and updateR().

Referenced by quda::BiCGstab::operator()(), and quda::MultiShiftCG::operator()().

Here is the call graph for this function:

Here is the caller graph for this function:

reorder_location()

QudaFieldLocation quda::reorder_location

(

)

Return whether data is reordered on the CPU or GPU. This can set at QUDA initialization using the environment variable QUDA_REORDER_LOCATION.

Returns

Reorder location

Definition at line 585 of file lattice_field.cpp.

References reorder_location_.

Referenced by quda::cudaCloverField::copy(), quda::cudaGaugeField::copy(), quda::cpuGaugeField::copy(), quda::cudaColorSpinorField::loadSpinorField(), quda::cudaGaugeField::saveCPUField(), and quda::cudaColorSpinorField::saveSpinorField().

Here is the caller graph for this function:

reorder_location_set()

void quda::reorder_location_set

(

QudaFieldLocation

reorder_location_

)

Set whether data is reorderd on the CPU or GPU. This can set at QUDA initialization using the environment variable QUDA_REORDER_LOCATION.

Parameters

reorder_location_

The location to set where data will be reordered

Definition at line 586 of file lattice_field.cpp.

References reorder_location_.

Referenced by initQudaDevice().

Here is the caller graph for this function:

report() [1/2]

static void quda::report ( const char *  type )

static

Definition at line 7 of file eig_solver.cpp.

References getVerbosity(), printfQuda, and QUDA_VERBOSE.

Referenced by quda::Eig_Solver::create(), and quda::Solver::create().

Here is the call graph for this function:

Here is the caller graph for this function:

report() [2/2]

static void quda::report ( const char *  type )

static

Definition at line 8 of file solver.cpp.

References getVerbosity(), printfQuda, and QUDA_VERBOSE.

Here is the call graph for this function:

Restrict()

void quda::Restrict	(	ColorSpinorField &	out,
		const ColorSpinorField &	in,
		const ColorSpinorField &	v,
		int	Nvec,
		const int *	fine_to_coarse,
		const int *	coarse_to_fine,
		const int *	spin_map,
		int	parity = QUDA_INVALID_PARITY
	)

Apply the restriction operator.

Parameters

[out]	out	Resulting coarsened field
[in]	in	Input field on fine grid
[in]	v	Matrix field containing the null-space components
[in]	Nvec	Number of null-space components
[in]	fine_to_coarse	Fine-to-coarse lookup table (linear indices)
[in]	spin_map	Spin blocking lookup table
[in]	parity	of the input fine field (if single parity input field)

Definition at line 509 of file restrictor.cu.

References checkPrecision, errorQuda, quda::ColorSpinorField::FieldOrder(), in, out, parity, QUDA_DOUBLE_PRECISION, and QUDA_SINGLE_PRECISION.

Referenced by quda::Transfer::R().

Here is the call graph for this function:

Here is the caller graph for this function:

s2d()

static __host__ __device__ double quda::s2d ( const short &  a )

inlinestatic

Definition at line 135 of file register_traits.h.

References a, and MAX_SHORT_INV.

s2f()

static __host__ __device__ float quda::s2f ( const short &  a )

inlinestatic

Definition at line 134 of file register_traits.h.

References a, and MAX_SHORT_INV.

Referenced by copy().

Here is the caller graph for this function:

safe_malloc_()

void * quda::safe_malloc_	(	const char *	func,
		const char *	file,
		int	line,
		size_t	size
	)

Perform a standard malloc() with error-checking. This function should only be called via the safe_malloc() macro, defined in malloc_quda.h

Definition at line 219 of file malloc.cpp.

References a, errorQuda, func, HOST, malloc(), memset(), printfQuda, ptr, size, and track_malloc().

Here is the call graph for this function:

saveProfile()

void quda::saveProfile

(

const std::string

label = ""

)

Save profile to disk.

Definition at line 472 of file tune.cpp.

References comm_rank(), count, ctime(), entry, getenv(), getVerbosity(), gitversion, param, printfQuda, quda_hash, QUDA_SUMMARIZE, quda_version, resource_path, serializeProfile(), serializeTrace(), strcmp(), strncpy(), time(), tmp, trace_list, traceEnabled(), tunecache, and warningQuda.

Referenced by endQuda(), newDeflationQuda(), and newMultigridQuda().

Here is the call graph for this function:

Here is the caller graph for this function:

saveTuneCache()

void quda::saveTuneCache

(

)

Write tunecache to disk.

Definition at line 388 of file tune.cpp.

References comm_rank(), ctime(), getVerbosity(), gitversion, initial_cache_size, printfQuda, quda_hash, QUDA_SUMMARIZE, quda_version, resource_path, serializeTuneCache(), time(), tunecache, and warningQuda.

Referenced by endQuda(), invertMultiShiftQuda(), invertMultiSrcQuda(), invertQuda(), lanczosQuda(), and newMultigridQuda().

Here is the call graph for this function:

Here is the caller graph for this function:

serializeProfile()

static void quda::serializeProfile ( std::ostream &  out, std::ostream &  async_out  )

static

Serialize tunecache to an ostream, useful for writing to a file or sending to other nodes.

Definition at line 181 of file tune.cpp.

References quda::TuneKey::aux, entry, quda::TuneKey::name, out, param, strcmp(), strncpy(), time(), tmp, tunecache, and quda::TuneKey::volume.

Referenced by saveProfile().

Here is the call graph for this function:

Here is the caller graph for this function:

serializeTrace()

static void quda::serializeTrace ( std::ostream &  out )

static

Serialize trace to an ostream, useful for writing to a file or sending to other nodes.

Definition at line 241 of file tune.cpp.

References quda::TuneKey::aux, it, quda::TuneKey::name, out, strcmp(), strncpy(), tmp, trace_list, and quda::TuneKey::volume.

Referenced by saveProfile().

Here is the call graph for this function:

Here is the caller graph for this function:

serializeTuneCache()

static void quda::serializeTuneCache ( std::ostream &  out )

static

Serialize tunecache to an ostream, useful for writing to a file or sending to other nodes.

Definition at line 154 of file tune.cpp.

References quda::TuneKey::aux, entry, quda::TuneKey::name, out, param, tunecache, and quda::TuneKey::volume.

Referenced by broadcastTuneCache(), and saveTuneCache().

Here is the caller graph for this function:

setDiracParam()

void quda::setDiracParam	(	DiracParam &	diracParam,
		QudaInvertParam *	inv_param,
		bool	pc
	)

Definition at line 1386 of file interface_quda.cpp.

References quda::GaugeField::Anisotropy(), quda::DiracParam::b_5, QudaInvertParam_s::b_5, quda::DiracParam::c_5, QudaInvertParam_s::c_5, quda::DiracParam::clover, cloverPrecise, quda::DiracParam::commDim, quda::DiracParam::dagger, QudaInvertParam_s::dagger, QudaInvertParam_s::dirac_order, QudaInvertParam_s::dslash_type, quda::DiracParam::epsilon, QudaInvertParam_s::epsilon, errorQuda, quda::DiracParam::fatGauge, quda::DiracParam::gauge, gaugeFatPrecise, gaugeLongPrecise, gaugePrecise, fused_exterior_ndeg_tm_dslash_cuda_gen::i, inv_param, quda::DiracParam::kappa, QudaInvertParam_s::kappa, kappa, quda::DiracParam::longGauge, quda::DiracParam::Ls, QudaInvertParam_s::Ls, quda::DiracParam::m5, QudaInvertParam_s::m5, quda::DiracParam::mass, QudaInvertParam_s::mass, QudaInvertParam_s::matpc_type, quda::DiracParam::matpcType, memcpy(), quda::DiracParam::mu, QudaInvertParam_s::mu, QUDA_ASQTAD_DIRAC, QUDA_ASQTAD_DSLASH, QUDA_ASQTADPC_DIRAC, QUDA_CLOVER_DIRAC, QUDA_CLOVER_WILSON_DSLASH, QUDA_CLOVERPC_DIRAC, QUDA_COVDEV_DSLASH, QUDA_CPS_WILSON_DIRAC_ORDER, QUDA_DOMAIN_WALL_4D_DSLASH, QUDA_DOMAIN_WALL_4DPC_DIRAC, QUDA_DOMAIN_WALL_DIRAC, QUDA_DOMAIN_WALL_DSLASH, QUDA_DOMAIN_WALLPC_DIRAC, QUDA_GAUGE_COVDEV_DIRAC, QUDA_GAUGE_LAPLACE_DIRAC, QUDA_GAUGE_LAPLACEPC_DIRAC, QUDA_LAPLACE_DSLASH, QUDA_MAX_DWF_LS, QUDA_MOBIUS_DOMAIN_WALL_DIRAC, QUDA_MOBIUS_DOMAIN_WALLPC_DIRAC, QUDA_MOBIUS_DWF_DSLASH, QUDA_STAGGERED_DIRAC, QUDA_STAGGERED_DSLASH, QUDA_STAGGEREDPC_DIRAC, QUDA_TWIST_NONDEG_DOUBLET, QUDA_TWIST_SINGLET, QUDA_TWISTED_CLOVER_DIRAC, QUDA_TWISTED_CLOVER_DSLASH, QUDA_TWISTED_CLOVERPC_DIRAC, QUDA_TWISTED_MASS_DIRAC, QUDA_TWISTED_MASS_DSLASH, QUDA_TWISTED_MASSPC_DIRAC, QUDA_WILSON_DIRAC, QUDA_WILSON_DSLASH, QUDA_WILSONPC_DIRAC, QudaInvertParam_s::twist_flavor, and quda::DiracParam::type.

