diff options
author | Gian Marco <gianmarco.iodice@arm.com> | 2017-12-16 19:33:50 +0000 |
---|---|---|
committer | Anthony Barbier <anthony.barbier@arm.com> | 2018-11-02 16:42:33 +0000 |
commit | 1d25ed54a948639d1894c8b021940df70005d519 (patch) | |
tree | 96a29126c5b61299d64496fad7f6844412ab2cca /arm_compute/runtime/CL/functions/CLConvolutionLayer.h | |
parent | 57b20109108a90113d29d21ce7d3c873ff19749c (diff) | |
download | ComputeLibrary-1d25ed54a948639d1894c8b021940df70005d519.tar.gz |
COMPMID-759 - CLGEMM optimization for McVail benchmarks
This patch introduces an optimization for CLGEMM on Bifrost
architectures which can bring to 40% of FMA utilization on
config 3 of McVail. The new CLGEMM does not require any reshape of
matrix A and matrix B.
This patch also adds the auto-config in CLConvolutionLayer and CLGEMM
and extends the interface for NEGEMM and CLGEMM.
Change-Id: Ibb354eda45e9ca64b14a99700fb21dff5989dda9
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/113716
Tested-by: Jenkins <bsgcomp@arm.com>
Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>
Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
Diffstat (limited to 'arm_compute/runtime/CL/functions/CLConvolutionLayer.h')
-rw-r--r-- | arm_compute/runtime/CL/functions/CLConvolutionLayer.h | 3 |
1 files changed, 1 insertions, 2 deletions
diff --git a/arm_compute/runtime/CL/functions/CLConvolutionLayer.h b/arm_compute/runtime/CL/functions/CLConvolutionLayer.h index a8a04a0bbf..3fe6604db9 100644 --- a/arm_compute/runtime/CL/functions/CLConvolutionLayer.h +++ b/arm_compute/runtime/CL/functions/CLConvolutionLayer.h @@ -138,10 +138,9 @@ private: CLTensor _gemm_output; CLTensor _tmp_output; - bool _append_bias; - bool _is_fully_connected_convolution; bool _are_weights_reshaped; bool _is_quantized; + bool _is_interleaved_transposed; }; } #endif /* __ARM_COMPUTE_CLCONVOLUTIONLAYER_H__ */ |