diff options
| author | Michael Tyler <michael.tyler@arm.com> | 2023-06-30 11:26:05 +0100 |
|---|---|---|
| committer | michael.tyler <michael.tyler@arm.com> | 2023-07-04 14:34:58 +0000 |
| commit | 8deee9bd9b9137c256c23b86be11dbf0466f3aa8 (patch) | |
| tree | ac80b3bdd992552b65e306b77f061484da0591ca /src/core/NEON/kernels/arm_conv/depthwise/depthwise_planar.hpp | |
| parent | 19844f605f5e5b71d05164711dee13f8652adafe (diff) | |
| download | ComputeLibrary-8deee9bd9b9137c256c23b86be11dbf0466f3aa8.tar.gz | |
Depthwise channel pre-multiplication
Resolves: COMPMID-6337
Change-Id: Ie9097b3f56e8071426c621386a5988bd7f7e8ef2
Signed-off-by: Michael Tyler <michael.tyler@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/9852
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Viet-Hoa Do <viet-hoa.do@arm.com>
Benchmark: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src/core/NEON/kernels/arm_conv/depthwise/depthwise_planar.hpp')
| -rw-r--r-- | src/core/NEON/kernels/arm_conv/depthwise/depthwise_planar.hpp | 4 |
1 files changed, 2 insertions, 2 deletions
diff --git a/src/core/NEON/kernels/arm_conv/depthwise/depthwise_planar.hpp b/src/core/NEON/kernels/arm_conv/depthwise/depthwise_planar.hpp index 567eab13f3..c3daaf04fe 100644 --- a/src/core/NEON/kernels/arm_conv/depthwise/depthwise_planar.hpp +++ b/src/core/NEON/kernels/arm_conv/depthwise/depthwise_planar.hpp @@ -153,7 +153,7 @@ class PlanarStrategy : public IPlanarStrategy<OutputStage> { return interleaves::PackingArguments( m_kernel_rows, m_kernel_cols, sizeof(TWeight), - false, sizeof(TAccum), // Don't pack the bias + false, sizeof(TAccum), true, // Don't pack the bias m_vl_type, sizeof(TAccum), 1, // Accumulator depth of 1 TODO [this] (unsigned int idx, unsigned int &x, unsigned int &y) -> bool { return this->get_kernel_packing_point(idx, x, y); } @@ -276,7 +276,7 @@ class DepthwisePlanar : public DepthwiseCommon<TInput, TWeight, TOutput> depthwise_depthfirst::stash_bias(this->m_os, biases); } - size_t get_working_size(unsigned int n_threads, unsigned int) const override + size_t get_working_size(unsigned int n_threads) const override { return this->get_working_size_per_thread() * n_threads; } |