diff options
author | Michele Di Giorgio <michele.digiorgio@arm.com> | 2021-06-21 14:45:44 +0100 |
---|---|---|
committer | Michele Di Giorgio <michele.digiorgio@arm.com> | 2021-06-29 15:58:33 +0000 |
commit | 53832b2bcce44c71fe31a618a81765294df55750 (patch) | |
tree | 2165c4703891202186df55b772353bab8606315d /arm_compute/runtime/NEON/functions | |
parent | 72b56875b9bb30a9ed1d2ad38ec51fc88e435c35 (diff) | |
download | ComputeLibrary-53832b2bcce44c71fe31a618a81765294df55750.tar.gz |
Port NEGEMM to memory injecting interface (Part 2)
- Port NEGEMMMatrixMultiplyKernel to the new API
Partially resolves: COMPMID-4402
Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Change-Id: I52b67055dc24bb3a417d6ec5aeeee86e21b74320
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5873
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'arm_compute/runtime/NEON/functions')
-rw-r--r-- | arm_compute/runtime/NEON/functions/NEFullyConnectedLayer.h | 2 | ||||
-rw-r--r-- | arm_compute/runtime/NEON/functions/NEGEMM.h | 2 |
2 files changed, 2 insertions, 2 deletions
diff --git a/arm_compute/runtime/NEON/functions/NEFullyConnectedLayer.h b/arm_compute/runtime/NEON/functions/NEFullyConnectedLayer.h index 22ec9e0fec..e409a61ba1 100644 --- a/arm_compute/runtime/NEON/functions/NEFullyConnectedLayer.h +++ b/arm_compute/runtime/NEON/functions/NEFullyConnectedLayer.h @@ -79,7 +79,7 @@ private: /** Basic function to compute a Fully Connected layer. This function calls the following kernels: * -# @ref NEIm2ColKernel (called when the input comes from a convolutional layer) * -# @ref NETranspose (if @p are_weights_reshaped is set to false and transpose_weights is set to true ) (called once) - * -# @ref NEGEMMMatrixMultiplyKernel or @ref NEGEMMLowpMatrixMultiplyCore (if quantized asymmetric) + * -# @ref NEGEMM or @ref NEGEMMLowpMatrixMultiplyCore (if quantized asymmetric) * -# @ref cpu::kernels::CpuGemmMatrixAdditionKernel or @ref NEGEMMLowpOutputStage (if quantized asymmetric) (if @p biases is not equal to nullptr) * * @note The fully connected layer accepts "weights" tensors only with 2 dimensions. diff --git a/arm_compute/runtime/NEON/functions/NEGEMM.h b/arm_compute/runtime/NEON/functions/NEGEMM.h index c1ae11bcbf..5daa0406a5 100644 --- a/arm_compute/runtime/NEON/functions/NEGEMM.h +++ b/arm_compute/runtime/NEON/functions/NEGEMM.h @@ -42,7 +42,7 @@ namespace arm_compute * Else: * -# @ref cpu::kernels::CpuGemmInterleave4x4Kernel (if the output tensor is a matrix) * -# @ref cpu::kernels::CpuGemmTranspose1xWKernel (if the output tensor is a matrix) - * -# @ref NEGEMMMatrixMultiplyKernel + * -# @ref cpu::kernels::CpuGemmMatrixMultiplyKernel * In both cases: * -# @ref cpu::kernels::CpuGemmMatrixAdditionKernel (if c != nullptr and beta != 0.0 and is not reshaped once) * Else: |