diff options
| author | Michalis Spyrou <michalis.spyrou@arm.com> | 2020-07-02 17:39:25 +0100 |
|---|---|---|
| committer | Michalis Spyrou <michalis.spyrou@arm.com> | 2020-07-07 09:14:00 +0000 |
| commit | 6eb73458c4869165c88d33c6a745a91cdc73a36a (patch) | |
| tree | 1f22bd141f420ad4e2906939bb4abf11fec3aea3 /src/runtime/NEON/functions/NENormalizationLayer.cpp | |
| parent | 1fad814022ba98506ba30b2e25601985e7ec5259 (diff) | |
| download | ComputeLibrary-6eb73458c4869165c88d33c6a745a91cdc73a36a.tar.gz | |
COMPMID-3373: Async support to NEArithmetic* kernels/functions (Pt. 2)
Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com>
Change-Id: Iec06adb535aaf7efb1838d921e8d6bb978b7b215
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3498
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src/runtime/NEON/functions/NENormalizationLayer.cpp')
| -rw-r--r-- | src/runtime/NEON/functions/NENormalizationLayer.cpp | 9 |
1 files changed, 4 insertions, 5 deletions
diff --git a/src/runtime/NEON/functions/NENormalizationLayer.cpp b/src/runtime/NEON/functions/NENormalizationLayer.cpp index f3a3ac6322..ab8cb656bd 100644 --- a/src/runtime/NEON/functions/NENormalizationLayer.cpp +++ b/src/runtime/NEON/functions/NENormalizationLayer.cpp @@ -33,7 +33,7 @@ namespace arm_compute { NENormalizationLayer::NENormalizationLayer(std::shared_ptr<IMemoryManager> memory_manager) - : _memory_group(std::move(memory_manager)), _norm_kernel(), _multiply_kernel(), _input_squared() + : _memory_group(std::move(memory_manager)), _norm_kernel(), _multiply_f(), _input_squared() { } @@ -49,7 +49,7 @@ void NENormalizationLayer::configure(const ITensor *input, ITensor *output, cons // Configure kernels _norm_kernel.configure(input, &_input_squared, output, norm_info); - _multiply_kernel.configure(input, input, &_input_squared, 1.0f, ConvertPolicy::SATURATE, RoundingPolicy::TO_ZERO); + _multiply_f.configure(input, input, &_input_squared, 1.0f, ConvertPolicy::SATURATE, RoundingPolicy::TO_ZERO); // Allocate the tensor once the configure methods have been called _input_squared.allocator()->allocate(); @@ -61,7 +61,7 @@ Status NENormalizationLayer::validate(const ITensorInfo *input, const ITensorInf ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(input, output); ARM_COMPUTE_RETURN_ON_ERROR(NENormalizationLayerKernel::validate(input, input, output, norm_info)); - ARM_COMPUTE_RETURN_ON_ERROR(NEPixelWiseMultiplicationKernel::validate(input, input, output, 1.0f, ConvertPolicy::SATURATE, RoundingPolicy::TO_ZERO)); + ARM_COMPUTE_RETURN_ON_ERROR(NEPixelWiseMultiplication::validate(input, input, output, 1.0f, ConvertPolicy::SATURATE, RoundingPolicy::TO_ZERO)); return Status{}; } @@ -69,8 +69,7 @@ Status NENormalizationLayer::validate(const ITensorInfo *input, const ITensorInf void NENormalizationLayer::run() { MemoryGroupResourceScope scope_mg(_memory_group); - - NEScheduler::get().schedule(&_multiply_kernel, Window::DimY); + _multiply_f.run(); NEScheduler::get().schedule(&_norm_kernel, Window::DimY); } }
\ No newline at end of file |