QUDA  1.0.0
All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Friends Macros
Public Member Functions | Protected Member Functions | Protected Attributes | List of all members
quda::Pack< Float, nColor, spin_project > Class Template Reference
Inheritance diagram for quda::Pack< Float, nColor, spin_project >:
Inheritance graph
[legend]
Collaboration diagram for quda::Pack< Float, nColor, spin_project >:
Collaboration graph
[legend]

Public Member Functions

 Pack (void *ghost[], const ColorSpinorField &in, MemoryLocation location, int nFace, bool dagger, int parity, double a, double b, double c)
 
virtual ~Pack ()
 
template<typename T , typename Arg >
void launch (T *f, const TuneParam &tp, Arg &arg, const cudaStream_t &stream)
 
void apply (const cudaStream_t &stream)
 
bool tuneSharedBytes () const
 
void initTuneParam (TuneParam &param) const
 
void defaultTuneParam (TuneParam &param) const
 
TuneKey tuneKey () const
 
int tuningIter () const
 
long long flops () const
 
long long bytes () const
 

Protected Member Functions

bool tuneGridDim () const
 
unsigned int maxGridSize () const
 
unsigned int minGridSize () const
 
int gridStep () const
 gridStep sets the step size when iterating the grid size in advanceGridDim. More...
 
bool tuneAuxDim () const
 
unsigned int minThreads () const
 
void fillAux ()
 

Protected Attributes

void ** ghost
 
const ColorSpinorFieldin
 
MemoryLocation location
 
const int nFace
 
const bool dagger
 
const int parity
 
const int nParity
 
int threads
 
const double a
 
const double b
 
const double c
 
int twist
 

Additional Inherited Members

- Private Member Functions inherited from quda::TunableVectorYZ
 TunableVectorYZ (unsigned int vector_length_y, unsigned int vector_length_z)
 
bool advanceBlockDim (TuneParam &param) const
 
void resizeVector (int y, int z) const
 
void resizeStep (int y, int z) const
 
- Private Member Functions inherited from quda::TunableVectorY
 TunableVectorY (unsigned int vector_length_y)
 
void resizeVector (int y) const
 
void resizeStep (int y) const
 
virtual unsigned int sharedBytesPerThread () const
 
virtual unsigned int sharedBytesPerBlock (const TuneParam &param) const
 
- Private Member Functions inherited from quda::Tunable
 Tunable ()
 
virtual ~Tunable ()
 
virtual void preTune ()
 
virtual void postTune ()
 
virtual std::string paramString (const TuneParam &param) const
 
virtual std::string perfString (float time) const
 
virtual bool advanceTuneParam (TuneParam &param) const
 
void checkLaunchParam (TuneParam &param)
 
CUresult jitifyError () const
 
CUresult & jitifyError ()
 
virtual bool advanceGridDim (TuneParam &param) const
 
virtual unsigned int maxBlockSize (const TuneParam &param) const
 
virtual int blockStep () const
 
virtual int blockMin () const
 
virtual void resetBlockDim (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,...)
 
- Private Attributes inherited from quda::TunableVectorY
unsigned int vector_length_y
 
unsigned int step_y
 
bool tune_block_x
 
- Private Attributes inherited from quda::Tunable
char aux [TuneKey::aux_n]
 
CUresult jitify_error
 

Detailed Description

template<typename Float, int nColor, bool spin_project>
class quda::Pack< Float, nColor, spin_project >

Definition at line 45 of file dslash_pack2.cu.

Constructor & Destructor Documentation

◆ Pack()

template<typename Float, int nColor, bool spin_project>
quda::Pack< Float, nColor, spin_project >::Pack ( void *  ghost[],
const ColorSpinorField in,
MemoryLocation  location,
int  nFace,
bool  dagger,
int  parity,
double  a,
double  b,
double  c 
)
inline

◆ ~Pack()

template<typename Float, int nColor, bool spin_project>
virtual quda::Pack< Float, nColor, spin_project >::~Pack ( )
inlinevirtual

Definition at line 189 of file dslash_pack2.cu.

