diff options
| author | Ramy Elgammal <ramy.elgammal@arm.com> | 2022-11-30 16:23:10 +0000 |
|---|---|---|
| committer | Ramy Elgammal <ramy.elgammal@arm.com> | 2022-12-09 13:57:49 +0000 |
| commit | df6a3b05842a98702437347ca269138ccd55f852 (patch) | |
| tree | d38b3cc83acfa0aa492b953b6a3c06104e0d76fc /src/dynamic_fusion/sketch/gpu/components/cl/ClComponentDirectConv2d.h | |
| parent | 86689cdd95f634fb374f3875f62a4cb3408e1699 (diff) | |
| download | ComputeLibrary-df6a3b05842a98702437347ca269138ccd55f852.tar.gz | |
Use heuristics for setting dynamic fusion direct conv2d tile sizes
Resolves: COMPMID-5735
Change-Id: I9958413b69c5052cfa205dd0e9457cc4953aaf35
Signed-off-by: Ramy Elgammal <ramy.elgammal@arm.com>
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/c/VisualCompute/ComputeLibrary/+/474818
Tested-by: bsgcomp <bsgcomp@arm.com>
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Comments-Addressed: bsgcomp <bsgcomp@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8724
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Benchmark: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src/dynamic_fusion/sketch/gpu/components/cl/ClComponentDirectConv2d.h')
| -rw-r--r-- | src/dynamic_fusion/sketch/gpu/components/cl/ClComponentDirectConv2d.h | 11 |
1 files changed, 9 insertions, 2 deletions
diff --git a/src/dynamic_fusion/sketch/gpu/components/cl/ClComponentDirectConv2d.h b/src/dynamic_fusion/sketch/gpu/components/cl/ClComponentDirectConv2d.h index fec22b84a5..c3a70ef3ae 100644 --- a/src/dynamic_fusion/sketch/gpu/components/cl/ClComponentDirectConv2d.h +++ b/src/dynamic_fusion/sketch/gpu/components/cl/ClComponentDirectConv2d.h @@ -25,6 +25,7 @@ #define SRC_DYNAMIC_FUSION_SKETCH_GPU_COMPONENTS_CL_CLCOMPONENTDIRECTCONV2D #include "arm_compute/core/Error.h" +#include "arm_compute/core/KernelDescriptors.h" #include "src/dynamic_fusion/sketch/gpu/components/IGpuKernelComponent.h" #include <memory> @@ -56,9 +57,15 @@ public: /** Get fast_relaxed_math flag */ bool fast_relaxed_math() const; + /** Set direct convolution descriptor */ + ClComponentDirectConv2dSettings &direct_conv_descriptor(const DirectConvComputeKernelInfo &desc); + /** Get direct convolution descriptor */ + DirectConvComputeKernelInfo direct_conv_descriptor() const; + private: - bool _export_to_cl_image{ false }; - bool _fast_relaxed_math{ true }; + bool _export_to_cl_image{ false }; + bool _fast_relaxed_math{ true }; + DirectConvComputeKernelInfo _desc{}; // Direct convolution descriptor }; /** Forward declaration */ |