diff options
author | Georgios Pinitas <georgios.pinitas@arm.com> | 2019-03-01 19:07:11 +0000 |
---|---|---|
committer | Georgios Pinitas <georgios.pinitas@arm.com> | 2019-03-19 11:53:45 +0000 |
commit | 146138378c1587b7297d245b7177641315f6180b (patch) | |
tree | 4689218b48a0884418099f373015429f2845ceaf /src/core/NEON/kernels/arm_gemm/gemv_batched.hpp | |
parent | 3dd5b6884a65c06bcb9d15589ee2dc2978e3b336 (diff) | |
download | ComputeLibrary-146138378c1587b7297d245b7177641315f6180b.tar.gz |
COMPMID-1995: Update RSH GEMM assembly kernels.
-Updates u8/s8 hybrid dot product kernels to work for any N and any K >=16.
-Adds hybrid FP32 kernels with generic and A55 variants.
-Adds SVE native kernels for fp16/u8/s8.
Change-Id: Ifc0eaba9e3c8ea5bb19d334e870e1b39e4e7e728
Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com>
Reviewed-on: https://review.mlplatform.org/c/863
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michalis Spyrou <michalis.spyrou@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 | 4 |
1 files changed, 1 insertions, 3 deletions
diff --git a/src/core/NEON/kernels/arm_gemm/gemv_batched.hpp b/src/core/NEON/kernels/arm_gemm/gemv_batched.hpp index 32d668f66d..b7f9de85c4 100644 --- a/src/core/NEON/kernels/arm_gemm/gemv_batched.hpp +++ b/src/core/NEON/kernels/arm_gemm/gemv_batched.hpp @@ -44,10 +44,9 @@ public: _subgemm = gemm<To,Tr>(newargs); } - using GemmCommon<To, Tr>::set_arrays; 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, @@ -86,7 +85,6 @@ public: return _subgemm->get_B_pretransposed_array_size(); } - using GemmCommon<To, Tr>::pretranspose_B_array; void pretranspose_B_array(void *buffer, const To *B, const int ldb, const int B_multi_stride) override { _subgemm->pretranspose_B_array(buffer, B, ldb, B_multi_stride); } |