Member Function Documentation

◆ apply()

template<typename Float, int nColor, bool spin_project>
void quda::Pack< Float, nColor, spin_project >::apply ( const cudaStream_t &  stream)
inlinevirtual

◆ bytes()

template<typename Float, int nColor, bool spin_project>
long long quda::Pack< Float, nColor, spin_project >::bytes ( ) const
inlinevirtual

◆ defaultTuneParam()

template<typename Float, int nColor, bool spin_project>
void quda::Pack< Float, nColor, spin_project >::defaultTuneParam ( TuneParam param) const
inlinevirtual

sets default values for when tuning is disabled

Reimplemented from quda::TunableVectorYZ.

Definition at line 309 of file dslash_pack2.cu.

References quda::TunableVectorYZ::defaultTuneParam(), quda::TuneParam::grid, quda::Host, quda::Tunable::maxDynamicSharedBytesPerBlock(), quda::Pack< Float, nColor, spin_project >::minGridSize(), and quda::TuneParam::shared_bytes.

Here is the call graph for this function:

◆ fillAux()

template<typename Float, int nColor, bool spin_project>
void quda::Pack< Float, nColor, spin_project >::fillAux ( )
inlineprotected

◆ flops()

template<typename Float, int nColor, bool spin_project>
long long quda::Pack< Float, nColor, spin_project >::flops ( ) const
inlinevirtual

◆ gridStep()

template<typename Float, int nColor, bool spin_project>
int quda::Pack< Float, nColor, spin_project >::gridStep ( ) const
inlineprotectedvirtual

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

Returns
Grid step size

Reimplemented from quda::Tunable.

Definition at line 104 of file dslash_pack2.cu.

References quda::commDim, quda::Tunable::gridStep(), quda::Host, and quda::ColorSpinorField::Ndim().

Here is the call graph for this function:

◆ initTuneParam()

template<typename Float, int nColor, bool spin_project>
void quda::Pack< Float, nColor, spin_project >::initTuneParam ( TuneParam param) const
inlinevirtual

◆ launch()

template<typename Float, int nColor, bool spin_project>
template<typename T , typename Arg >
void quda::Pack< Float, nColor, spin_project >::launch ( T *  f,
const TuneParam tp,
Arg arg,
const cudaStream_t &  stream 
)
inline

Definition at line 192 of file dslash_pack2.cu.

References quda::TuneParam::block, deviceProp, quda::TuneParam::grid, quda::qudaLaunchKernel(), quda::Tunable::setMaxDynamicSharedBytesPerBlock(), and quda::TuneParam::shared_bytes.

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

Here is the call graph for this function:
Here is the caller graph for this function:

◆ maxGridSize()

template<typename Float, int nColor, bool spin_project>
unsigned int quda::Pack< Float, nColor, spin_project >::maxGridSize ( ) const
inlineprotectedvirtual

Reimplemented from quda::Tunable.

Definition at line 64 of file dslash_pack2.cu.

References quda::commDim, quda::Host, quda::Tunable::maxGridSize(), and quda::ColorSpinorField::Ndim().

Here is the call graph for this function:

◆ minGridSize()

template<typename Float, int nColor, bool spin_project>
unsigned int quda::Pack< Float, nColor, spin_project >::minGridSize ( ) const
inlineprotectedvirtual

Reimplemented from quda::Tunable.

Definition at line 84 of file dslash_pack2.cu.

References quda::commDim, quda::Host, quda::Tunable::minGridSize(), and quda::ColorSpinorField::Ndim().

Referenced by quda::Pack< Float, nColor, spin_project >::defaultTuneParam(), and quda::Pack< Float, nColor, spin_project >::initTuneParam().

Here is the call graph for this function:
Here is the caller graph for this function:

◆ minThreads()

template<typename Float, int nColor, bool spin_project>
unsigned int quda::Pack< Float, nColor, spin_project >::minThreads ( ) const
inlineprotectedvirtual

Reimplemented from quda::Tunable.

Definition at line 123 of file dslash_pack2.cu.

References quda::Pack< Float, nColor, spin_project >::threads.

◆ tuneAuxDim()

template<typename Float, int nColor, bool spin_project>
bool quda::Pack< Float, nColor, spin_project >::tuneAuxDim ( ) const
inlineprotectedvirtual

Reimplemented from quda::Tunable.