Referenced by cloverQuda(), computeCloverForceQuda(), computeStaggeredForceQuda(), createDirac(), quda::deflated_solver::deflated_solver(), dslashQuda(), dslashQuda_4dpc(), dslashQuda_mdwf(), init(), lanczosQuda(), MatDagMatQuda(), MatQuda(), setDiracPreParam(), and setDiracSloppyParam().

Here is the call graph for this function:

Here is the caller graph for this function:

setDiracPreParam()

void quda::setDiracPreParam	(	DiracParam &	diracParam,
		QudaInvertParam *	inv_param,
		const bool	pc,
		bool	comms
	)

Definition at line 1485 of file interface_quda.cpp.

References quda::DiracParam::clover, cloverPrecondition, quda::DiracParam::commDim, QudaInvertParam_s::dslash_type, QudaInvertParam_s::dslash_type_precondition, quda::DiracParam::fatGauge, quda::DiracParam::gauge, gaugeExtended, gaugeFatExtended, gaugeFatPrecondition, gaugeLongExtended, gaugeLongPrecondition, gaugePrecondition, fused_exterior_ndeg_tm_dslash_cuda_gen::i, inv_param, QudaInvertParam_s::inv_type, quda::DiracParam::longGauge, QudaInvertParam_s::overlap, QUDA_ASQTAD_DSLASH, QUDA_PCG_INVERTER, QUDA_STAGGERED_DIRAC, QUDA_STAGGERED_DSLASH, QUDA_STAGGEREDPC_DIRAC, setDiracParam(), and quda::DiracParam::type.

Referenced by createDirac(), quda::multigrid_solver::multigrid_solver(), and updateMultigridQuda().

Here is the call graph for this function:

Here is the caller graph for this function:

setDiracSloppyParam()

void quda::setDiracSloppyParam	(	DiracParam &	diracParam,
		QudaInvertParam *	inv_param,
		bool	pc
	)

Definition at line 1469 of file interface_quda.cpp.

References quda::DiracParam::clover, cloverSloppy, quda::DiracParam::commDim, quda::DiracParam::fatGauge, quda::DiracParam::gauge, gaugeFatSloppy, gaugeLongSloppy, gaugeSloppy, fused_exterior_ndeg_tm_dslash_cuda_gen::i, inv_param, quda::DiracParam::longGauge, and setDiracParam().

Referenced by createDirac(), quda::deflated_solver::deflated_solver(), quda::multigrid_solver::multigrid_solver(), and updateMultigridQuda().

Here is the call graph for this function:

Here is the caller graph for this function:

setIdentity() [1/3]

template<class T , int N>

__device__ __host__ void quda::setIdentity ( Matrix< T, N > *  m )

inline

Definition at line 543 of file quda_matrix.h.

References fused_exterior_ndeg_tm_dslash_cuda_gen::i.

Referenced by bdSVD(), computeOvrImpSTOUTStep(), constructHHMat(), exponentiate_iQ(), getRealBidiagMatrix(), and smallSVD().

Here is the caller graph for this function:

setIdentity() [2/3]

template<int N>

__device__ __host__ void quda::setIdentity ( Matrix< float2, N > *  m )

inline

Definition at line 559 of file quda_matrix.h.

References fused_exterior_ndeg_tm_dslash_cuda_gen::i.

setIdentity() [3/3]

template<int N>

__device__ __host__ void quda::setIdentity ( Matrix< double2, N > *  m )

inline

Definition at line 575 of file quda_matrix.h.

References fused_exterior_ndeg_tm_dslash_cuda_gen::i.

setKernelPackT()

void quda::setKernelPackT

(

bool

pack

)

Parameters

pack	Sets whether to use a kernel to pack the T dimension

Definition at line 59 of file dslash_quda.cu.

References kernelPackT.

Referenced by anonymous_namespace{dslash_policy.cuh}::DslashPolicyTune::apply(), anonymous_namespace{dslash_policy.cuh}::DslashPolicyTune::DslashPolicyTune(), dslashQuda(), dslashQuda_4dpc(), dslashQuda_mdwf(), quda::cudaColorSpinorField::exchangeGhost(), init(), invertMultiShiftQuda(), invertMultiSrcQuda(), invertQuda(), lanczosQuda(), MatDagMatQuda(), MatQuda(), set_kernel_pack_t_(), and twistedMassDslashCuda().

Here is the caller graph for this function:

setPackComms()

void quda::setPackComms

(

const int *

commDim

)

Sets commDim array used in dslash_pack.cu

Definition at line 41 of file dslash_pack.cu.

Referenced by DslashCuda::DslashCuda().

Here is the caller graph for this function:

setPolicyTuning()

void quda::setPolicyTuning

(

bool

policy_tuning_

)

Definition at line 457 of file tune.cpp.

References policy_tuning.

Referenced by quda::DslashCoarsePolicyTune::DslashCoarsePolicyTune(), anonymous_namespace{dslash_policy.cuh}::DslashPolicyTune::DslashPolicyTune(), quda::blas::TileSizeTune< ReducerDiagonal, writeDiagonal, ReducerOffDiagonal, writeOffDiagonal >::TileSizeTune(), quda::DslashCoarsePolicyTune::~DslashCoarsePolicyTune(), anonymous_namespace{dslash_policy.cuh}::DslashPolicyTune::~DslashPolicyTune(), and quda::blas::TileSizeTune< ReducerDiagonal, writeDiagonal, ReducerOffDiagonal, writeOffDiagonal >::~TileSizeTune().

Here is the caller graph for this function:

setTransferGPU()

void quda::setTransferGPU

(

bool

)

setUnitarizeLinksConstants()

void quda::setUnitarizeLinksConstants	(	double	unitarize_eps,
		double	max_error,
		bool	allow_svd,
		bool	svd_only,
		double	svd_rel_error,
		double	svd_abs_error
	)

Referenced by computeKSLinkQuda(), GaugeAlgTest::SetReunitarizationConsts(), setReunitarizationConsts(), and unitarize_link_test().

Here is the caller graph for this function:

setZero() [1/3]

template<class T , int N>

__device__ __host__ void quda::setZero ( Matrix< T, N > *  m )

inline

Definition at line 592 of file quda_matrix.h.

References fused_exterior_ndeg_tm_dslash_cuda_gen::i.

Referenced by computeStapleRectangle(), and exponentiate_iQ().

Here is the caller graph for this function:

setZero() [2/3]

template<int N>

__device__ __host__ void quda::setZero ( Matrix< float2, N > *  m )

inline

Definition at line 607 of file quda_matrix.h.

References fused_exterior_ndeg_tm_dslash_cuda_gen::i.

setZero() [3/3]

template<int N>

__device__ __host__ void quda::setZero ( Matrix< double2, N > *  m )

inline

Definition at line 622 of file quda_matrix.h.

References fused_exterior_ndeg_tm_dslash_cuda_gen::i.

shiftColorSpinorField()

void quda::shiftColorSpinorField	(	cudaColorSpinorField &	dst,
		const cudaColorSpinorField &	src,
		const unsigned int	parity,
		const unsigned int	dim,
		const int	shift
	)

Definition at line 207 of file shift_quark_field.cu.

