diff options
author | Georgios Pinitas <georgios.pinitas@arm.com> | 2021-06-25 12:13:49 +0100 |
---|---|---|
committer | Georgios Pinitas <georgios.pinitas@arm.com> | 2021-06-29 16:26:41 +0000 |
commit | 4a578b923ed000c67fe0bc1433f945aea634ca9c (patch) | |
tree | b7bb041d2e7bfb4b909199f1b889585d237c665d /arm_compute/runtime/CL/functions/CLGEMMLowpMatrixMultiplyCore.h | |
parent | 53832b2bcce44c71fe31a618a81765294df55750 (diff) | |
download | ComputeLibrary-4a578b923ed000c67fe0bc1433f945aea634ca9c.tar.gz |
Port the ClGemmLowp kernels to the new API
Ported kernels:
- CLGEMMLowpMatrixMultiplyNativeKernel
- CLGEMMLowpMatrixMultiplyReshapedKernel
- CLGEMMLowpMatrixMultiplyReshapedOnlyRHSKernel
- CLGEMMLowpOffsetContributionKernel
- CLGEMMLowpOffsetContributionOutputStageKernel
- CLGEMMLowpQuantizeDownInt32ScaleByFixedPointKernel
- CLGEMMLowpQuantizeDownInt32ScaleByFloatKernel
- CLGEMMLowpQuantizeDownInt32ScaleKernel
Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com>
Change-Id: I9d5a744d6a2dd2f2726fdfb291bad000b6970de2
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5870
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'arm_compute/runtime/CL/functions/CLGEMMLowpMatrixMultiplyCore.h')
-rw-r--r-- | arm_compute/runtime/CL/functions/CLGEMMLowpMatrixMultiplyCore.h | 31 |
1 files changed, 16 insertions, 15 deletions
diff --git a/arm_compute/runtime/CL/functions/CLGEMMLowpMatrixMultiplyCore.h b/arm_compute/runtime/CL/functions/CLGEMMLowpMatrixMultiplyCore.h index 3d2dbdb104..e62db8e644 100644 --- a/arm_compute/runtime/CL/functions/CLGEMMLowpMatrixMultiplyCore.h +++ b/arm_compute/runtime/CL/functions/CLGEMMLowpMatrixMultiplyCore.h @@ -34,17 +34,17 @@ class CLCompileContext; class IMemoryManager; class ICLTensor; class ITensorInfo; -class CLGEMMLowpMatrixMultiplyNativeKernel; -class CLGEMMLowpMatrixMultiplyReshapedOnlyRHSKernel; -class CLGEMMLowpOffsetContributionKernel; -class CLGEMMLowpOffsetContributionOutputStageKernel; -class CLGEMMLowpMatrixAReductionKernel; -class CLGEMMLowpMatrixBReductionKernel; namespace opencl { namespace kernels { class ClGemmReshapeRhsMatrixKernel; +class ClGemmLowpMatrixMultiplyNativeKernel; +class ClGemmLowpMatrixMultiplyReshapedOnlyRhsKernel; +class ClGemmLowpOffsetContributionKernel; +class ClGemmLowpOffsetContributionOutputStageKernel; +class ClGemmLowpMatrixAReductionKernel; +class ClGemmLowpMatrixBReductionKernel; } // namespace kernels } // namespace opencl @@ -150,14 +150,14 @@ private: MemoryGroup _memory_group; // Kernels used - std::unique_ptr<opencl::kernels::ClCastKernel> _weights_to_qasymm8; - std::unique_ptr<CLGEMMLowpMatrixMultiplyNativeKernel> _mm_native_kernel; - std::unique_ptr<CLGEMMLowpMatrixMultiplyReshapedOnlyRHSKernel> _mm_reshaped_only_rhs_kernel; - std::unique_ptr<opencl::kernels::ClGemmReshapeRhsMatrixKernel> _mtx_b_reshape_kernel; - std::unique_ptr<CLGEMMLowpMatrixAReductionKernel> _mtx_a_reduction_kernel; - std::unique_ptr<CLGEMMLowpMatrixBReductionKernel> _mtx_b_reduction_kernel; - std::unique_ptr<CLGEMMLowpOffsetContributionKernel> _offset_contribution_kernel; - std::unique_ptr<CLGEMMLowpOffsetContributionOutputStageKernel> _offset_contribution_output_stage_kernel; + std::unique_ptr<opencl::kernels::ClCastKernel> _weights_to_qasymm8; + std::unique_ptr<opencl::kernels::ClGemmLowpMatrixMultiplyNativeKernel> _mm_native_kernel; + std::unique_ptr<opencl::kernels::ClGemmLowpMatrixMultiplyReshapedOnlyRhsKernel> _mm_reshaped_only_rhs_kernel; + std::unique_ptr<opencl::kernels::ClGemmReshapeRhsMatrixKernel> _mtx_b_reshape_kernel; + std::unique_ptr<opencl::kernels::ClGemmLowpMatrixAReductionKernel> _mtx_a_reduction_kernel; + std::unique_ptr<opencl::kernels::ClGemmLowpMatrixBReductionKernel> _mtx_b_reduction_kernel; + std::unique_ptr<opencl::kernels::ClGemmLowpOffsetContributionKernel> _offset_contribution_kernel; + std::unique_ptr<opencl::kernels::ClGemmLowpOffsetContributionOutputStageKernel> _offset_contribution_output_stage_kernel; // Temporary tensors CLTensor _qasymm8_weights; @@ -171,7 +171,8 @@ private: // Tensor pointers const ICLTensor *_matrix_a; const ICLTensor *_original_b; - const ICLTensor *_output; + const ICLTensor *_c; + ICLTensor *_output; int32_t _a_offset; int32_t _b_offset; |