Compute Library
 21.11
Public Member Functions | Static Public Attributes
ICPPKernel Class Referenceabstract

Common interface for all kernels implemented in C++. More...

#include <ICPPKernel.h>

Collaboration diagram for ICPPKernel:
[legend]

Public Member Functions

virtual ~ICPPKernel ()=default
 Default destructor. More...
 
virtual void run (const Window &window, const ThreadInfo &info)
 Execute the kernel on the passed window. More...
 
virtual void run_nd (const Window &window, const ThreadInfo &info, const Window &thread_locator)
 legacy compatibility layer for implemantions which do not support thread_locator In these cases we simply narrow the interface down the legacy version More...
 
virtual void run_op (ITensorPack &tensors, const Window &window, const ThreadInfo &info)
 Execute the kernel on the passed window. More...
 
virtual size_t get_mws (const CPUInfo &platform, size_t thread_count) const
 Return minimum workload size of the relevant kernel. More...
 
virtual const char * name () const =0
 Name of the kernel. More...
 
- Public Member Functions inherited from IKernel
 IKernel ()
 Constructor. More...
 
virtual ~IKernel ()=default
 Destructor. More...
 
virtual bool is_parallelisable () const
 Indicates whether or not the kernel is parallelisable. More...
 
virtual BorderSize border_size () const
 The size of the border for that kernel. More...
 
const Window & window () const
 The maximum window the kernel can be executed on. More...
 
bool is_window_configured () const
 Function to check if the embedded window of this kernel has been configured. More...
 

Static Public Attributes

static constexpr size_t default_mws = 128
 
static constexpr size_t small_network_mws = 256
 

Detailed Description

Common interface for all kernels implemented in C++.

Definition at line 38 of file ICPPKernel.h.

Constructor & Destructor Documentation

◆ ~ICPPKernel()

virtual ~ICPPKernel ( )
virtualdefault

Default destructor.

Member Function Documentation

◆ get_mws()

virtual size_t get_mws ( const CPUInfo &  platform,
size_t  thread_count 
) const
inlinevirtual

Return minimum workload size of the relevant kernel.

Parameters
[in]platformThe CPU platform used to create the context.
[in]thread_countNumber of threads in the execution.
Returns
[out] mws Minimum workload size for requsted configuration.

Reimplemented in CpuGemmAssemblyWrapperKernel< TypeInput, TypeOutput >, CpuDepthwiseConv2dAssemblyWrapperKernel, CpuIm2ColKernel, NEPadLayerKernel, CpuAddKernel, CpuReshapeKernel, and CpuActivationKernel.

Definition at line 101 of file ICPPKernel.h.

References ARM_COMPUTE_UNUSED, ICPPKernel::default_mws, and ICPPKernel::name().

102  {
103  ARM_COMPUTE_UNUSED(platform, thread_count);
104 
105  return default_mws;
106  }
ARM_COMPUTE_UNUSED
#define ARM_COMPUTE_UNUSED(...)
To avoid unused variables warnings.
Definition: Error.h:152
arm_compute::ICPPKernel::default_mws
static constexpr size_t default_mws
Definition: ICPPKernel.h:41

◆ name()

virtual const char* name ( ) const
pure virtual

Name of the kernel.