References arg(), commDimPartitioned(), deg_tm_dslash_cuda_gen::dagger, dim, errorQuda, quda::ColorSpinorField::Even(), dslash::gatherEnd, quda::ColorSpinorField::Odd(), offset, dslash::packEnd, parity, quda::LatticeField::Precision(), QUDA_BACKWARDS, QUDA_CPU_FIELD_LOCATION, QUDA_DOUBLE_PRECISION, QUDA_FORWARDS, QUDA_FULL_SITE_SUBSET, QUDA_SINGLE_PRECISION, qudaEventRecord(), qudaStreamWaitEvent(), dslash::scatterEnd, shift, quda::ColorSpinorField::SiteSubset(), src, and streams.

Here is the call graph for this function:

shiftColorSpinorFieldExternalKernel()

template<typename FloatN , int N, typename Output , typename Input >

__global__ void quda::shiftColorSpinorFieldExternalKernel

(

ShiftQuarkArg< Output, Input >

arg

)

Definition at line 93 of file shift_quark_field.cu.

References arg(), blockDim, coord, gridDim, idx, and x.

Here is the call graph for this function:

shiftColorSpinorFieldKernel()

template<typename FloatN , int N, typename Output , typename Input >

__global__ void quda::shiftColorSpinorFieldKernel

(

ShiftQuarkArg< Output, Input >

arg

)

Definition at line 68 of file shift_quark_field.cu.

References arg(), blockDim, gridDim, idx, neighborIndex(), shift, and x.

Here is the call graph for this function:

sin() [1/4]

template<typename ValueType >

__host__ __device__ ValueType quda::sin ( ValueType  x )

inline

Definition at line 40 of file complex_quda.h.

References sin(), and x.

Referenced by cos(), cosh(), exponentiate_iQ(), genericSource(), genGauss(), new_load_half(), polar(), polarSu3(), quda::Trig< isHalf, T >::Sin(), sin(), quda::Trig< isHalf, T >::SinCos(), sinh(), and tan().

Here is the call graph for this function:

Here is the caller graph for this function:

sin() [2/4]

template<class P >

void quda::sin	(	P &	p,
		int	d,
		int	n,
		int	offset
	)

Insert a sinusoidal wave sin ( n * (x[d] / X[d]) * pi ) in dimension d

Definition at line 55 of file color_spinor_util.cu.

References c, coord, d, double, getCoords(), mode, n, offset, p, parity, s, sin(), and X.

Here is the call graph for this function:

sin() [3/4]

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::sin ( const complex< ValueType > &  z )

inline

Definition at line 1018 of file complex_quda.h.

References cos(), cosh(), sin(), sinh(), and z.

Here is the call graph for this function:

sin() [4/4]

template<>

__host__ __device__ complex<float> quda::sin ( const complex< float > &  z )

inline

Definition at line 1026 of file complex_quda.h.

References cosf(), coshf(), sinf(), sinhf(), and z.

Referenced by sin().

Here is the call graph for this function:

Here is the caller graph for this function:

sinh() [1/3]

template<typename ValueType >

__host__ __device__ ValueType quda::sinh ( ValueType  x )

inline

Definition at line 75 of file complex_quda.h.

References sinh(), and x.

Referenced by cos(), cosh(), sin(), and sinh().

Here is the call graph for this function:

Here is the caller graph for this function:

sinh() [2/3]

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::sinh ( const complex< ValueType > &  z )

inline

Definition at line 1034 of file complex_quda.h.

References cos(), cosh(), sin(), sinh(), and z.

Here is the call graph for this function:

sinh() [3/3]

template<>

__host__ __device__ complex<float> quda::sinh ( const complex< float > &  z )

inline

Definition at line 1042 of file complex_quda.h.

References cosf(), coshf(), sinf(), sinhf(), and z.

Referenced by sinh().

Here is the call graph for this function:

Here is the caller graph for this function:

siteChecksum()

template<typename Arg >

__device__ __host__ uint64_t quda::siteChecksum ( const Arg &  arg, int  d, int  parity, int  x_cb  )

inline

Definition at line 17 of file checksum.cu.

References arg(), quda::Matrix< T, N >::checksum(), d, nColor, and parity.

Referenced by ChecksumCPU().

Here is the call graph for this function:

Here is the caller graph for this function:

solve()

void quda::solve	(	Complex *	psi,
		std::vector< ColorSpinorField *> &	p,
		std::vector< ColorSpinorField *> &	q,
		ColorSpinorField &	b
	)

Solve the equation A p_k psi_k = b by minimizing the residual and using Gaussian elimination.

Parameters

psi[out]	Array of coefficients
p[in]	Search direction vectors
q[in]	Search direction vectors with the operator applied

Definition at line 64 of file inv_mre.cpp.

References abs(), b, quda::blas::cDotProduct(), conj(), fused_exterior_ndeg_tm_dslash_cuda_gen::i, and p.

Referenced by invertMultiSrcQuda(), invertQuda(), and quda::MinResExt::operator()().

Here is the call graph for this function:

Here is the caller graph for this function:

spinorGauss() [1/2]

void quda::spinorGauss	(	ColorSpinorField &	src,
		int	seed
	)

Definition at line 149 of file spinor_gauss.cu.

References quda::RNG::Init(), quda::RNG::Release(), and src.

Here is the call graph for this function:

spinorGauss() [2/2]

void quda::spinorGauss	(	ColorSpinorField &	src,
		RNG &	randstates
	)

Definition at line 126 of file spinor_gauss.cu.

References errorQuda, QUDA_DOUBLE_PRECISION, QUDA_SINGLE_PRECISION, and src.

sqrt() [1/3]

template<typename ValueType >

__host__ __device__ ValueType quda::sqrt ( ValueType  x )

inline

Definition at line 105 of file complex_quda.h.

References sqrt(), and x.

Referenced by acosh(), asinh(), quda::linalg::Cholesky< Mat, T, N, fast >::Cholesky(), quda::IncEigCG::eigCGsolve(), exponentiate_iQ(), quda::GMResDR::FlexArnoldiProcedure(), quda::MG::generateNullVectors(), genGauss(), quda::Deflation::increment(), invertMultiShiftQuda(), invertMultiSrcQuda(), invertQuda(), l2(), new_save_half(), quda::Lanczos::operator()(), quda::Deflation::operator()(), quda::CG::operator()(), quda::CGNR::operator()(), quda::MPCG::operator()(), quda::PreconCG::operator()(), quda::BiCGstab::operator()(), quda::SimpleBiCGstab::operator()(), quda::MPBiCGstab::operator()(), quda::BiCGstabL::operator()(), quda::GCR::operator()(), quda::MR::operator()(), quda::SD::operator()(), quda::MultiShiftCG::operator()(), quda::MinResExt::operator()(), quda::IncEigCG::operator()(), quda::GMResDR::operator()(), quda::Solver::PrintStats(), quda::Solver::PrintSummary(), quda::Deflation::reduce(), reliable(), quda::BiCGstabL::reliable(), quda::GMResDR::RestartVZH(), quda::CG::solve(), sqrt(), test(), quda::gauge::Reconstruct< 8, Float >::Unpack(), quda::Deflation::verify(), and quda::MG::verify().

Here is the call graph for this function:

Here is the caller graph for this function:

sqrt() [2/3]

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::sqrt ( const complex< ValueType > &  z )

inline

Definition at line 1050 of file complex_quda.h.

References abs(), arg(), polar(), sqrt(), and z.

Here is the call graph for this function:

sqrt() [3/3]

template<typename ValueType >

__host__ __device__ complex<float> quda::sqrt ( const complex< float > &  z )

inline

Definition at line 1056 of file complex_quda.h.

References abs(), arg(), polar(), sqrtf(), and z.

Referenced by sqrt().

Here is the call graph for this function:

Here is the caller graph for this function:

staggeredDslashCuda()

void quda::staggeredDslashCuda	(	cudaColorSpinorField *	out,
		const cudaGaugeField &	gauge,
		const cudaColorSpinorField *	in,
		const int	parity,
		const int	dagger,
		const cudaColorSpinorField *	x,
		const double &	k,
		const int *	commDim,
		TimeProfile &	profile
	)

Definition at line 152 of file dslash_staggered.cu.

