quda::Tunable Class Reference

#include <tune_quda.h>

Inheritance diagram for quda::Tunable:

Public Member Functions
	Tunable ()
virtual	~Tunable ()
virtual TuneKey	tuneKey () const =0
virtual void	apply (const cudaStream_t &stream)=0
virtual void	preTune ()
virtual void	postTune ()
virtual int	tuningIter () const
virtual std::string	paramString (const TuneParam &param) const
virtual std::string	perfString (float time) const
virtual void	initTuneParam (TuneParam &param) const
virtual void	defaultTuneParam (TuneParam &param) const
virtual bool	advanceTuneParam (TuneParam &param) const
void	checkLaunchParam (TuneParam &param)
CUresult	jitifyError () const
CUresult &	jitifyError ()

Protected Member Functions
virtual long long	flops () const =0
virtual long long	bytes () const
virtual unsigned int	sharedBytesPerThread () const =0
virtual unsigned int	sharedBytesPerBlock (const TuneParam &param) const =0
virtual unsigned int	minThreads () const
virtual bool	tuneGridDim () const
virtual bool	tuneAuxDim () const
virtual bool	tuneSharedBytes () const
virtual bool	advanceGridDim (TuneParam &param) const
virtual unsigned int	maxBlockSize (const TuneParam &param) const
virtual unsigned int	maxGridSize () const
virtual unsigned int	minGridSize () const
virtual int	gridStep () const
	gridStep sets the step size when iterating the grid size in advanceGridDim. More...
virtual int	blockStep () const
virtual int	blockMin () const
virtual void	resetBlockDim (TuneParam &param) const
virtual bool	advanceBlockDim (TuneParam &param) const
unsigned int	maxBlocksPerSM () const
	For some reason this can't be queried from the device properties, so here we set set this. Based on Table 14 of the CUDA Programming Guide 10.0 (Technical Specifications per Compute Capability) More...
template<typename F >
void	setMaxDynamicSharedBytesPerBlock (F *func) const
	Enable the maximum dynamic shared bytes for the kernel "func" (values given by maxDynamicSharedBytesPerBlock()). More...
unsigned int	maxDynamicSharedBytesPerBlock () const
	This can't be correctly queried in CUDA for all architectures so here we set set this. Based on Table 14 of the CUDA Programming Guide 10.0 (Technical Specifications per Compute Capability). More...
virtual unsigned int	maxSharedBytesPerBlock () const
	The maximum shared memory that a CUDA thread block can use in the autotuner. This isn't necessarily the same as maxDynamicSharedMemoryPerBlock since that may need explicit opt in to enable (by calling setMaxDynamicSharedBytes for the kernel in question). If the CUDA kernel in question does this opt in then this function can be overloaded to return maxDynamicSharedBytesPerBlock. More...
virtual bool	advanceSharedBytes (TuneParam &param) const
virtual bool	advanceAux (TuneParam &param) const
int	writeAuxString (const char *format,...)

Protected Attributes
char	aux [TuneKey::aux_n]
CUresult	jitify_error

Detailed Description

Definition at line 59 of file tune_quda.h.

Constructor & Destructor Documentation

Tunable()

quda::Tunable::Tunable ( )

inline

Definition at line 279 of file tune_quda.h.

~Tunable()

virtual quda::Tunable::~Tunable ( )

inlinevirtual

Definition at line 280 of file tune_quda.h.

References quda::stream.

Member Function Documentation

advanceAux()

virtual bool quda::Tunable::advanceAux ( TuneParam &  param ) const

inlineprotectedvirtual

Reimplemented in quda::dslash::DslashPolicyTune< Dslash >, quda::blas::TileSizeTune< ReducerDiagonal, writeDiagonal, ReducerOffDiagonal, writeOffDiagonal >, quda::DslashCoarsePolicyTune, quda::CalculateY< from_coarse, Float, fineSpin, fineColor, coarseSpin, coarseColor, Arg >, and quda::GenericPackGhostLauncher< Float, block_float, Ns, Ms, Nc, Mc, Arg >.

Definition at line 263 of file tune_quda.h.

advanceBlockDim()

virtual bool quda::Tunable::advanceBlockDim ( TuneParam &  param ) const

inlineprotectedvirtual

Reimplemented in quda::TunableVectorYZ, quda::TunableVectorY, quda::TunableLocalParity, quda::ShiftColorSpinorField< Output, Input >, and quda::GenericPackGhostLauncher< Float, block_float, Ns, Ms, Nc, Mc, Arg >.

Definition at line 124 of file tune_quda.h.

References quda::TuneParam::block, and quda::TuneParam::grid.

Referenced by quda::GenericPackGhostLauncher< Float, block_float, Ns, Ms, Nc, Mc, Arg >::advanceBlockDim(), quda::TunableLocalParity::advanceBlockDim(), quda::TunableVectorY::advanceBlockDim(), quda::TunableVectorYZ::advanceBlockDim(), quda::blas::BlasCuda< FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Functor >::advanceSharedBytes(), and quda::blas::copy_ns::CopyCuda< FloatN, N, Output, Input >::advanceSharedBytes().

Here is the caller graph for this function:

advanceGridDim()

virtual bool quda::Tunable::advanceGridDim ( TuneParam &  param ) const

inlineprotectedvirtual

Reimplemented in quda::blas::MultiReduceCuda< NXZ, doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, Reducer >, and quda::ShiftColorSpinorField< Output, Input >.

Definition at line 77 of file tune_quda.h.

References quda::TuneParam::grid.

Referenced by quda::blas::MultiReduceCuda< NXZ, doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, Reducer >::advanceGridDim().

Here is the caller graph for this function:

advanceSharedBytes()

virtual bool quda::Tunable::advanceSharedBytes ( TuneParam &  param ) const

inlineprotectedvirtual

The goal here is to throttle the number of thread blocks per SM by over-allocating shared memory (in order to improve L2 utilization, etc.). We thus request the smallest amount of dynamic shared memory that guarantees throttling to a given number of blocks, in order to allow some extra leeway.