Returns
Kernel name

Implemented in CpuWinogradConv2dTransformWeightsKernel< T, OutputTileRows, OutputTileCols, KernelRows, KernelCols >, CpuWinogradConv2dTransformOutputKernel< T, OutputTileRows, OutputTileCols, KernelRows, KernelCols >, CpuWinogradConv2dTransformInputKernel< T, OutputTileRows, OutputTileCols, KernelRows, KernelCols >, CpuComplexMulKernel, CpuGemmLowpMatrixBReductionKernel, CpuGemmLowpOffsetContributionOutputStageKernel, CpuLogits1DSoftmaxKernel< IS_LOG >, CpuIm2ColKernel, CpuGemmTranspose1xWKernel, CpuWeightsReshapeKernel, CpuGemmLowpQuantizeDownInt32ToInt8ScaleByFixedPointKernel, CpuGemmLowpQuantizeDownInt32ToUint8ScaleByFixedPointKernel, CpuMulKernel, CpuCol2ImKernel, CpuGemmLowpQuantizeDownInt32ToInt16ScaleByFixedPointKernel, CpuGemmLowpQuantizeDownInt32ScaleKernel, CpuDepthwiseConv2dAssemblyWrapperKernel, CpuGemmLowpOffsetContributionKernel, CpuGemmInterleave4x4Kernel, CpuCastKernel, CpuDepthwiseConv2dNativeKernel, CpuDirectConv3dKernel, CpuGemmLowpMatrixMultiplyKernel, CpuActivationKernel, CpuAddKernel, CpuDirectConv2dOutputStageKernel, CpuGemmMatrixMultiplyKernel, CpuSubKernel, CpuGemmLowpMatrixAReductionKernel, CpuGemmMatrixAdditionKernel, CpuGemmAssemblyWrapperKernel< TypeInput, TypeOutput >, CpuConcatenateDepthKernel, CpuDirectConv2dKernel, CpuScaleKernel, CpuConvertFullyConnectedWeightsKernel, CpuFloorKernel, NELogicalKernel, CpuElementwiseUnaryKernel, CpuPool2dKernel, CpuQuantizeKernel, CpuConcatenateHeightKernel, CpuConcatenateWidthKernel, CpuConcatenateBatchKernel, CpuPermuteKernel, CpuCopyKernel, CpuReshapeKernel, CpuConvertQuantizedSignednessKernel, CpuDequantizeKernel, CpuLogits1DMaxKernel, CpuTransposeKernel, NEGatherKernel, CpuElementwiseKernel, NECol2ImKernel, CpuPool2dAssemblyWrapperKernel, CpuFillKernel, NETileKernel, NESelectKernel, NEReductionOperationKernel, NEFFTRadixStageKernel, NERangeKernel, CPPNonMaximumSuppressionKernel, NEBitwiseAndKernel, NEBitwiseNotKernel, NEBitwiseOrKernel, NEBitwiseXorKernel, NEMeanStdDevNormalizationKernel, NEStridedSliceKernel, CPPPermuteKernel, NECropKernel, NEFFTScaleKernel, NEFillBorderKernel, CPPUpsampleKernel, NEBatchNormalizationLayerKernel, NEFFTDigitReverseKernel, NESpaceToBatchLayerKernel, NEStackLayerKernel, CPPBoxWithNonMaximaSuppressionLimitKernel, NEChannelShuffleLayerKernel, NEDepthToSpaceLayerKernel, NEFuseBatchNormalizationKernel, NEInstanceNormalizationLayerKernel, NENormalizationLayerKernel, NEQLSTMLayerNormalizationKernel, NERemapKernel, NEReorgLayerKernel, NEROIAlignLayerKernel, NESpaceToDepthLayerKernel, CPPTopKVKernel, NEBatchToSpaceLayerKernel, NEBoundingBoxTransformKernel, NEL2NormalizeLayerKernel, NEMaxUnpoolingLayerKernel, NEPadLayerKernel, NEPriorBoxLayerKernel, NEReverseKernel, NEComputeAllAnchorsKernel, and NEROIPoolingLayerKernel.

Referenced by ICPPKernel::get_mws(), and SchedulerClock< output_timestamps >::id().

◆ run()

virtual void run ( const Window &  window,
const ThreadInfo &  info 
)
inlinevirtual

Execute the kernel on the passed window.

Warning
If is_parallelisable() returns false then the passed window must be equal to window()
Note
The window has to be a region within the window returned by the window() method
The width of the window has to be a multiple of num_elems_processed_per_iteration().
Parameters
[in]windowRegion on which to execute the kernel. (Must be a region of the window returned by window())
[in]infoInfo about executing thread and CPU.