References deg_tm_dslash_cuda_gen::dagger, deg_tm_dslash_cuda_gen::dslash, errorQuda, fused_exterior_ndeg_tm_dslash_cuda_gen::i, in, out, parity, QUDA_DOUBLE_PRECISION, QUDA_HALF_PRECISION, QUDA_MAX_DIM, QUDA_SINGLE_PRECISION, and x.

Referenced by quda::DiracStaggered::Dslash(), and quda::DiracStaggered::DslashXpay().

Here is the caller graph for this function:

store_streaming_double2()

__device__ void quda::store_streaming_double2 ( double2 *  addr, double  x, double  y  )

inline

Definition at line 49 of file inline_ptx.h.

References __PTR, x, and y.

Referenced by vector_store().

Here is the caller graph for this function:

store_streaming_float2()

__device__ void quda::store_streaming_float2 ( float2 *  addr, float  x, float  y  )

inline

Definition at line 54 of file inline_ptx.h.

References __PTR, x, and y.

Referenced by vector_store().

Here is the caller graph for this function:

store_streaming_float4()

__device__ void quda::store_streaming_float4 ( float4 *  addr, float  x, float  y, float  z, float  w  )

inline

Definition at line 39 of file inline_ptx.h.

References __PTR, w, x, y, and z.

Referenced by vector_store().

Here is the caller graph for this function:

store_streaming_short2()

__device__ void quda::store_streaming_short2 ( short2 *  addr, short  x, short  y  )

inline

Definition at line 59 of file inline_ptx.h.

References __PTR, x, and y.

Referenced by vector_store().

Here is the caller graph for this function:

store_streaming_short4()

__device__ void quda::store_streaming_short4 ( short4 *  addr, short  x, short  y, short  z, short  w  )

inline

Definition at line 44 of file inline_ptx.h.

References __PTR, w, x, y, and z.

Referenced by vector_store().

Here is the caller graph for this function:

STOUTStep()

void quda::STOUTStep	(	GaugeField &	dataDs,
		const GaugeField &	dataOr,
		double	rho
	)

Apply STOUT smearing to the gauge field

Parameters

dataDs	Output smeared field
dataOr	Input gauge field
rho	smearing parameter

Definition at line 300 of file gauge_stout.cu.

References errorQuda, float, quda::GaugeField::isNative(), quda::GaugeField::Order(), quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, QUDA_HALF_PRECISION, QUDA_SINGLE_PRECISION, and quda::GaugeField::Reconstruct().

Referenced by performSTOUTnStep().

Here is the call graph for this function:

Here is the caller graph for this function:

str_end()

constexpr const char* quda::str_end ( const char *  str )

inline

Definition at line 45 of file malloc_quda.h.

Referenced by file_name().

Here is the caller graph for this function:

str_slant()

constexpr bool quda::str_slant ( const char *  str )

inline

Definition at line 46 of file malloc_quda.h.

Referenced by file_name().

Here is the caller graph for this function:

SubTraceUnit()

template<class T >

__device__ __host__ void quda::SubTraceUnit ( Matrix< T, 3 > &  a )

inline

Definition at line 1015 of file quda_matrix.h.

References a.

tan() [1/2]

template<typename ValueType >

__host__ __device__ ValueType quda::tan ( ValueType  x )

inline

Definition at line 45 of file complex_quda.h.

References tan(), and x.

Here is the call graph for this function:

tan() [2/2]

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::tan ( const complex< ValueType > &  z )

inline

Definition at line 1062 of file complex_quda.h.

References cos(), sin(), and z.

Referenced by tan().

Here is the call graph for this function:

Here is the caller graph for this function:

tanh() [1/2]

template<typename ValueType >

__host__ __device__ ValueType quda::tanh ( ValueType  x )

inline

Definition at line 80 of file complex_quda.h.

References tanh(), and x.

Here is the call graph for this function:

tanh() [2/2]

template<typename ValueType >

__host__ __device__ complex< ValueType > quda::tanh ( const complex< ValueType > &  z )

inline

Definition at line 1068 of file complex_quda.h.

References exp(), and z.

Referenced by tanh().

Here is the call graph for this function:

Here is the caller graph for this function:

timeInterval()

double quda::timeInterval	(	struct timeval	start,
		struct timeval	end
	)

Definition at line 18 of file inv_gcr_quda.cpp.

References end, and start.

traceEnabled()

bool quda::traceEnabled

(

)

Definition at line 75 of file tune.cpp.

References enable_trace, getenv(), quda::blas::init(), and strcmp().

Referenced by saveProfile(), and tuneLaunch().

Here is the call graph for this function:

Here is the caller graph for this function:

track_free()

static void quda::track_free ( const AllocType &  type, void *  ptr  )

static

Definition at line 119 of file malloc.cpp.

References alloc, DEVICE, MAPPED, PINNED, ptr, size, total_bytes, total_host_bytes, and total_pinned_bytes.

Referenced by device_free_(), device_pinned_free_(), and host_free_().

Here is the caller graph for this function:

track_malloc()

static void quda::track_malloc ( const AllocType &  type, const MemAlloc &  a, void *  ptr  )

static

Definition at line 97 of file malloc.cpp.

References a, alloc, DEVICE, MAPPED, max_total_bytes, max_total_host_bytes, max_total_pinned_bytes, PINNED, ptr, total_bytes, total_host_bytes, and total_pinned_bytes.

Referenced by device_malloc_(), device_pinned_malloc_(), mapped_malloc_(), pinned_malloc_(), and safe_malloc_().

Here is the caller graph for this function:

tuneLaunch()

TuneParam & quda::tuneLaunch	(	Tunable &	tunable,
		QudaTune	enabled,
		QudaVerbosity	verbosity
	)

Return the optimal launch parameters for a given kernel, either by retrieving them from tunecache or autotuning on the spot.

Definition at line 603 of file tune.cpp.

References quda::Tunable::advanceTuneParam(), quda::Tunable::apply(), quda::TuneKey::aux, broadcastTuneCache(), quda::Tunable::checkLaunchParam(), comm_rank(), quda::TuneParam::comment, commGlobalReduction(), ctime(), quda::Tunable::defaultTuneParam(), end, errorQuda, fused_exterior_ndeg_tm_dslash_cuda_gen::i, quda::Tunable::initTuneParam(), it, last_key, launchTimer, quda::TuneKey::name, param, quda::Tunable::paramString(), quda::Tunable::perfString(), policyTuning(), quda::Tunable::postTune(), quda::Tunable::preTune(), printfQuda, profile_count, QUDA_DEBUG_VERBOSE, QUDA_PROFILE_COMPUTE, QUDA_PROFILE_EPILOGUE, QUDA_PROFILE_INIT, QUDA_PROFILE_PREAMBLE, QUDA_PROFILE_TOTAL, QUDA_TUNE_NO, QUDA_TUNE_YES, QUDA_VERBOSE, start, quda::TuneParam::time, time(), trace_list, traceEnabled(), tunecache, quda::Tunable::tuneKey(), tuning, quda::Tunable::tuningIter(), verbosity, and quda::TuneKey::volume.

Referenced by quda::CopySpinor< FloatOut, FloatIn, Ns, Nc, OutOrder, InOrder >::apply(), quda::QudaMemCopy::apply(), quda::blas::copy_ns::CopyCuda< FloatN, N, Output, Input >::apply(), quda::GaussSpinor< FloatIn, Ns, Nc, InOrder >::apply(), BlasCuda< FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, Functor >::apply(), quda::CopyGaugeEx< FloatOut, FloatIn, length, OutOrder, InOrder >::apply(), quda::GenericPackGhostLauncher< Float, Ns, Ms, Nc, Mc, Arg >::apply(), ReduceCuda< doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Reducer >::apply(), quda::ShiftColorSpinorField< Output, Input >::apply(), quda::CopyColorSpinor< FloatOut, FloatIn, Ns, Nc, Arg >::apply(), quda::WuppertalSmearing< Float, Ns, Nc, Arg >::apply(), MultiBlasCuda< NXZ, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, Functor >::apply(), quda::Laplace< Float, nDim, nColor, Arg >::apply(), quda::ExtractGhost< Float, length, nDim, Order >::apply(), quda::ExtractGhostEx< Float, length, nDim, dim, Order >::apply(), quda::CopyGauge< FloatOut, FloatIn, length, OutOrder, InOrder, isGhost >::apply(), quda::CopyColorSpinor< FloatOut, FloatIn, 4, Nc, Arg >::apply(), quda::Gamma< ValueType, basis, dir >::apply(), quda::CopySpinorEx< FloatOut, FloatIn, Ns, Nc, OutOrder, InOrder, Basis, extend >::apply(), MultiReduceCuda< NXZ, doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, Reducer >::apply(), quda::TwistGamma< Float, nColor, Arg >::apply(), quda::blas::TileSizeTune< ReducerDiagonal, writeDiagonal, ReducerOffDiagonal, writeOffDiagonal >::apply(), quda::Clover< Float, nSpin, nColor, Arg >::apply(), quda::ProjectSU3< Float, G >::apply(), quda::TwistClover< Float, nSpin, nColor, Arg >::apply(), quda::GaugeOvrImpSTOUT< Float, GaugeOr, GaugeDs >::apply(), quda::DslashCoarsePolicyTune::apply(), quda::CalculateY< from_coarse, Float, fineSpin, fineColor, coarseSpin, coarseColor, Arg >::apply(), quda::CalculateYhat< Float, n, Arg >::apply(), and anonymous_namespace{dslash_policy.cuh}::DslashPolicyTune::apply().