Reimplemented in quda::CalculateY< from_coarse, Float, fineSpin, fineColor, coarseSpin, coarseColor, Arg >, quda::CopySpinorEx< FloatOut, FloatIn, Ns, Nc, OutOrder, InOrder, Basis, extend >, quda::CopyColorSpinor< 4, Arg >, quda::blas::ReduceCuda< doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Reducer >, quda::CopyColorSpinor< Ns, Arg >, quda::blas::MultiReduceCuda< NXZ, doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, Reducer >, quda::blas::copy_ns::CopyCuda< FloatN, N, Output, Input >, quda::CopySpinor< FloatOut, FloatIn, Ns, Nc, OutOrder, InOrder >, and quda::blas::BlasCuda< FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Functor >.

Definition at line 238 of file tune_quda.h.

References quda::TuneParam::block, deviceProp, and quda::TuneParam::shared_bytes.

Referenced by quda::CalculateY< from_coarse, Float, fineSpin, fineColor, coarseSpin, coarseColor, Arg >::advanceSharedBytes().

Here is the caller graph for this function:

advanceTuneParam()

virtual bool quda::Tunable::advanceTuneParam ( TuneParam &  param ) const

inlinevirtual

Reimplemented in quda::dslash::DslashPolicyTune< Dslash >, quda::blas::TileSizeTune< ReducerDiagonal, writeDiagonal, ReducerOffDiagonal, writeOffDiagonal >, quda::DslashCoarsePolicyTune, quda::CalculateY< from_coarse, Float, fineSpin, fineColor, coarseSpin, coarseColor, Arg >, quda::QudaMemCopy, quda::SpinorNoise< real, Ns, Nc, type, Arg >, and quda::CopyGauge< FloatOut, FloatIn, length, Arg >.

Definition at line 335 of file tune_quda.h.

Referenced by quda::CopyGauge< FloatOut, FloatIn, length, Arg >::advanceTuneParam(), quda::SpinorNoise< real, Ns, Nc, type, Arg >::advanceTuneParam(), quda::CalculateY< from_coarse, Float, fineSpin, fineColor, coarseSpin, coarseColor, Arg >::advanceTuneParam(), and quda::tuneLaunch().

Here is the caller graph for this function:

apply()

virtual void quda::Tunable::apply ( const cudaStream_t &  stream )

pure virtual

Implemented in quda::dslash::DslashPolicyTune< Dslash >, quda::blas::TileSizeTune< ReducerDiagonal, writeDiagonal, ReducerOffDiagonal, writeOffDiagonal >, quda::TwistClover< Float, nSpin, nColor, Arg >, quda::DslashCoarsePolicyTune, quda::Clover< Float, nSpin, nColor, Arg >, quda::ProjectSU3< Float, G >, quda::TwistGamma< Float, nColor, Arg >, quda::KSLongLinkForce< Float, Result, Oprod, Gauge >, quda::Gamma< ValueType, basis, dir >, quda::CopySpinorEx< FloatOut, FloatIn, Ns, Nc, OutOrder, InOrder, Basis, extend >, quda::blas::ReduceCuda< doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Reducer >, quda::CopyColorSpinor< 4, Arg >, quda::ExtractGhostEx< Float, length, nDim, dim, Order >, quda::ExtractGhost< nDim, Arg >, quda::Pack< Float, nColor, spin_project >, quda::CalculateY< from_coarse, Float, fineSpin, fineColor, coarseSpin, coarseColor, Arg >, quda::blas::MultiReduceCuda< NXZ, doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, Reducer >, quda::CopyColorSpinor< Ns, Arg >, quda::GaugeOvrImpSTOUT< Float, Arg >, quda::WuppertalSmearing< Float, Ns, Nc, Arg >, quda::ShiftColorSpinorField< Output, Input >, quda::CopyGaugeEx< FloatOut, FloatIn, length, OutOrder, InOrder >, quda::Dslash5< Float, nColor, Arg >, quda::KSForceComplete< Float, Oprod, Gauge, Mom >, quda::GaugeGauss< Float, Arg >, quda::blas::MultiBlas< NXZ, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, Functor, T >, quda::SpinorNoise< real, Ns, Nc, type, Arg >, quda::blas::BlasCuda< FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Functor >, quda::GenericPackGhostLauncher< Float, block_float, Ns, Ms, Nc, Mc, Arg >, quda::CopyGauge< FloatOut, FloatIn, length, Arg >, quda::blas::copy_ns::CopyCuda< FloatN, N, Output, Input >, quda::QudaMemCopy, quda::CopySpinor< FloatOut, FloatIn, Ns, Nc, OutOrder, InOrder >, quda::NdegTwistedMassPreconditioned< Float, nDim, nColor, Arg >, quda::Laplace< Float, nDim, nColor, Arg >, quda::TwistedMassPreconditioned< Float, nDim, nColor, Arg >, quda::Wilson< Float, nDim, nColor, Arg >, quda::DomainWall4D< Float, nDim, nColor, Arg >, quda::TwistedCloverPreconditioned< Float, nDim, nColor, Arg >, quda::WilsonCloverPreconditioned< Float, nDim, nColor, Arg >, quda::NdegTwistedMass< Float, nDim, nColor, Arg >, quda::TwistedClover< Float, nDim, nColor, Arg >, quda::WilsonClover< Float, nDim, nColor, Arg >, quda::DomainWall5D< Float, nDim, nColor, Arg >, quda::Staggered< Float, nDim, nColor, Arg >, quda::Staggered< Float, nDim, nColor, Arg >, quda::TwistedMass< Float, nDim, nColor, Arg >, and quda::GaugePlaq< Float, Gauge >.

