diff options
| author | SiCongLi <sicong.li@arm.com> | 2021-10-24 19:12:33 +0100 |
|---|---|---|
| committer | SiCong Li <sicong.li@arm.com> | 2021-11-02 10:41:11 +0000 |
| commit | afa19725f7f3feb2c21a6aed02ade49d08e3097b (patch) | |
| tree | d796239f542cf447060e6fa6d1240fecb3d6c7a6 /src/core/CL/cl_kernels/common/gemm.cl | |
| parent | 579ca84bd8ef5a91eded65c4dc5e0b9f7de8bef1 (diff) | |
| download | ComputeLibrary-afa19725f7f3feb2c21a6aed02ade49d08e3097b.tar.gz | |
Add post ops to ClGemmMatrixMultiplyReshapedOnlyRHSKernel and ClGemmMatrixMultiplyNativeKernel Part 3
Partially resolves: COMPMID-4435
Change-Id: Ifc5affa3a24a70942ca2d001380205df09b03ad7
Signed-off-by: SiCongLi <sicong.li@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6550
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src/core/CL/cl_kernels/common/gemm.cl')
| -rw-r--r-- | src/core/CL/cl_kernels/common/gemm.cl | 5 |
1 files changed, 5 insertions, 0 deletions
diff --git a/src/core/CL/cl_kernels/common/gemm.cl b/src/core/CL/cl_kernels/common/gemm.cl index 9732588b31..6502dd496a 100644 --- a/src/core/CL/cl_kernels/common/gemm.cl +++ b/src/core/CL/cl_kernels/common/gemm.cl @@ -1000,6 +1000,7 @@ __kernel void gemm_reshape_rhs_matrix_t(TENSOR3D_DECLARATION(src), /** This OpenCL kernel computes the matrix multiplication between 2 matrices. * The LHS matrix is NOT reshaped * The RHS is reshaped with @ref CLGEMMReshapeRHSMatrixKernel and the block K0xN0 is transposed + * @note This kernel is duplicated in /experimental/gemm_fused_post_ops/act_eltwise_op_act/gemm_mm_reshaped_only_rhs.cl * * @note If the first two dimensions of NDRange have been dispatched with "dummy_work_items" support, the option -DDUMMY_WORK_ITEMS must be passed at compile time. * @note The GEMM's dimensions (M,N and K) must be passed at compile time using -DM, -DN and and -DK (e.g. -DM=52, -DN=30 and -DK=90) @@ -1294,6 +1295,7 @@ __kernel void gemm_mm_reshaped_only_rhs_t(IMAGE_DECLARATION(lhs), /** This OpenCL kernel computes the matrix multiplication between 2 matrices. The RHS matrix is stored in OpenCL image * The LHS matrix is NOT reshaped * The RHS is reshaped with @ref CLGEMMReshapeRHSMatrixKernel and the block K0xN0 is transposed + * @note This kernel is duplicated in /experimental/gemm_fused_post_ops/act_eltwise_op_act/gemm_mm_reshaped_only_rhs.cl * * @note -DOPENCL_IMAGE_SUPPORT must be passed at compile time in order to compile this OpenCL kernel * @note If the first two dimensions of NDRange have been dispatched with "dummy_work_items" support, the option -DDUMMY_WORK_ITEMS must be passed at compile time. @@ -1720,6 +1722,7 @@ __kernel void gemm_mm_reshaped_only_rhs_t_texture(IMAGE_DECLARATION(lhs), /** This OpenCL kernel computes the matrix multiplication between 2 matrices. * The LHS matrix is NOT reshaped * The RHS is reshaped with @ref CLGEMMReshapeRHSMatrixKernel and the block K0xN0 is NOT transposed + * @note This kernel is duplicated in /experimental/gemm_fused_post_ops/act_eltwise_op_act/gemm_mm_reshaped_only_rhs.cl * * @note If the first two dimensions of NDRange have been dispatched with "dummy_work_items" support, the option -DDUMMY_WORK_ITEMS must be passed at compile time. * @note The GEMM's dimensions (M,N and K) must be passed at compile time using -DM, -DN and and -DK (e.g. -DM=52, -DN=30 and -DK=90). @@ -2038,6 +2041,7 @@ __kernel void gemm_mm_reshaped_only_rhs_nt(IMAGE_DECLARATION(lhs), /** This OpenCL kernel computes the matrix multiplication between 2 matrices. * The LHS matrix is NOT reshaped * The RHS is reshaped with @ref CLGEMMReshapeRHSMatrixKernel and the block K0xN0 is NOT transposed + * @note This kernel is duplicated in /experimental/gemm_fused_post_ops/act_eltwise_op_act/gemm_mm_reshaped_only_rhs.cl * * @note -DOPENCL_IMAGE_SUPPORT must be passed at compile time in order to compile this OpenCL kernel * @note If the first two dimensions of NDRange have been dispatched with "dummy_work_items" support, the option -DDUMMY_WORK_ITEMS must be passed at compile time. @@ -4025,6 +4029,7 @@ __kernel void gemm_mm_reshaped_lhs_t_rhs_nt_texture(IMAGE_DECLARATION(lhs), /** This OpenCL kernel computes the matrix multiplication between 2 matrices. * The LHS matrix is NOT reshaped * The RHS matrix is NOT reshaped + * @note This kernel is duplicated in /experimental/gemm_fused_post_ops/act_eltwise_op_act/gemm_mm_native.cl * * @note If the first two dimensions of NDRange have been dispatched with "dummy_work_items" support, the option -DDUMMY_WORK_ITEMS must be passed at compile time. * @note The GEMM's dimensions (M,N and K) must be passed at compile time using -DM, -DN and and -DK (e.g. -DM=52, -DN=30 and -DK=90) |