Here is the call graph for this function:

Here is the caller graph for this function:

twistCloverApply()

template<bool inverse, typename Float , int nSpin, int nColor, typename Arg >

__device__ __host__ void quda::twistCloverApply ( Arg &  arg, int  x_cb, int  parity  )

inline

Definition at line 604 of file dslash_quda.cu.

References arg(), in, Mat(), nColor, out, and parity.

Here is the call graph for this function:

twistCloverCPU()

template<bool inverse, typename Float , int nSpin, int nColor, typename Arg >

void quda::twistCloverCPU

(

Arg &

arg

)

Definition at line 648 of file dslash_quda.cu.

References arg(), for(), and parity.

Here is the call graph for this function:

twistCloverGPU()

template<bool inverse, typename Float , int nSpin, int nColor, typename Arg >

__global__ void quda::twistCloverGPU

(

Arg

arg

)

Definition at line 656 of file dslash_quda.cu.

References arg(), blockDim, if(), and parity.

Here is the call graph for this function:

twistedCloverDslashCuda()

void quda::twistedCloverDslashCuda	(	cudaColorSpinorField *	out,
		const cudaGaugeField &	gauge,
		const FullClover *	clover,
		const FullClover *	cloverInv,
		const cudaColorSpinorField *	in,
		const int	parity,
		const int	dagger,
		const cudaColorSpinorField *	x,
		const QudaTwistCloverDslashType	type,
		const double &	kappa,
		const double &	mu,
		const double &	epsilon,
		const double &	k,
		const int *	commDim,
		TimeProfile &	profile
	)

Definition at line 215 of file dslash_twisted_clover.cu.

References dslash_cuda_gen::clover, deg_tm_dslash_cuda_gen::dagger, deg_tm_dslash_cuda_gen::dslash, errorQuda, fused_exterior_ndeg_tm_dslash_cuda_gen::i, in, kappa, mu, out, parity, QUDA_DOUBLE_PRECISION, QUDA_HALF_PRECISION, QUDA_SINGLE_PRECISION, QUDA_TWIST_SINGLET, and x.

Referenced by quda::DiracTwistedCloverPC::Dslash(), quda::DiracTwistedCloverPC::DslashXpay(), quda::DiracTwistedClover::M(), and quda::DiracTwistedCloverPC::M().

Here is the caller graph for this function:

twistedMassDslashCuda()

void quda::twistedMassDslashCuda	(	cudaColorSpinorField *	out,
		const cudaGaugeField &	gauge,
		const cudaColorSpinorField *	in,
		const int	parity,
		const int	dagger,
		const cudaColorSpinorField *	x,
		const QudaTwistDslashType	type,
		const double &	kappa,
		const double &	mu,
		const double &	epsilon,
		const double &	k,
		const int *	commDim,
		TimeProfile &	profile
	)

Definition at line 169 of file dslash_twisted_mass.cu.

References deg_tm_dslash_cuda_gen::dagger, deg_tm_dslash_cuda_gen::dslash, errorQuda, in, kappa, mu, out, parity, QUDA_DEG_TWIST_INV_DSLASH, QUDA_DOUBLE_PRECISION, QUDA_HALF_PRECISION, QUDA_SINGLE_PRECISION, setKernelPackT(), and x.

Referenced by quda::DiracTwistedMass::TwistedDslash(), and quda::DiracTwistedMass::TwistedDslashXpay().

Here is the call graph for this function:

Here is the caller graph for this function:

twistGammaCPU()

template<bool doublet, typename Float , int nColor, typename Arg >

void quda::twistGammaCPU

(

Arg

arg

)

Definition at line 300 of file dslash_quda.cu.

References arg(), in, and parity.

Here is the call graph for this function:

twistGammaGPU()

template<bool doublet, typename Float , int nColor, int d, typename Arg >

__global__ void quda::twistGammaGPU

(

Arg

arg

)

Definition at line 321 of file dslash_quda.cu.

References arg(), blockDim, d, in, and parity.

Here is the call graph for this function:

u32toa()

void quda::u32toa ( char *  buffer, uint32_t  value  )

inline

Definition at line 45 of file uint_to_char.h.

References a, b, c, gDigitsLut, fused_exterior_ndeg_tm_dslash_cuda_gen::i, and value.

Referenced by i32toa().

Here is the caller graph for this function:

u64toa()

void quda::u64toa ( char *  buffer, uint64_t  value  )

inline

Definition at line 127 of file uint_to_char.h.

References a, b, c, gDigitsLut, fused_exterior_ndeg_tm_dslash_cuda_gen::i, and value.

Referenced by i64toa(), quda::blas::TileSizeTune< ReducerDiagonal, writeDiagonal, ReducerOffDiagonal, writeOffDiagonal >::TileSizeTune(), and quda::QudaMemCopy::tuneKey().

Here is the caller graph for this function:

unitarizeLinks() [1/2]

void quda::unitarizeLinks	(	cudaGaugeField &	outfield,
		const cudaGaugeField &	infield,
		int *	fails
	)

Definition at line 495 of file unitarize_links_quda.cu.

References errorQuda, quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, and QUDA_SINGLE_PRECISION.

Referenced by GaugeAlgTest::CallUnitarizeLinks(), CallUnitarizeLinks(), computeKSLinkQuda(), unitarize_link_test(), and unitarizeLinks().

Here is the call graph for this function:

Here is the caller graph for this function:

unitarizeLinks() [2/2]

void quda::unitarizeLinks	(	cudaGaugeField &	outfield,
		int *	fails
	)

Definition at line 512 of file unitarize_links_quda.cu.

References links, and unitarizeLinks().

Here is the call graph for this function:

unitarizeLinksCPU()

void quda::unitarizeLinksCPU	(	cpuGaugeField &	outfield,
		const cpuGaugeField &	infield
	)

Referenced by TEST().

Here is the caller graph for this function:

updateAlphaZeta()

void quda::updateAlphaZeta	(	double *	alpha,
		double *	zeta,
		double *	zeta_old,
		const double *	r2,
		const double *	beta,
		const double	pAp,
		const double *	offset,
		const int	nShift,
		const int	j_low
	)

Compute the new values of alpha and zeta

Definition at line 127 of file inv_multi_cg_quda.cpp.

References offset, and QUDA_MAX_MULTI_SHIFT.

Referenced by quda::MultiShiftCG::operator()().

Here is the caller graph for this function:

updateAp()

void quda::updateAp	(	Complex **	beta,
		std::vector< ColorSpinorField *>	Ap,
		int	begin,
		int	size,
		int	k
	)

Definition at line 70 of file inv_gcr_quda.cpp.

References quda::blas::caxpy(), fused_exterior_ndeg_tm_dslash_cuda_gen::i, and size.

Referenced by orthoDir().

Here is the call graph for this function:

Here is the caller graph for this function:

updateGaugeField()

void quda::updateGaugeField	(	GaugeField &	out,
		double	dt,
		const GaugeField &	in,
		const GaugeField &	mom,
		bool	conj_mom,
		bool	exact
	)

Evolve the gauge field by step size dt using the momentuim field

Parameters

out	Updated gauge field
dt	Step size
in	Input gauge field
mom	Momentum field
conj_mom	Whether we conjugate the momentum in the exponential
exact	Calculate exact exponential or use an expansion

Definition at line 308 of file gauge_update_quda.cu.