Referenced by quda::dslash::DslashBasic< Dslash >::operator()(), quda::dslash::DslashFusedExterior< Dslash >::operator()(), quda::dslash::DslashGDR< Dslash >::operator()(), quda::dslash::DslashFusedGDR< Dslash >::operator()(), quda::dslash::DslashGDRRecv< Dslash >::operator()(), quda::dslash::DslashFusedGDRRecv< Dslash >::operator()(), quda::dslash::DslashZeroCopyPack< Dslash >::operator()(), quda::dslash::DslashFusedZeroCopyPack< Dslash >::operator()(), quda::dslash::DslashZeroCopyPackGDRRecv< Dslash >::operator()(), quda::dslash::DslashFusedZeroCopyPackGDRRecv< Dslash >::operator()(), quda::dslash::DslashZeroCopy< Dslash >::operator()(), quda::dslash::DslashFusedZeroCopy< Dslash >::operator()(), quda::dslash::DslashNC< Dslash >::operator()(), and quda::tuneLaunch().

Here is the caller graph for this function:

blockMin()

virtual int quda::Tunable::blockMin ( ) const

inlineprotectedvirtual

Reimplemented in quda::GenericPackGhostLauncher< Float, block_float, Ns, Ms, Nc, Mc, Arg >, quda::Dslash< Float >, and quda::Dslash5< Float, nColor, Arg >.

Definition at line 106 of file tune_quda.h.

References deviceProp.

Referenced by quda::GenericPackGhostLauncher< Float, block_float, Ns, Ms, Nc, Mc, Arg >::blockMin().

Here is the caller graph for this function:

blockStep()

virtual int quda::Tunable::blockStep ( ) const

inlineprotectedvirtual

Reimplemented in quda::GenericPackGhostLauncher< Float, block_float, Ns, Ms, Nc, Mc, Arg >, quda::Dslash< Float >, and quda::Dslash5< Float, nColor, Arg >.

Definition at line 105 of file tune_quda.h.

References deviceProp.

Referenced by quda::GenericPackGhostLauncher< Float, block_float, Ns, Ms, Nc, Mc, Arg >::blockStep().

Here is the caller graph for this function:

bytes()

virtual long long quda::Tunable::bytes ( ) const

inlineprotectedvirtual

Reimplemented in quda::dslash::DslashPolicyTune< Dslash >, quda::blas::TileSizeTune< ReducerDiagonal, writeDiagonal, ReducerOffDiagonal, writeOffDiagonal >, quda::DslashCoarsePolicyTune, quda::TwistClover< Float, nSpin, nColor, Arg >, quda::ProjectSU3< Float, G >, quda::Clover< Float, nSpin, nColor, Arg >, quda::TwistGamma< Float, nColor, Arg >, quda::Dslash< Float >, quda::KSLongLinkForce< Float, Result, Oprod, Gauge >, quda::Pack< Float, nColor, spin_project >, quda::blas::ReduceCuda< doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Reducer >, quda::CopyColorSpinor< 4, Arg >, quda::blas::MultiReduceCuda< NXZ, doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, Reducer >, quda::CopySpinorEx< FloatOut, FloatIn, Ns, Nc, OutOrder, InOrder, Basis, extend >, quda::Gamma< ValueType, basis, dir >, quda::ExtractGhostEx< Float, length, nDim, dim, Order >, quda::blas::MultiBlas< NXZ, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, Functor, T >, quda::ExtractGhost< nDim, Arg >, quda::GaugeOvrImpSTOUT< Float, Arg >, quda::CopyColorSpinor< Ns, Arg >, quda::ShiftColorSpinorField< Output, Input >, quda::GenericPackGhostLauncher< Float, block_float, Ns, Ms, Nc, Mc, Arg >, quda::WuppertalSmearing< Float, Ns, Nc, Arg >, quda::CopyGaugeEx< FloatOut, FloatIn, length, OutOrder, InOrder >, quda::blas::BlasCuda< FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Functor >, quda::KSForceComplete< Float, Oprod, Gauge, Mom >, quda::GaugeGauss< Float, Arg >, quda::QudaMemCopy, quda::CopyGauge< FloatOut, FloatIn, length, Arg >, quda::SpinorNoise< real, Ns, Nc, type, Arg >, quda::Laplace< Float, nDim, nColor, Arg >, quda::Staggered< Float, nDim, nColor, Arg >, quda::blas::copy_ns::CopyCuda< FloatN, N, Output, Input >, quda::TwistedCloverPreconditioned< Float, nDim, nColor, Arg >, quda::WilsonCloverPreconditioned< Float, nDim, nColor, Arg >, quda::CalculateY< from_coarse, Float, fineSpin, fineColor, coarseSpin, coarseColor, Arg >, quda::CopySpinor< FloatOut, FloatIn, Ns, Nc, OutOrder, InOrder >, quda::DomainWall5D< Float, nDim, nColor, Arg >, quda::TwistedClover< Float, nDim, nColor, Arg >, quda::WilsonClover< Float, nDim, nColor, Arg >, quda::Dslash5< Float, nColor, Arg >, and quda::GaugePlaq< Float, Gauge >.

Definition at line 63 of file tune_quda.h.

References param.

checkLaunchParam()

void quda::Tunable::checkLaunchParam ( TuneParam &  param )

inline

Check the launch parameters of the kernel to ensure that they are valid for the current device.

Definition at line 344 of file tune_quda.h.

References quda::TuneParam::block, deviceProp, errorQuda, and quda::TuneParam::grid.

Referenced by quda::tuneLaunch().

Here is the caller graph for this function:

defaultTuneParam()

virtual void quda::Tunable::defaultTuneParam ( TuneParam &  param ) const

inlinevirtual

sets default values for when tuning is disabled

Reimplemented in quda::dslash::DslashPolicyTune< Dslash >, quda::blas::TileSizeTune< ReducerDiagonal, writeDiagonal, ReducerOffDiagonal, writeOffDiagonal >, quda::DslashCoarsePolicyTune, quda::CalculateY< from_coarse, Float, fineSpin, fineColor, coarseSpin, coarseColor, Arg >, quda::TunableVectorYZ, quda::TunableVectorY, quda::TunableLocalParity, quda::Pack< Float, nColor, spin_project >, quda::blas::ReduceCuda< doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Reducer >, quda::blas::MultiReduceCuda< NXZ, doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, Reducer >, quda::blas::MultiBlas< NXZ, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, Functor, T >, quda::Dslash5< Float, nColor, Arg >, quda::ShiftColorSpinorField< Output, Input >, quda::GenericPackGhostLauncher< Float, block_float, Ns, Ms, Nc, Mc, Arg >, quda::blas::BlasCuda< FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Functor >, quda::NdegTwistedMassPreconditioned< Float, nDim, nColor, Arg >, and quda::blas::copy_ns::CopyCuda< FloatN, N, Output, Input >.