Definition at line 122 of file dslash_pack2.cu.

◆ tuneGridDim()

template<typename Float, int nColor, bool spin_project>
bool quda::Pack< Float, nColor, spin_project >::tuneGridDim ( ) const
inlineprotectedvirtual

Reimplemented from quda::Tunable.

Definition at line 62 of file dslash_pack2.cu.

◆ tuneKey()

template<typename Float, int nColor, bool spin_project>
TuneKey quda::Pack< Float, nColor, spin_project >::tuneKey ( ) const
inlinevirtual

Implements quda::Tunable.

Definition at line 321 of file dslash_pack2.cu.

References quda::Tunable::aux, and quda::LatticeField::VolString().

Here is the call graph for this function:

◆ tuneSharedBytes()

template<typename Float, int nColor, bool spin_project>
bool quda::Pack< Float, nColor, spin_project >::tuneSharedBytes ( ) const
inlinevirtual

Reimplemented from quda::Tunable.

Definition at line 289 of file dslash_pack2.cu.

References quda::Tunable::maxDynamicSharedBytesPerBlock(), and quda::Tunable::maxSharedBytesPerBlock().

Here is the call graph for this function:

◆ tuningIter()

template<typename Float, int nColor, bool spin_project>
int quda::Pack< Float, nColor, spin_project >::tuningIter ( ) const
inlinevirtual

Reimplemented from quda::Tunable.

Definition at line 323 of file dslash_pack2.cu.

Member Data Documentation

◆ a

template<typename Float, int nColor, bool spin_project>
const double quda::Pack< Float, nColor, spin_project >::a
protected

Definition at line 57 of file dslash_pack2.cu.

Referenced by quda::PackGhost().

◆ b

template<typename Float, int nColor, bool spin_project>
const double quda::Pack< Float, nColor, spin_project >::b
protected

Definition at line 58 of file dslash_pack2.cu.

Referenced by quda::PackGhost().

◆ c

template<typename Float, int nColor, bool spin_project>
const double quda::Pack< Float, nColor, spin_project >::c
protected

Definition at line 59 of file dslash_pack2.cu.

Referenced by quda::PackGhost().

◆ dagger

template<typename Float, int nColor, bool spin_project>
const bool quda::Pack< Float, nColor, spin_project >::dagger
protected

Definition at line 53 of file dslash_pack2.cu.

Referenced by quda::PackGhost().

◆ ghost

template<typename Float, int nColor, bool spin_project>
void** quda::Pack< Float, nColor, spin_project >::ghost
protected

Definition at line 49 of file dslash_pack2.cu.

Referenced by quda::PackGhost().

◆ in

template<typename Float, int nColor, bool spin_project>
const ColorSpinorField& quda::Pack< Float, nColor, spin_project >::in
protected

Definition at line 50 of file dslash_pack2.cu.

Referenced by quda::PackGhost().

◆ location

template<typename Float, int nColor, bool spin_project>
MemoryLocation quda::Pack< Float, nColor, spin_project >::location
protected

Definition at line 51 of file dslash_pack2.cu.

Referenced by quda::PackGhost().

◆ nFace

template<typename Float, int nColor, bool spin_project>
const int quda::Pack< Float, nColor, spin_project >::nFace
protected

Definition at line 52 of file dslash_pack2.cu.

Referenced by quda::PackGhost().

◆ nParity

template<typename Float, int nColor, bool spin_project>
const int quda::Pack< Float, nColor, spin_project >::nParity
protected

Definition at line 55 of file dslash_pack2.cu.

◆ parity

template<typename Float, int nColor, bool spin_project>
const int quda::Pack< Float, nColor, spin_project >::parity
protected

Definition at line 54 of file dslash_pack2.cu.

Referenced by quda::PackGhost().

◆ threads

template<typename Float, int nColor, bool spin_project>
int quda::Pack< Float, nColor, spin_project >::threads
protected

◆ twist

template<typename Float, int nColor, bool spin_project>
int quda::Pack< Float, nColor, spin_project >::twist
protected

Definition at line 60 of file dslash_pack2.cu.


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