Reimplemented in NESpaceToBatchLayerKernel, NEFuseBatchNormalizationKernel, NEBatchNormalizationLayerKernel, NECol2ImKernel, NECropKernel, NEBatchToSpaceLayerKernel, NEROIAlignLayerKernel, CPPNonMaximumSuppressionKernel, NEStackLayerKernel, NEBoundingBoxTransformKernel, NEFFTRadixStageKernel, NEReductionOperationKernel, NESelectKernel, NEFillBorderKernel, NEGatherKernel, NEL2NormalizeLayerKernel, NEPadLayerKernel, NEMeanStdDevNormalizationKernel, NEMaxUnpoolingLayerKernel, NENormalizationLayerKernel, NERangeKernel, NEFFTDigitReverseKernel, CPPPermuteKernel, CPPTopKVKernel, NEReorgLayerKernel, NEFFTScaleKernel, NEComputeAllAnchorsKernel, NEInstanceNormalizationLayerKernel, NEQLSTMLayerNormalizationKernel, CpuGemmAssemblyWrapperKernel< TypeInput, TypeOutput >, NEPriorBoxLayerKernel, NEChannelShuffleLayerKernel, NEDepthToSpaceLayerKernel, NEReverseKernel, NESpaceToDepthLayerKernel, NETileKernel, CPPBoxWithNonMaximaSuppressionLimitKernel, NEROIPoolingLayerKernel, NERemapKernel, NEBitwiseAndKernel, NEBitwiseOrKernel, NEBitwiseXorKernel, CPPUpsampleKernel, and NEBitwiseNotKernel.

Definition at line 58 of file ICPPKernel.h.

References ARM_COMPUTE_ERROR, and ARM_COMPUTE_UNUSED.

Referenced by IScheduler::num_threads_hint(), ICPPKernel::run_nd(), and SingleThreadScheduler::schedule().

59  {
60  ARM_COMPUTE_UNUSED(window, info);
61  ARM_COMPUTE_ERROR("default implementation of legacy run() virtual member function invoked");
62  }
arm_compute::IKernel::window
const Window & window() const
The maximum window the kernel can be executed on.
Definition: IKernel.cpp:28
ARM_COMPUTE_ERROR
#define ARM_COMPUTE_ERROR(msg)
Print the given message then throw an std::runtime_error.
Definition: Error.h:352
ARM_COMPUTE_UNUSED
#define ARM_COMPUTE_UNUSED(...)
To avoid unused variables warnings.
Definition: Error.h:152
arm_compute::test::validation::info
ScaleKernelInfo info(interpolation_policy, default_border_mode, PixelValue(), sampling_policy, false)

◆ run_nd()

virtual void run_nd ( const Window &  window,
const ThreadInfo &  info,
const Window &  thread_locator 
)
inlinevirtual

legacy compatibility layer for implemantions which do not support thread_locator In these cases we simply narrow the interface down the legacy version

Parameters
[in]windowRegion on which to execute the kernel. (Must be a region of the window returned by window())
[in]infoInfo about executing thread and CPU.
[in]thread_locatorSpecifies "where" the current thread is in the multi-dimensional space

Reimplemented in CpuGemmAssemblyWrapperKernel< TypeInput, TypeOutput >.

Definition at line 71 of file ICPPKernel.h.

References ARM_COMPUTE_UNUSED, and ICPPKernel::run().

Referenced by IScheduler::num_threads_hint().

72  {
73  ARM_COMPUTE_UNUSED(thread_locator);
74  run(window, info);
75  }
arm_compute::IKernel::window
const Window & window() const
The maximum window the kernel can be executed on.
Definition: IKernel.cpp:28
ARM_COMPUTE_UNUSED
#define ARM_COMPUTE_UNUSED(...)
To avoid unused variables warnings.
Definition: Error.h:152
arm_compute::ICPPKernel::run
virtual void run(const Window &window, const ThreadInfo &info)
Execute the kernel on the passed window.
Definition: ICPPKernel.h:58
arm_compute::test::validation::info
ScaleKernelInfo info(interpolation_policy, default_border_mode, PixelValue(), sampling_policy, false)

◆ run_op()

virtual void run_op ( ITensorPack &  tensors,
const Window &  window,
const ThreadInfo &  info 
)
inlinevirtual

Execute the kernel on the passed window.

Warning
If is_parallelisable() returns false then the passed window must be equal to window()
Note
The window has to be a region within the window returned by the window() method
The width of the window has to be a multiple of num_elems_processed_per_iteration().
Parameters
[in]tensorsA vector containing the tensors to operate on.
[in]windowRegion on which to execute the kernel. (Must be a region of the window returned by window())
[in]infoInfo about executing thread and CPU.