Definition at line 329 of file tune_quda.h.

References quda::TuneParam::grid.

Referenced by quda::blas::copy_ns::CopyCuda< FloatN, N, Output, Input >::defaultTuneParam(), quda::blas::MultiReduceCuda< NXZ, doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, Reducer >::defaultTuneParam(), quda::blas::ReduceCuda< doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Reducer >::defaultTuneParam(), quda::TunableLocalParity::defaultTuneParam(), quda::TunableVectorY::defaultTuneParam(), quda::DslashCoarsePolicyTune::defaultTuneParam(), quda::blas::TileSizeTune< ReducerDiagonal, writeDiagonal, ReducerOffDiagonal, writeOffDiagonal >::defaultTuneParam(), quda::dslash::DslashPolicyTune< Dslash >::defaultTuneParam(), and quda::tuneLaunch().

Here is the caller graph for this function:

flops()

virtual long long quda::Tunable::flops ( ) const

protectedpure virtual

Implemented in quda::dslash::DslashPolicyTune< Dslash >, quda::blas::TileSizeTune< ReducerDiagonal, writeDiagonal, ReducerOffDiagonal, writeOffDiagonal >, quda::DslashCoarsePolicyTune, quda::TwistClover< Float, nSpin, nColor, Arg >, quda::ProjectSU3< Float, G >, quda::Clover< Float, nSpin, nColor, Arg >, quda::TwistGamma< Float, nColor, Arg >, quda::KSLongLinkForce< Float, Result, Oprod, Gauge >, quda::Pack< Float, nColor, spin_project >, quda::Dslash< Float >, quda::blas::ReduceCuda< doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Reducer >, quda::CopyColorSpinor< 4, Arg >, quda::CopySpinorEx< FloatOut, FloatIn, Ns, Nc, OutOrder, InOrder, Basis, extend >, quda::Gamma< ValueType, basis, dir >, quda::blas::MultiReduceCuda< NXZ, doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, Reducer >, quda::ExtractGhostEx< Float, length, nDim, dim, Order >, quda::blas::MultiBlas< NXZ, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, Functor, T >, quda::ExtractGhost< nDim, Arg >, quda::GaugeOvrImpSTOUT< Float, Arg >, quda::CopyColorSpinor< Ns, Arg >, quda::ShiftColorSpinorField< Output, Input >, quda::GenericPackGhostLauncher< Float, block_float, Ns, Ms, Nc, Mc, Arg >, quda::WuppertalSmearing< Float, Ns, Nc, Arg >, quda::CopyGaugeEx< FloatOut, FloatIn, length, OutOrder, InOrder >, quda::blas::BlasCuda< FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Functor >, quda::KSForceComplete< Float, Oprod, Gauge, Mom >, quda::GaugeGauss< Float, Arg >, quda::QudaMemCopy, quda::CopyGauge< FloatOut, FloatIn, length, Arg >, quda::SpinorNoise< real, Ns, Nc, type, Arg >, quda::NdegTwistedMassPreconditioned< Float, nDim, nColor, Arg >, quda::blas::copy_ns::CopyCuda< FloatN, N, Output, Input >, quda::CopySpinor< FloatOut, FloatIn, Ns, Nc, OutOrder, InOrder >, quda::Staggered< Float, nDim, nColor, Arg >, quda::TwistedMassPreconditioned< Float, nDim, nColor, Arg >, quda::TwistedCloverPreconditioned< Float, nDim, nColor, Arg >, quda::WilsonCloverPreconditioned< Float, nDim, nColor, Arg >, quda::Laplace< Float, nDim, nColor, Arg >, quda::NdegTwistedMass< Float, nDim, nColor, Arg >, quda::TwistedClover< Float, nDim, nColor, Arg >, quda::WilsonClover< Float, nDim, nColor, Arg >, quda::TwistedMass< Float, nDim, nColor, Arg >, quda::DomainWall5D< Float, nDim, nColor, Arg >, quda::CalculateY< from_coarse, Float, fineSpin, fineColor, coarseSpin, coarseColor, Arg >, quda::GaugePlaq< Float, Gauge >, and quda::Dslash5< Float, nColor, Arg >.

gridStep()

virtual int quda::Tunable::gridStep ( ) const

inlineprotectedvirtual

gridStep sets the step size when iterating the grid size in advanceGridDim.

Returns

Grid step size

Reimplemented in quda::Pack< Float, nColor, spin_project >.

Definition at line 103 of file tune_quda.h.

Referenced by quda::Pack< Float, nColor, spin_project >::gridStep().

Here is the caller graph for this function:

initTuneParam()

virtual void quda::Tunable::initTuneParam ( TuneParam &  param ) const

inlinevirtual

Reimplemented in quda::dslash::DslashPolicyTune< Dslash >, quda::blas::TileSizeTune< ReducerDiagonal, writeDiagonal, ReducerOffDiagonal, writeOffDiagonal >, quda::DslashCoarsePolicyTune, quda::CalculateY< from_coarse, Float, fineSpin, fineColor, coarseSpin, coarseColor, Arg >, quda::TunableVectorYZ, quda::TunableVectorY, quda::TunableLocalParity, quda::Pack< Float, nColor, spin_project >, quda::blas::ReduceCuda< doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Reducer >, quda::blas::MultiReduceCuda< NXZ, doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, Reducer >, quda::blas::MultiBlas< NXZ, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, Functor, T >, quda::Dslash5< Float, nColor, Arg >, quda::ShiftColorSpinorField< Output, Input >, quda::GenericPackGhostLauncher< Float, block_float, Ns, Ms, Nc, Mc, Arg >, quda::blas::BlasCuda< FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Functor >, quda::blas::copy_ns::CopyCuda< FloatN, N, Output, Input >, and quda::NdegTwistedMassPreconditioned< Float, nDim, nColor, Arg >.

