diff options
author | Georgios Pinitas <georgios.pinitas@arm.com> | 2019-06-27 17:00:52 +0100 |
---|---|---|
committer | Georgios Pinitas <georgios.pinitas@arm.com> | 2019-07-26 11:55:15 +0000 |
commit | cfa2bba98169cb5ab1945462514be1b6badf7d98 (patch) | |
tree | 1635e6e9463e9798c7195f0aa71b5df3f2650df1 /src/core/NEON/kernels/arm_gemm/gemv_batched.hpp | |
parent | f59b16f42ef68bde877b70816ffb953d64c8baa3 (diff) | |
download | ComputeLibrary-cfa2bba98169cb5ab1945462514be1b6badf7d98.tar.gz |
COMPMID-2178: Update GEMM assembly code.
Perform offset reduction and requantization within the assembly wrapper.
Change-Id: I5d5b3e1f6f9ef4c71805362c57f88ff199c027a3
Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com>
Reviewed-on: https://review.mlplatform.org/c/1541
Comments-Addressed: Pablo Marquez <pablo.tello@arm.com>
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src/core/NEON/kernels/arm_gemm/gemv_batched.hpp')
-rw-r--r-- | src/core/NEON/kernels/arm_gemm/gemv_batched.hpp | 3 |
1 files changed, 2 insertions, 1 deletions
diff --git a/src/core/NEON/kernels/arm_gemm/gemv_batched.hpp b/src/core/NEON/kernels/arm_gemm/gemv_batched.hpp index b7f9de85c4..8dae7c3098 100644 --- a/src/core/NEON/kernels/arm_gemm/gemv_batched.hpp +++ b/src/core/NEON/kernels/arm_gemm/gemv_batched.hpp @@ -41,12 +41,13 @@ public: GemmArgs<Tr> newargs = args; newargs._Msize = args._nbatches; newargs._nbatches = 1; + newargs._cfg = nullptr; _subgemm = gemm<To,Tr>(newargs); } void set_arrays(const To *A, const int lda, const int A_batch_stride, const int A_multi_stride, const To *B, const int ldb, const int B_multi_stride, - Tr *C, const int ldc, const int C_batch_stride, const int C_multi_stride) override { + Tr *C, const int ldc, const int C_batch_stride, const int C_multi_stride) override { /* A and C's batch stride becomes their new row stride. New batch stride is 0 as nbatches for subgemm is always 1. */ _subgemm->set_arrays(A, A_batch_stride, 0, A_multi_stride, B, ldb, B_multi_stride, |