References errorQuda, in, quda::LatticeField::Location(), out, quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, and QUDA_SINGLE_PRECISION.

Referenced by updateGaugeFieldQuda().

Here is the call graph for this function:

Here is the caller graph for this function:

updateMomentum()

void quda::updateMomentum	(	GaugeField &	mom,
		double	coeff,
		GaugeField &	force
	)

Update the momentum field from the force field

mom = mom - coeff * [force]_TA

where [A]_TA means the traceless anti-hermitian projection of A

Parameters

mom	Momentum field
force	Force field

Definition at line 224 of file momentum.cu.

References checkCudaError, dw_dslash_4D_cuda_gen::coeff(), errorQuda, quda::GaugeField::Order(), quda::LatticeField::Precision(), QUDA_DOUBLE_PRECISION, and QUDA_FLOAT2_GAUGE_ORDER.

Referenced by computeCloverForceQuda(), computeHISQForceQuda(), and computeStaggeredForceQuda().

Here is the call graph for this function:

Here is the caller graph for this function:

updateSolution()

void quda::updateSolution	(	ColorSpinorField &	x,
		const Complex *	alpha,
		Complex **const	beta,
		double *	gamma,
		int	k,
		std::vector< ColorSpinorField *>	p
	)

Definition at line 141 of file inv_gcr_quda.cpp.

References backSubs(), quda::blas::caxpy(), delta, gamma(), fused_exterior_ndeg_tm_dslash_cuda_gen::i, p, X, and x.

Referenced by quda::GCR::operator()().

Here is the call graph for this function:

Here is the caller graph for this function:

vector_load()

template<typename VectorType >

__device__ __host__ VectorType quda::vector_load ( void *  ptr, int  idx  )

inline

Definition at line 275 of file register_traits.h.

References idx, and ptr.

vector_store() [1/6]

template<typename VectorType >

__device__ __host__ void quda::vector_store ( void *  ptr, int  idx, const VectorType &  value  )

inline

Definition at line 285 of file register_traits.h.

References idx, ptr, and value.

Referenced by quda::clover::FloatNOrder< Float, length, N, huge_alloc >::save(), quda::colorspinor::FloatNOrder< Float, Ns, Nc, N, huge_alloc >::save(), quda::gauge::FloatNOrder< Float, length, N, reconLenParam, stag_phase, huge_alloc >::save(), quda::colorspinor::FloatNOrder< Float, Ns, Nc, N, huge_alloc >::saveGhost(), quda::gauge::FloatNOrder< Float, length, N, reconLenParam, stag_phase, huge_alloc >::saveGhost(), and quda::gauge::FloatNOrder< Float, length, N, reconLenParam, stag_phase, huge_alloc >::saveGhostEx().

Here is the caller graph for this function:

vector_store() [2/6]

template<>

__device__ __host__ void quda::vector_store ( void *  ptr, int  idx, const double2 &  value  )

inline

Definition at line 290 of file register_traits.h.

References idx, ptr, store_streaming_double2(), and value.

Here is the call graph for this function:

vector_store() [3/6]

template<>

__device__ __host__ void quda::vector_store ( void *  ptr, int  idx, const float4 &  value  )

inline

Definition at line 299 of file register_traits.h.

References idx, ptr, store_streaming_float4(), and value.

Here is the call graph for this function:

vector_store() [4/6]

template<>

__device__ __host__ void quda::vector_store ( void *  ptr, int  idx, const float2 &  value  )

inline

Definition at line 308 of file register_traits.h.

References idx, ptr, store_streaming_float2(), and value.

Here is the call graph for this function:

vector_store() [5/6]

template<>

__device__ __host__ void quda::vector_store ( void *  ptr, int  idx, const short4 &  value  )

inline

Definition at line 317 of file register_traits.h.

References idx, ptr, store_streaming_short4(), and value.

Here is the call graph for this function:

vector_store() [6/6]

template<>

__device__ __host__ void quda::vector_store ( void *  ptr, int  idx, const short2 &  value  )

inline

Definition at line 326 of file register_traits.h.

References idx, ptr, store_streaming_short2(), and value.

Here is the call graph for this function:

wilsonDslashCuda()

void quda::wilsonDslashCuda	(	cudaColorSpinorField *	out,
		const cudaGaugeField &	gauge,
		const cudaColorSpinorField *	in,
		const int	oddBit,
		const int	daggerBit,
		const cudaColorSpinorField *	x,
		const double &	k,
		const int *	commDim,
		TimeProfile &	profile
	)

Definition at line 108 of file dslash_wilson.cu.

References deg_tm_dslash_cuda_gen::dagger, deg_tm_dslash_cuda_gen::dslash, errorQuda, in, out, parity, QUDA_DOUBLE_PRECISION, QUDA_HALF_PRECISION, QUDA_SINGLE_PRECISION, and x.

Referenced by quda::DiracWilson::Dslash(), and quda::DiracWilson::DslashXpay().

Here is the caller graph for this function:

writeLinkVariableToArray() [1/2]

template<class T , class U >

__device__ void quda::writeLinkVariableToArray ( const Matrix< T, 3 > &  link, const int  dir, const int  idx, const int  stride, U *const  array  )

inline

Definition at line 816 of file quda_matrix.h.

References array, quda::Matrix< T, N >::data, fused_exterior_ndeg_tm_dslash_cuda_gen::i, and idx.

writeLinkVariableToArray() [2/2]

__device__ void quda::writeLinkVariableToArray ( const Matrix< complex< double >, 3 > &  link, const int  dir, const int  idx, const int  stride, float2 *const  array  )

inline

Definition at line 829 of file quda_matrix.h.

References array, fused_exterior_ndeg_tm_dslash_cuda_gen::i, and idx.

writeMatrixToArray()

template<class T , int N, class U >

__device__ void quda::writeMatrixToArray ( const Matrix< T, N > &  mat, const int  idx, const int  stride, U *const  array  )

inline

Definition at line 785 of file quda_matrix.h.

References array, fused_exterior_ndeg_tm_dslash_cuda_gen::i, idx, and mat().

Here is the call graph for this function:

writeMomentumToArray()

template<class T , class U >

__device__ void quda::writeMomentumToArray ( const Matrix< T, 3 > &  mom, const int  dir, const int  idx, const U  coeff, const int  stride, T *const  array  )

inline

Definition at line 881 of file quda_matrix.h.

References array, dw_dslash_4D_cuda_gen::coeff(), quda::Matrix< T, N >::data, and idx.

Here is the call graph for this function:

wuppertalStep() [1/2]

void quda::wuppertalStep	(	ColorSpinorField &	out,
		const ColorSpinorField &	in,
		int	parity,
		const GaugeField &	U,
		double	A,
		double	B
	)

Apply a generic Wuppertal smearing step Computes out(x) = A*in(x) + B* (U_{-}(x)in(x+mu) + U^(x-mu)in(x-mu))

Parameters

[out]	out	The out result field
[in]	in	The in spinor field
[in]	U	The gauge field
[in]	A	The scaling factor for in(x)
[in]	B	The scaling factor for (U_{-}(x)in(x+mu) + U^(x-mu)in(x-mu))

Definition at line 189 of file color_spinor_wuppertal.cu.

References quda::WuppertalSmearing< Float, Ns, Nc, Arg >::apply(), arg(), in, out, and parity.

Referenced by performWuppertalnStep(), and wuppertalStep().

Here is the call graph for this function:

Here is the caller graph for this function:

wuppertalStep() [2/2]

void quda::wuppertalStep	(	ColorSpinorField &	out,
		const ColorSpinorField &	in,
		int	parity,
		const GaugeField &	U,
		double	alpha
	)

Apply a standard Wuppertal smearing step Computes out(x) = 1/(1+6*alpha)*(in(x) + alpha* (U_{-}(x)in(x+mu) + U^(x-mu)in(x-mu)))

Parameters

[out]	out	The out result field
[in]	in	The in spinor field
[in]	U	The gauge field
[in]	alpha	The smearing parameter

Definition at line 294 of file color_spinor_wuppertal.cu.

References in, out, parity, and wuppertalStep().

Here is the call graph for this function:

wuppertalStepCPU()

template<typename Float , int Ns, int Nc, typename Arg >

void quda::wuppertalStepCPU

(

Arg

arg

)

Definition at line 119 of file color_spinor_wuppertal.cu.

References arg(), for(), and parity.