Definition at line 304 of file tune_quda.h.

References quda::TuneParam::block, deviceProp, errorQuda, quda::TuneParam::grid, and quda::TuneParam::shared_bytes.

Referenced by quda::blas::BlasCuda< FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Functor >::defaultTuneParam(), quda::blas::copy_ns::CopyCuda< FloatN, N, Output, Input >::initTuneParam(), quda::blas::BlasCuda< FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Functor >::initTuneParam(), quda::blas::MultiReduceCuda< NXZ, doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, Reducer >::initTuneParam(), quda::blas::ReduceCuda< doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Reducer >::initTuneParam(), quda::TunableLocalParity::initTuneParam(), quda::TunableVectorY::initTuneParam(), quda::DslashCoarsePolicyTune::initTuneParam(), quda::blas::TileSizeTune< ReducerDiagonal, writeDiagonal, ReducerOffDiagonal, writeOffDiagonal >::initTuneParam(), quda::dslash::DslashPolicyTune< Dslash >::initTuneParam(), and quda::tuneLaunch().

Here is the caller graph for this function:

jitifyError() [1/2]

CUresult quda::Tunable::jitifyError ( ) const

inline

Definition at line 375 of file tune_quda.h.

Referenced by quda::blas::multiReduceLaunch(), quda::blas::reduceLaunch(), and quda::tuneLaunch().

Here is the caller graph for this function:

jitifyError() [2/2]

CUresult& quda::Tunable::jitifyError ( )

inline

Definition at line 376 of file tune_quda.h.

maxBlockSize()

virtual unsigned int quda::Tunable::maxBlockSize ( const TuneParam &  param ) const

inlineprotectedvirtual

Reimplemented in quda::TunableLocalParity, and quda::blas::MultiReduceCuda< NXZ, doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, Reducer >.

Definition at line 94 of file tune_quda.h.

References quda::TuneParam::block, and deviceProp.

maxBlocksPerSM()

unsigned int quda::Tunable::maxBlocksPerSM ( ) const

inlineprotected

For some reason this can't be queried from the device properties, so here we set set this. Based on Table 14 of the CUDA Programming Guide 10.0 (Technical Specifications per Compute Capability)

Returns

The maximum number of simultaneously resident blocks per SM

Definition at line 153 of file tune_quda.h.

References deviceProp, and warningQuda.

maxDynamicSharedBytesPerBlock()

unsigned int quda::Tunable::maxDynamicSharedBytesPerBlock ( ) const

inlineprotected

This can't be correctly queried in CUDA for all architectures so here we set set this. Based on Table 14 of the CUDA Programming Guide 10.0 (Technical Specifications per Compute Capability).

Returns

The maximum dynamic shared memory to CUDA thread block

Definition at line 198 of file tune_quda.h.

References deviceProp, and warningQuda.

Referenced by quda::Pack< Float, nColor, spin_project >::defaultTuneParam(), quda::Pack< Float, nColor, spin_project >::initTuneParam(), quda::Dslash5< Float, nColor, Arg >::maxSharedBytesPerBlock(), quda::Dslash< Float >::maxSharedBytesPerBlock(), and quda::Pack< Float, nColor, spin_project >::tuneSharedBytes().

Here is the caller graph for this function:

maxGridSize()

virtual unsigned int quda::Tunable::maxGridSize ( ) const

inlineprotectedvirtual

Reimplemented in quda::Pack< Float, nColor, spin_project >.

Definition at line 95 of file tune_quda.h.

References deviceProp.

Referenced by quda::Pack< Float, nColor, spin_project >::maxGridSize().

Here is the caller graph for this function:

maxSharedBytesPerBlock()

virtual unsigned int quda::Tunable::maxSharedBytesPerBlock ( ) const

inlineprotectedvirtual

The maximum shared memory that a CUDA thread block can use in the autotuner. This isn't necessarily the same as maxDynamicSharedMemoryPerBlock since that may need explicit opt in to enable (by calling setMaxDynamicSharedBytes for the kernel in question). If the CUDA kernel in question does this opt in then this function can be overloaded to return maxDynamicSharedBytesPerBlock.

Returns

The maximum shared bytes limit per block the autotung will utilize.

Reimplemented in quda::Dslash< Float >, and quda::Dslash5< Float, nColor, Arg >.

Definition at line 229 of file tune_quda.h.

References deviceProp.

Referenced by quda::Dslash5< Float, nColor, Arg >::maxSharedBytesPerBlock(), and quda::Pack< Float, nColor, spin_project >::tuneSharedBytes().

Here is the caller graph for this function:

minGridSize()

virtual unsigned int quda::Tunable::minGridSize ( ) const

inlineprotectedvirtual

Reimplemented in quda::Pack< Float, nColor, spin_project >.

Definition at line 96 of file tune_quda.h.

Referenced by quda::Pack< Float, nColor, spin_project >::minGridSize().

Here is the caller graph for this function:

minThreads()

virtual unsigned int quda::Tunable::minThreads ( ) const

inlineprotectedvirtual

Reimplemented in quda::TwistClover< Float, nSpin, nColor, Arg >, quda::Clover< Float, nSpin, nColor, Arg >, quda::ProjectSU3< Float, G >, quda::TwistGamma< Float, nColor, Arg >, quda::KSLongLinkForce< Float, Result, Oprod, Gauge >, quda::Gamma< ValueType, basis, dir >, quda::CopySpinorEx< FloatOut, FloatIn, Ns, Nc, OutOrder, InOrder, Basis, extend >, quda::CopyColorSpinor< 4, Arg >, quda::ExtractGhostEx< Float, length, nDim, dim, Order >, quda::ExtractGhost< nDim, Arg >, quda::CopyColorSpinor< Ns, Arg >, quda::GaugeOvrImpSTOUT< Float, Arg >, quda::WuppertalSmearing< Float, Ns, Nc, Arg >, quda::CalculateY< from_coarse, Float, fineSpin, fineColor, coarseSpin, coarseColor, Arg >, quda::Pack< Float, nColor, spin_project >, quda::CopyGaugeEx< FloatOut, FloatIn, length, OutOrder, InOrder >, quda::KSForceComplete< Float, Oprod, Gauge, Mom >, quda::GaugeGauss< Float, Arg >, quda::SpinorNoise< real, Ns, Nc, type, Arg >, quda::Dslash5< Float, nColor, Arg >, quda::Dslash< Float >, quda::CopySpinor< FloatOut, FloatIn, Ns, Nc, OutOrder, InOrder >, quda::GenericPackGhostLauncher< Float, block_float, Ns, Ms, Nc, Mc, Arg >, and quda::CopyGauge< FloatOut, FloatIn, length, Arg >.