Reimplemented in CpuWinogradConv2dTransformWeightsKernel< T, OutputTileRows, OutputTileCols, KernelRows, KernelCols >, CpuWinogradConv2dTransformOutputKernel< T, OutputTileRows, OutputTileCols, KernelRows, KernelCols >, CpuWinogradConv2dTransformInputKernel< T, OutputTileRows, OutputTileCols, KernelRows, KernelCols >, CpuComplexMulKernel, CpuGemmLowpMatrixBReductionKernel, CpuGemmLowpOffsetContributionOutputStageKernel, CpuLogits1DSoftmaxKernel< IS_LOG >, NEStridedSliceKernel, CpuIm2ColKernel, CpuGemmTranspose1xWKernel, CpuWeightsReshapeKernel, NEFillBorderKernel, CpuGemmLowpQuantizeDownInt32ToInt8ScaleByFixedPointKernel, CpuGemmLowpQuantizeDownInt32ToUint8ScaleByFixedPointKernel, CpuMulKernel, CpuPool2dAssemblyWrapperKernel, CpuCol2ImKernel, CpuGemmLowpQuantizeDownInt32ToInt16ScaleByFixedPointKernel, CpuGemmLowpQuantizeDownInt32ScaleKernel, CpuDepthwiseConv2dAssemblyWrapperKernel, CpuGemmLowpOffsetContributionKernel, CpuGemmInterleave4x4Kernel, CpuCastKernel, CpuDepthwiseConv2dNativeKernel, CpuDirectConv3dKernel, CpuGemmLowpMatrixMultiplyKernel, CpuActivationKernel, CpuAddKernel, CpuDirectConv2dOutputStageKernel, CpuGemmMatrixMultiplyKernel, CpuSubKernel, CpuGemmLowpMatrixAReductionKernel, CpuGemmMatrixAdditionKernel, CpuConcatenateDepthKernel, CpuDirectConv2dKernel, CpuScaleKernel, CpuConvertFullyConnectedWeightsKernel, CpuFloorKernel, NELogicalKernel, CpuElementwiseUnaryKernel, CpuPool2dKernel, CpuQuantizeKernel, CpuConcatenateHeightKernel, CpuConcatenateWidthKernel, CpuConcatenateBatchKernel, CpuPermuteKernel, CpuCopyKernel, CpuReshapeKernel, CpuConvertQuantizedSignednessKernel, CpuDequantizeKernel, CpuLogits1DMaxKernel, CpuTransposeKernel, CpuElementwiseKernel, and CpuFillKernel.

Definition at line 89 of file ICPPKernel.h.

References ARM_COMPUTE_UNUSED.

Referenced by CpuWinogradConv2dTransformInputKernel< T, OutputTileRows, OutputTileCols, KernelRows, KernelCols >::name(), CpuWinogradConv2dTransformOutputKernel< T, OutputTileRows, OutputTileCols, KernelRows, KernelCols >::name(), CpuWinogradConv2dTransformWeightsKernel< T, OutputTileRows, OutputTileCols, KernelRows, KernelCols >::name(), IScheduler::num_threads_hint(), SingleThreadScheduler::schedule_op(), and OMPScheduler::schedule_op().

90  {
91  ARM_COMPUTE_UNUSED(tensors, window, info);
92  }
arm_compute::IKernel::window
const Window & window() const
The maximum window the kernel can be executed on.
Definition: IKernel.cpp:28
ARM_COMPUTE_UNUSED
#define ARM_COMPUTE_UNUSED(...)
To avoid unused variables warnings.
Definition: Error.h:152
arm_compute::test::validation::info
ScaleKernelInfo info(interpolation_policy, default_border_mode, PixelValue(), sampling_policy, false)

Field Documentation

◆ default_mws

constexpr size_t default_mws = 128
static

Definition at line 41 of file ICPPKernel.h.

Referenced by ICPPKernel::get_mws().

◆ small_network_mws

constexpr size_t small_network_mws = 256
static

Definition at line 42 of file ICPPKernel.h.

Referenced by CpuActivationKernel::get_mws(), CpuReshapeKernel::get_mws(), CpuAddKernel::get_mws(), NEPadLayerKernel::get_mws(), CpuIm2ColKernel::get_mws(), CpuDepthwiseConv2dAssemblyWrapperKernel::get_mws(), CpuGemmAssemblyWrapperKernel< TypeInput, TypeOutput >::get_mws(), and CpuPool2dAssemblyWrapperKernel::is_configured().

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