Here is the call graph for this function:

wuppertalStepGPU()

template<typename Float , int Ns, int Nc, typename Arg >

__global__ void quda::wuppertalStepGPU

(

Arg

arg

)

Definition at line 135 of file color_spinor_wuppertal.cu.

References arg(), blockDim, and parity.

Here is the call graph for this function:

zero() [1/10]

__device__ __host__ void quda::zero ( double &  a )

inline

Definition at line 14 of file float_vector.h.

References a.

zero() [2/10]

__device__ __host__ void quda::zero ( double2 &  a )

inline

Definition at line 15 of file float_vector.h.

References a.

zero() [3/10]

__device__ __host__ void quda::zero ( double3 &  a )

inline

Definition at line 16 of file float_vector.h.

References a.

zero() [4/10]

__device__ __host__ void quda::zero ( double4 &  a )

inline

Definition at line 17 of file float_vector.h.

References a.

zero() [5/10]

__device__ __host__ void quda::zero ( float &  a )

inline

Definition at line 19 of file float_vector.h.

References a.

zero() [6/10]

__device__ __host__ void quda::zero ( float2 &  a )

inline

Definition at line 20 of file float_vector.h.

References a.

zero() [7/10]

__device__ __host__ void quda::zero ( float3 &  a )

inline

Definition at line 21 of file float_vector.h.

References a.

zero() [8/10]

__device__ __host__ void quda::zero ( float4 &  a )

inline

Definition at line 22 of file float_vector.h.

References a.

zero() [9/10]

template<typename T >

static void quda::zero ( T  d[], int  N  )

static

Definition at line 52 of file inv_mpcg_quda.cpp.

References fused_exterior_ndeg_tm_dslash_cuda_gen::i.

zero() [10/10]

template<typename scalar , int n>

__device__ __host__ void quda::zero ( vector_type< scalar, n > &  v )

inline

Definition at line 82 of file cub_helper.cuh.

References quda::vector_type< scalar, n >::data, fused_exterior_ndeg_tm_dslash_cuda_gen::i, and n.

Referenced by quda::clover::Accessor< Float, nColor, nSpin, QUDA_PACKED_CLOVER_ORDER >::Accessor(), quda::ShiftUpdate::apply(), applyThirdTerm(), quda::GMResDR::FlexArnoldiProcedure(), quda::MG::generateNullVectors(), quda::HMatrix< T, N >::HMatrix(), quda::MG::loadVectors(), quda::Matrix< T, N >::Matrix(), quda::MG::MG(), quda::Lanczos::operator()(), quda::clover::Accessor< Float, nColor, nSpin, QUDA_PACKED_CLOVER_ORDER >::operator()(), quda::MPCG::operator()(), quda::PreconCG::operator()(), quda::MPBiCGstab::operator()(), quda::SD::operator()(), quda::GMResDR::operator()(), reduce2d(), quda::vector_type< scalar, n >::vector_type(), quda::Deflation::verify(), and quda::MG::verify().

Here is the caller graph for this function:

Variable Documentation

alloc

std::map<void *, MemAlloc> quda::alloc[N_ALLOC_TYPE]

static

Definition at line 51 of file malloc.cpp.

Referenced by assertAllMemFree(), device_free_(), quda::gauge::Accessor< Float, nColor, QUDA_FLOAT2_GAUGE_ORDER >::device_norm2(), device_pinned_free_(), host_free_(), quda::colorspinor::FieldOrderCB< Float, nSpin, nColor, nVec, order >::norm2(), print_alloc(), track_free(), and track_malloc().

apiTimer

TimeProfile quda::apiTimer("CUDA API calls (driver)")

static

Referenced by printAPIProfile().

bidirectional_debug

bool quda::bidirectional_debug = false

static

Definition at line 7 of file coarse_op.cuh.

Referenced by calculateY().

complete_recv_back

bool quda::complete_recv_back[QUDA_MAX_DIM] = { }

static

Definition at line 1150 of file cuda_color_spinor_field.cu.

Referenced by quda::cudaColorSpinorField::commsQuery().

complete_recv_fwd

bool quda::complete_recv_fwd[QUDA_MAX_DIM] = { }

static

Definition at line 1149 of file cuda_color_spinor_field.cu.

Referenced by quda::cudaColorSpinorField::commsQuery().

complete_send_back

bool quda::complete_send_back[QUDA_MAX_DIM] = { }

static

Definition at line 1152 of file cuda_color_spinor_field.cu.

Referenced by quda::cudaColorSpinorField::commsQuery().

complete_send_fwd

bool quda::complete_send_fwd[QUDA_MAX_DIM] = { }

static

Definition at line 1151 of file cuda_color_spinor_field.cu.

Referenced by quda::cudaColorSpinorField::commsQuery().

config

int quda::config = 0

static

Definition at line 957 of file dslash_coarse.cu.

Referenced by quda::DslashCoarsePolicyTune::apply(), quda::DslashCoarsePolicyTune::defaultTuneParam(), quda::DslashCoarsePolicyTune::DslashCoarsePolicyTune(), and quda::DslashCoarsePolicyTune::initTuneParam().

count

__device__ unsigned int quda::count[QUDA_MAX_MULTI_REDUCE] = { }

Definition at line 118 of file cub_helper.cuh.

Referenced by quda::ShiftUpdate::apply(), quda::BiCGstabLUpdate::apply(), quda::blas::caxpy_recurse(), quda::blas::caxpyz_recurse(), device_free_(), device_pinned_free_(), host_free_(), quda::blas::multiReduce_recurse(), quda::TimeProfile::Print(), quda::TimeProfile::PrintGlobal(), qudaMemcpy_(), qudaMemcpyAsync_(), reduce2d(), reduceRow(), and saveProfile().

debug

bool quda::debug = false

static

Definition at line 11 of file multigrid.cpp.

Referenced by quda::MG::operator()().

dslash_init

bool quda::dslash_init = false

static

Definition at line 955 of file dslash_coarse.cu.

Referenced by quda::DslashCoarsePolicyTune::DslashCoarsePolicyTune().

enable_trace

bool quda::enable_trace = false

static

Definition at line 73 of file tune.cpp.

Referenced by traceEnabled().

gDigitsLut

const char quda::gDigitsLut[200]

static

Initial value:

= {
    '0','0','0','1','0','2','0','3','0','4','0','5','0','6','0','7','0','8','0','9',
    '1','0','1','1','1','2','1','3','1','4','1','5','1','6','1','7','1','8','1','9',
    '2','0','2','1','2','2','2','3','2','4','2','5','2','6','2','7','2','8','2','9',
    '3','0','3','1','3','2','3','3','3','4','3','5','3','6','3','7','3','8','3','9',
    '4','0','4','1','4','2','4','3','4','4','4','5','4','6','4','7','4','8','4','9',
    '5','0','5','1','5','2','5','3','5','4','5','5','5','6','5','7','5','8','5','9',
    '6','0','6','1','6','2','6','3','6','4','6','5','6','6','6','7','6','8','6','9',
    '7','0','7','1','7','2','7','3','7','4','7','5','7','6','7','7','7','8','7','9',
    '8','0','8','1','8','2','8','3','8','4','8','5','8','6','8','7','8','8','8','9',
    '9','0','9','1','9','2','9','3','9','4','9','5','9','6','9','7','9','8','9','9'
  }

Definition at line 32 of file uint_to_char.h.

Referenced by u32toa(), and u64toa().

initial_cache_size

size_t quda::initial_cache_size = 0

static

Definition at line 92 of file tune.cpp.

Referenced by loadTuneCache(), and saveTuneCache().

isLastBlockDone

__shared__ bool quda::isLastBlockDone

Definition at line 119 of file cub_helper.cuh.

Referenced by reduce2d(), and reduceRow().

isLastWarpDone

__shared__ volatile bool quda::isLastWarpDone[16]

Definition at line 166 of file cub_helper.cuh.

it

map::iterator quda::it

static

Definition at line 91 of file tune.cpp.

Referenced by quda::pool::device_malloc_(), quda::pool::flush_device(), quda::pool::flush_pinned(), quda::MPCG::operator()(), quda::MPBiCGstab::operator()(), quda::pool::pinned_malloc_(), serializeTrace(), and tuneLaunch().

kernelPackT

bool quda::kernelPackT = false

static

Definition at line 57 of file dslash_quda.cu.

Referenced by getKernelPackT(), and setKernelPackT().

last_key

TuneKey quda::last_key

static

Definition at line 24 of file tune.cpp.