Definition at line 72 of file tune_quda.h.

paramString()

virtual std::string quda::Tunable::paramString ( const TuneParam &  param ) const

inlinevirtual

Definition at line 287 of file tune_quda.h.

References param.

Referenced by quda::tuneLaunch().

Here is the caller graph for this function:

perfString()

virtual std::string quda::Tunable::perfString ( float  time ) const

inlinevirtual

Definition at line 294 of file tune_quda.h.

References quda::blas::bytes, quda::blas::flops, and quda::TuneParam::time.

Referenced by quda::tuneLaunch().

Here is the caller graph for this function:

postTune()

virtual void quda::Tunable::postTune ( )

inlinevirtual

Reimplemented in quda::dslash::DslashPolicyTune< Dslash >, quda::blas::TileSizeTune< ReducerDiagonal, writeDiagonal, ReducerOffDiagonal, writeOffDiagonal >, quda::CalculateY< from_coarse, Float, fineSpin, fineColor, coarseSpin, coarseColor, Arg >, quda::TwistClover< Float, nSpin, nColor, Arg >, quda::Clover< Float, nSpin, nColor, Arg >, quda::ProjectSU3< Float, G >, quda::TwistGamma< Float, nColor, Arg >, quda::Dslash< Float >, quda::Gamma< ValueType, basis, dir >, quda::blas::ReduceCuda< doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Reducer >, quda::blas::MultiReduceCuda< NXZ, doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, Reducer >, quda::blas::MultiBlas< NXZ, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, Functor, T >, quda::GaugeOvrImpSTOUT< Float, Arg >, quda::GaugeGauss< Float, Arg >, quda::SpinorNoise< real, Ns, Nc, type, Arg >, quda::blas::BlasCuda< FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Functor >, and quda::blas::copy_ns::CopyCuda< FloatN, N, Output, Input >.

Definition at line 284 of file tune_quda.h.

Referenced by quda::tuneLaunch().

Here is the caller graph for this function:

preTune()

virtual void quda::Tunable::preTune ( )

inlinevirtual

Reimplemented in quda::dslash::DslashPolicyTune< Dslash >, quda::blas::TileSizeTune< ReducerDiagonal, writeDiagonal, ReducerOffDiagonal, writeOffDiagonal >, quda::CalculateY< from_coarse, Float, fineSpin, fineColor, coarseSpin, coarseColor, Arg >, quda::TwistClover< Float, nSpin, nColor, Arg >, quda::Clover< Float, nSpin, nColor, Arg >, quda::ProjectSU3< Float, G >, quda::TwistGamma< Float, nColor, Arg >, quda::Dslash< Float >, quda::Gamma< ValueType, basis, dir >, quda::blas::ReduceCuda< doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Reducer >, quda::blas::MultiReduceCuda< NXZ, doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, Reducer >, quda::blas::MultiBlas< NXZ, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, Functor, T >, quda::GaugeOvrImpSTOUT< Float, Arg >, quda::GaugeGauss< Float, Arg >, quda::SpinorNoise< real, Ns, Nc, type, Arg >, quda::blas::BlasCuda< FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Functor >, and quda::blas::copy_ns::CopyCuda< FloatN, N, Output, Input >.

Definition at line 283 of file tune_quda.h.

Referenced by quda::tuneLaunch().

Here is the caller graph for this function:

resetBlockDim()

virtual void quda::Tunable::resetBlockDim ( TuneParam &  param ) const

inlineprotectedvirtual

Definition at line 108 of file tune_quda.h.

References quda::TuneParam::block, deviceProp, and errorQuda.

setMaxDynamicSharedBytesPerBlock()

template<typename F >

void quda::Tunable::setMaxDynamicSharedBytesPerBlock ( F *  func ) const

inlineprotected

Enable the maximum dynamic shared bytes for the kernel "func" (values given by maxDynamicSharedBytesPerBlock()).

Parameters

[in]

func

Function pointer to the kernel we want to enable max shared memory per block for

Definition at line 181 of file tune_quda.h.

Referenced by quda::Dslash< Float >::launch(), quda::Dslash5< Float, nColor, Arg >::launch(), and quda::Pack< Float, nColor, spin_project >::launch().

Here is the caller graph for this function:

sharedBytesPerBlock()

virtual unsigned int quda::Tunable::sharedBytesPerBlock ( const TuneParam &  param ) const

protectedpure virtual

Implemented in quda::dslash::DslashPolicyTune< Dslash >, quda::blas::TileSizeTune< ReducerDiagonal, writeDiagonal, ReducerOffDiagonal, writeOffDiagonal >, quda::DslashCoarsePolicyTune, quda::ProjectSU3< Float, G >, quda::TunableVectorY, quda::TunableLocalParity, quda::KSLongLinkForce< Float, Result, Oprod, Gauge >, quda::CopySpinorEx< FloatOut, FloatIn, Ns, Nc, OutOrder, InOrder, Basis, extend >, quda::CopyColorSpinor< 4, Arg >, quda::ExtractGhostEx< Float, length, nDim, dim, Order >, quda::blas::ReduceCuda< doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Reducer >, quda::ExtractGhost< nDim, Arg >, quda::CalculateY< from_coarse, Float, fineSpin, fineColor, coarseSpin, coarseColor, Arg >, quda::CopyColorSpinor< Ns, Arg >, quda::blas::MultiReduceCuda< NXZ, doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, Reducer >, quda::CopyGaugeEx< FloatOut, FloatIn, length, OutOrder, InOrder >, quda::KSForceComplete< Float, Oprod, Gauge, Mom >, quda::SpinorNoise< real, Ns, Nc, type, Arg >, quda::CopySpinor< FloatOut, FloatIn, Ns, Nc, OutOrder, InOrder >, quda::blas::copy_ns::CopyCuda< FloatN, N, Output, Input >, quda::blas::BlasCuda< FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Functor >, quda::QudaMemCopy, and quda::CopyGauge< FloatOut, FloatIn, length, Arg >.

