diff options
author | Gian Marco Iodice <gianmarco.iodice@arm.com> | 2020-05-07 10:26:15 +0100 |
---|---|---|
committer | Gian Marco Iodice <gianmarco.iodice@arm.com> | 2020-05-07 11:57:50 +0000 |
commit | 2886c757389c0ccca20a8689daf8180a730ecbc9 (patch) | |
tree | c54445b7bfb89818b80389315ffd373a4990a0fe | |
parent | 2cfd3f754ec707c25f129c0026686035a0dcfc54 (diff) | |
download | armcl-2886c757389c0ccca20a8689daf8180a730ecbc9.tar.gz armcl-2886c757389c0ccca20a8689daf8180a730ecbc9.tar.bz2 armcl-2886c757389c0ccca20a8689daf8180a730ecbc9.zip |
COMPMID-3459: Updated heuristic in CLGEMMReshapedOnlyRHSKernel
Change-Id: Ib958649d0193c148f45d72404da21d3f02dc5183
Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3154
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>
-rw-r--r-- | src/core/CL/gemm/reshaped_only_rhs/CLGEMMReshapedOnlyRHSKernelConfigurationValhall.cpp | 16 |
1 files changed, 15 insertions, 1 deletions
diff --git a/src/core/CL/gemm/reshaped_only_rhs/CLGEMMReshapedOnlyRHSKernelConfigurationValhall.cpp b/src/core/CL/gemm/reshaped_only_rhs/CLGEMMReshapedOnlyRHSKernelConfigurationValhall.cpp index d93a912c0..819a297a2 100644 --- a/src/core/CL/gemm/reshaped_only_rhs/CLGEMMReshapedOnlyRHSKernelConfigurationValhall.cpp +++ b/src/core/CL/gemm/reshaped_only_rhs/CLGEMMReshapedOnlyRHSKernelConfigurationValhall.cpp @@ -135,7 +135,21 @@ std::pair<GEMMLHSMatrixInfo, GEMMRHSMatrixInfo> CLGEMMReshapedOnlyRHSKernelConfi else { const int h0 = std::max(std::min(static_cast<int>(n / 4), static_cast<int>(256)), static_cast<int>(1)); - return configure_lhs_rhs_info(m, n, 4, 4, 4, 1, h0, false, true, false, false); + if(n >= 64) + { + return configure_lhs_rhs_info(m, n, 4, 4, 4, 1, h0, false, true, false, false); + } + else + { + if(k >= 512) + { + return configure_lhs_rhs_info(m, n, 2, 4, 16, 1, h0, false, true, false, false); + } + else + { + return configure_lhs_rhs_info(m, n, 2, 4, 8, 1, h0, false, true, false, false); + } + } } } |