Referenced by getLastTuneKey(), and tuneLaunch().

launchTimer

TimeProfile quda::launchTimer("tuneLaunch")

static

Referenced by printLaunchTimer(), and tuneLaunch().

max_eigcg_cycles

int quda::max_eigcg_cycles = 4

static

Definition at line 44 of file inv_eigcg_quda.cpp.

Referenced by quda::IncEigCG::operator()().

max_total_bytes

long quda::max_total_bytes[N_ALLOC_TYPE] = {0}

static

Definition at line 53 of file malloc.cpp.

Referenced by device_allocated_peak(), host_allocated_peak(), mapped_allocated_peak(), pinned_allocated_peak(), printPeakMemUsage(), and track_malloc().

max_total_host_bytes

long quda::max_total_host_bytes

static

Definition at line 54 of file malloc.cpp.

Referenced by printPeakMemUsage(), and track_malloc().

max_total_pinned_bytes

long quda::max_total_pinned_bytes

static

Definition at line 55 of file malloc.cpp.

Referenced by printPeakMemUsage(), and track_malloc().

Nstream

const int quda::Nstream = 9

Definition at line 330 of file quda_internal.h.

Referenced by ApplyClover(), ApplyGamma(), ApplyTwistClover(), ApplyTwistGamma(), anonymous_namespace{dslash_policy.cuh}::commsComplete(), anonymous_namespace{dslash_policy.cuh}::completeDslash(), contractCuda(), createDslashEvents(), destroyDslashEvents(), endQuda(), quda::blas::init(), initQudaMemory(), anonymous_namespace{dslash_policy.cuh}::DslashBasic::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashPthreads::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashFusedExterior::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashGDR::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashFusedGDR::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashGDRRecv::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashFusedGDRRecv::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashZeroCopyPack::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashFusedZeroCopyPack::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashZeroCopyPackGDRRecv::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashFusedZeroCopyPackGDRRecv::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashZeroCopy::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashFusedZeroCopy::operator()(), anonymous_namespace{dslash_policy.cuh}::DslashNC::operator()(), quda::cudaColorSpinorField::packExtended(), and quda::cudaColorSpinorField::zero().

pinned_allocator

auto quda::pinned_allocator = [] (size_t bytes ) { return static_cast<Complex*>(pool_pinned_malloc(bytes)); }

static

Definition at line 24 of file deflation.cpp.

Referenced by quda::Deflation::reduce(), and quda::Deflation::verify().

pinned_deleter

auto quda::pinned_deleter = [] (Complex *hptr) { pool_pinned_free(hptr); }

static

Definition at line 25 of file deflation.cpp.

Referenced by quda::Deflation::reduce(), and quda::Deflation::verify().

policy

std::vector<DslashCoarsePolicy> quda::policy

static

Definition at line 956 of file dslash_coarse.cu.

Referenced by quda::DslashCoarsePolicyTune::advanceAux(), quda::DslashCoarsePolicyTune::apply(), ApplyCoarse(), quda::DslashCoarsePolicyTune::DslashCoarsePolicyTune(), anonymous_namespace{dslash_policy.cuh}::DslashPolicyTune::DslashPolicyTune(), and quda::DslashCoarseLaunch::operator()().

policy_tuning

bool quda::policy_tuning = false

static

Definition at line 452 of file tune.cpp.

Referenced by policyTuning(), and setPolicyTuning().

profile_count

bool quda::profile_count = true

static

Definition at line 105 of file tune.cpp.

Referenced by disableProfileCount(), enableProfileCount(), and tuneLaunch().

quda_hash

const std::string quda::quda_hash = QUDA_HASH

static

Definition at line 88 of file tune.cpp.

Referenced by loadTuneCache(), saveProfile(), and saveTuneCache().

quda_version

const std::string quda::quda_version = STR(QUDA_VERSION_MAJOR) "." STR(QUDA_VERSION_MINOR) "." STR(QUDA_VERSION_SUBMINOR)

static

Definition at line 96 of file tune.cpp.

Referenced by initQudaDevice(), loadTuneCache(), saveProfile(), and saveTuneCache().

reorder_location_

QudaFieldLocation quda::reorder_location_ = QUDA_CUDA_FIELD_LOCATION

static

Definition at line 583 of file lattice_field.cpp.

Referenced by reorder_location(), and reorder_location_set().

resource_path

std::string quda::resource_path

static

Definition at line 89 of file tune.cpp.

Referenced by loadTuneCache(), saveProfile(), and saveTuneCache().

stream

cudaStream_t* quda::stream

Definition at line 898 of file cuda_color_spinor_field.cu.

Referenced by quda::CopySpinor< FloatOut, FloatIn, Ns, Nc, OutOrder, InOrder >::apply(), quda::blas::copy_ns::CopyCuda< FloatN, N, Output, Input >::apply(), quda::GaussSpinor< FloatIn, Ns, Nc, InOrder >::apply(), quda::CopyGaugeEx< FloatOut, FloatIn, length, OutOrder, InOrder >::apply(), quda::GenericPackGhostLauncher< Float, Ns, Ms, Nc, Mc, Arg >::apply(), quda::CopyColorSpinor< FloatOut, FloatIn, Ns, Nc, Arg >::apply(), quda::WuppertalSmearing< Float, Ns, Nc, Arg >::apply(), quda::Laplace< Float, nDim, nColor, Arg >::apply(), quda::ExtractGhost< Float, length, nDim, Order >::apply(), quda::ExtractGhostEx< Float, length, nDim, dim, Order >::apply(), quda::CopyGauge< FloatOut, FloatIn, length, OutOrder, InOrder, isGhost >::apply(), quda::CopyColorSpinor< FloatOut, FloatIn, 4, Nc, Arg >::apply(), quda::Gamma< ValueType, basis, dir >::apply(), quda::CopySpinorEx< FloatOut, FloatIn, Ns, Nc, OutOrder, InOrder, Basis, extend >::apply(), quda::TwistGamma< Float, nColor, Arg >::apply(), quda::Clover< Float, nSpin, nColor, Arg >::apply(), quda::ProjectSU3< Float, G >::apply(), quda::TwistClover< Float, nSpin, nColor, Arg >::apply(), quda::cudaColorSpinorField::gather(), quda::cudaColorSpinorField::pack(), quda::cudaColorSpinorField::packExtended(), quda::cudaColorSpinorField::packGhost(), quda::cudaColorSpinorField::packGhostExtended(), qudaEventRecord(), qudaLaunchKernel(), qudaMemcpy2DAsync_(), qudaMemcpyAsync_(), qudaStreamSynchronize(), qudaStreamWaitEvent(), quda::cudaColorSpinorField::scatter(), quda::cudaColorSpinorField::scatterExtended(), quda::cudaColorSpinorField::sendGhost(), quda::cudaColorSpinorField::sendStart(), quda::cudaColorSpinorField::streamInit(), quda::cudaColorSpinorField::unpackGhost(), and quda::cudaColorSpinorField::unpackGhostExtended().

total_bytes

long quda::total_bytes[N_ALLOC_TYPE] = {0}

static

Definition at line 52 of file malloc.cpp.

Referenced by track_free(), and track_malloc().

total_host_bytes

long quda::total_host_bytes

static

Definition at line 54 of file malloc.cpp.

Referenced by track_free(), and track_malloc().

total_pinned_bytes

long quda::total_pinned_bytes

static

Definition at line 55 of file malloc.cpp.

Referenced by track_free(), and track_malloc().

trace_list

std::list<TraceKey> quda::trace_list

static

Definition at line 72 of file tune.cpp.

Referenced by saveProfile(), serializeTrace(), and tuneLaunch().

tunecache

map quda::tunecache

static

Definition at line 90 of file tune.cpp.

Referenced by deserializeTuneCache(), flushProfile(), getTuneCache(), loadTuneCache(), saveProfile(), saveTuneCache(), serializeProfile(), serializeTuneCache(), and tuneLaunch().

tuning

bool quda::tuning = false

static

tuning in progress?

Definition at line 101 of file tune.cpp.

Referenced by activeTuning(), and tuneLaunch().

unscaled_shifts

double quda::unscaled_shifts[QUDA_MAX_MULTI_SHIFT]

static

Definition at line 1530 of file interface_quda.cpp.

Referenced by invertMultiShiftQuda(), and massRescale().