sharedBytesPerThread()

virtual unsigned int quda::Tunable::sharedBytesPerThread ( ) const

protectedpure virtual

Implemented in quda::dslash::DslashPolicyTune< Dslash >, quda::blas::TileSizeTune< ReducerDiagonal, writeDiagonal, ReducerOffDiagonal, writeOffDiagonal >, quda::DslashCoarsePolicyTune, quda::ProjectSU3< Float, G >, quda::TunableVectorY, quda::TunableLocalParity, quda::KSLongLinkForce< Float, Result, Oprod, Gauge >, quda::CopySpinorEx< FloatOut, FloatIn, Ns, Nc, OutOrder, InOrder, Basis, extend >, quda::CopyColorSpinor< 4, Arg >, quda::ExtractGhostEx< Float, length, nDim, dim, Order >, quda::blas::ReduceCuda< doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Reducer >, quda::ExtractGhost< nDim, Arg >, quda::CopyColorSpinor< Ns, Arg >, quda::blas::MultiReduceCuda< NXZ, doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, Reducer >, quda::ShiftColorSpinorField< Output, Input >, quda::CopyGaugeEx< FloatOut, FloatIn, length, OutOrder, InOrder >, quda::KSForceComplete< Float, Oprod, Gauge, Mom >, quda::SpinorNoise< real, Ns, Nc, type, Arg >, quda::Dslash5< Float, nColor, Arg >, quda::blas::copy_ns::CopyCuda< FloatN, N, Output, Input >, quda::CopySpinor< FloatOut, FloatIn, Ns, Nc, OutOrder, InOrder >, quda::blas::BlasCuda< FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Functor >, quda::NdegTwistedMassPreconditioned< Float, nDim, nColor, Arg >, quda::QudaMemCopy, and quda::CopyGauge< FloatOut, FloatIn, length, Arg >.

tuneAuxDim()

virtual bool quda::Tunable::tuneAuxDim ( ) const

inlineprotectedvirtual

Reimplemented in quda::dslash::DslashPolicyTune< Dslash >, quda::DslashCoarsePolicyTune, quda::CalculateY< from_coarse, Float, fineSpin, fineColor, coarseSpin, coarseColor, Arg >, quda::Pack< Float, nColor, spin_project >, and quda::GenericPackGhostLauncher< Float, block_float, Ns, Ms, Nc, Mc, Arg >.

Definition at line 74 of file tune_quda.h.

tuneGridDim()

virtual bool quda::Tunable::tuneGridDim ( ) const

inlineprotectedvirtual

Reimplemented in quda::dslash::DslashPolicyTune< Dslash >, quda::TwistClover< Float, nSpin, nColor, Arg >, quda::DslashCoarsePolicyTune, quda::Clover< Float, nSpin, nColor, Arg >, quda::ProjectSU3< Float, G >, quda::TunableLocalParity, quda::TwistGamma< Float, nColor, Arg >, quda::KSLongLinkForce< Float, Result, Oprod, Gauge >, quda::Gamma< ValueType, basis, dir >, quda::CopySpinorEx< FloatOut, FloatIn, Ns, Nc, OutOrder, InOrder, Basis, extend >, quda::CopyColorSpinor< 4, Arg >, quda::ExtractGhostEx< Float, length, nDim, dim, Order >, quda::ExtractGhost< nDim, Arg >, quda::CalculateY< from_coarse, Float, fineSpin, fineColor, coarseSpin, coarseColor, Arg >, quda::CopyColorSpinor< Ns, Arg >, quda::GaugeOvrImpSTOUT< Float, Arg >, quda::WuppertalSmearing< Float, Ns, Nc, Arg >, quda::CopyGaugeEx< FloatOut, FloatIn, length, OutOrder, InOrder >, quda::GaugeGauss< Float, Arg >, quda::KSForceComplete< Float, Oprod, Gauge, Mom >, quda::SpinorNoise< real, Ns, Nc, type, Arg >, quda::Dslash5< Float, nColor, Arg >, quda::Dslash< Float >, quda::Pack< Float, nColor, spin_project >, quda::CopySpinor< FloatOut, FloatIn, Ns, Nc, OutOrder, InOrder >, quda::GenericPackGhostLauncher< Float, block_float, Ns, Ms, Nc, Mc, Arg >, quda::CopyGauge< FloatOut, FloatIn, length, Arg >, and quda::GaugePlaq< Float, Gauge >.

Definition at line 73 of file tune_quda.h.

tuneKey()

virtual TuneKey quda::Tunable::tuneKey ( ) const

pure virtual

Implemented in quda::dslash::DslashPolicyTune< Dslash >, quda::blas::TileSizeTune< ReducerDiagonal, writeDiagonal, ReducerOffDiagonal, writeOffDiagonal >, quda::TwistClover< Float, nSpin, nColor, Arg >, quda::DslashCoarsePolicyTune, quda::CalculateY< from_coarse, Float, fineSpin, fineColor, coarseSpin, coarseColor, Arg >, quda::Clover< Float, nSpin, nColor, Arg >, quda::ProjectSU3< Float, G >, quda::TwistGamma< Float, nColor, Arg >, quda::KSLongLinkForce< Float, Result, Oprod, Gauge >, quda::Pack< Float, nColor, spin_project >, quda::Gamma< ValueType, basis, dir >, quda::CopyColorSpinor< 4, Arg >, quda::CopySpinorEx< FloatOut, FloatIn, Ns, Nc, OutOrder, InOrder, Basis, extend >, quda::blas::ReduceCuda< doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Reducer >, quda::ExtractGhostEx< Float, length, nDim, dim, Order >, quda::ExtractGhost< nDim, Arg >, quda::Dslash5< Float, nColor, Arg >, quda::GaugeOvrImpSTOUT< Float, Arg >, quda::ShiftColorSpinorField< Output, Input >, quda::CopyColorSpinor< Ns, Arg >, quda::WuppertalSmearing< Float, Ns, Nc, Arg >, quda::blas::MultiReduceCuda< NXZ, doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, Reducer >, quda::Staggered< Float, nDim, nColor, Arg >, quda::GenericPackGhostLauncher< Float, block_float, Ns, Ms, Nc, Mc, Arg >, quda::Laplace< Float, nDim, nColor, Arg >, quda::CopyGaugeEx< FloatOut, FloatIn, length, OutOrder, InOrder >, quda::KSForceComplete< Float, Oprod, Gauge, Mom >, quda::GaugeGauss< Float, Arg >, quda::TwistedCloverPreconditioned< Float, nDim, nColor, Arg >, quda::WilsonCloverPreconditioned< Float, nDim, nColor, Arg >, quda::NdegTwistedMassPreconditioned< Float, nDim, nColor, Arg >, quda::QudaMemCopy, quda::SpinorNoise< real, Ns, Nc, type, Arg >, quda::CopyGauge< FloatOut, FloatIn, length, Arg >, quda::blas::MultiBlas< NXZ, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, Functor, T >, quda::TwistedClover< Float, nDim, nColor, Arg >, quda::WilsonClover< Float, nDim, nColor, Arg >, quda::DomainWall5D< Float, nDim, nColor, Arg >, quda::blas::BlasCuda< FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Functor >, quda::TwistedMassPreconditioned< Float, nDim, nColor, Arg >, quda::NdegTwistedMass< Float, nDim, nColor, Arg >, quda::CopySpinor< FloatOut, FloatIn, Ns, Nc, OutOrder, InOrder >, quda::TwistedMass< Float, nDim, nColor, Arg >, quda::blas::copy_ns::CopyCuda< FloatN, N, Output, Input >, quda::DomainWall4D< Float, nDim, nColor, Arg >, quda::Staggered< Float, nDim, nColor, Arg >, quda::Wilson< Float, nDim, nColor, Arg >, and quda::GaugePlaq< Float, Gauge >.

Referenced by quda::dslash::DslashPolicyTune< Dslash >::tuneKey(), and quda::tuneLaunch().

Here is the caller graph for this function:

tuneSharedBytes()

virtual bool quda::Tunable::tuneSharedBytes ( ) const

inlineprotectedvirtual

Reimplemented in quda::KSLongLinkForce< Float, Result, Oprod, Gauge >, quda::Pack< Float, nColor, spin_project >, quda::KSForceComplete< Float, Oprod, Gauge, Mom >, and quda::blas::MultiBlas< NXZ, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, Functor, T >.

Definition at line 75 of file tune_quda.h.

tuningIter()

virtual int quda::Tunable::tuningIter ( ) const

inlinevirtual

Reimplemented in quda::dslash::DslashPolicyTune< Dslash >, quda::DslashCoarsePolicyTune, quda::Pack< Float, nColor, spin_project >, quda::blas::ReduceCuda< doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Reducer >, quda::blas::MultiReduceCuda< NXZ, doubleN, ReduceType, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, Reducer >, quda::blas::MultiBlas< NXZ, FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, Functor, T >, quda::blas::BlasCuda< FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Functor >, quda::blas::copy_ns::CopyCuda< FloatN, N, Output, Input >, and quda::Dslash< Float >.

Definition at line 285 of file tune_quda.h.

Referenced by quda::tuneLaunch().

Here is the caller graph for this function:

writeAuxString()

int quda::Tunable::writeAuxString ( const char *  format,   ...  )

inlineprotected

Definition at line 267 of file tune_quda.h.

References quda::TuneKey::aux_n, and errorQuda.

Member Data Documentation

aux

char quda::Tunable::aux[TuneKey::aux_n]

protected

Definition at line 265 of file tune_quda.h.

Referenced by quda::blas::BlasCuda< FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Functor >::BlasCuda(), quda::CalculateY< from_coarse, Float, fineSpin, fineColor, coarseSpin, coarseColor, Arg >::CalculateY(), quda::Dslash5< Float, nColor, Arg >::Dslash5(), quda::Pack< Float, nColor, spin_project >::fillAux(), quda::GaugePlaq< Float, Gauge >::GaugePlaq(), quda::GenericPackGhostLauncher< Float, block_float, Ns, Ms, Nc, Mc, Arg >::GenericPackGhostLauncher(), quda::QudaMemCopy::QudaMemCopy(), quda::GaugePlaq< Float, Gauge >::tuneKey(), quda::blas::BlasCuda< FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Functor >::tuneKey(), quda::QudaMemCopy::tuneKey(), quda::GenericPackGhostLauncher< Float, block_float, Ns, Ms, Nc, Mc, Arg >::tuneKey(), quda::Dslash5< Float, nColor, Arg >::tuneKey(), quda::Pack< Float, nColor, spin_project >::tuneKey(), and quda::CalculateY< from_coarse, Float, fineSpin, fineColor, coarseSpin, coarseColor, Arg >::tuneKey().

jitify_error

CUresult quda::Tunable::jitify_error

protected

This is the return result from kernels launched using jitify

Definition at line 276 of file tune_quda.h.

Referenced by quda::GaugePlaq< Float, Gauge >::apply(), quda::DomainWall4D< Float, nDim, nColor, Arg >::apply(), quda::GenericPackGhostLauncher< Float, block_float, Ns, Ms, Nc, Mc, Arg >::apply(), quda::blas::BlasCuda< FloatN, M, SpinorX, SpinorY, SpinorZ, SpinorW, SpinorV, Functor >::apply(), quda::CalculateY< from_coarse, Float, fineSpin, fineColor, coarseSpin, coarseColor, Arg >::apply(), and quda::Dslash< Float >::instantiate().

---

The documentation for this class was generated from the following file:

• include/tune_quda.h