diff options
| author | David Mansell <David.Mansell@arm.com> | 2018-07-06 14:52:52 +0100 |
|---|---|---|
| committer | Anthony Barbier <anthony.barbier@arm.com> | 2018-11-02 16:54:10 +0000 |
| commit | d93991e290618a685b67506c78090350e6aee43f (patch) | |
| tree | 1d5c3b3017cfccd3f0ec3f24e8e11334cf977ce3 /src/core/NEON/kernels/arm_gemm/transforms/a64_transpose_interleave_12way_16bit.hpp | |
| parent | dec32a9edd4b3c6dc55c60d7436e79af6be58c3d (diff) | |
| download | ComputeLibrary-d93991e290618a685b67506c78090350e6aee43f.tar.gz | |
COMPMID-1380: Pre-work for SVE support.
This patch makes the needed infrastructure changes to allow SVE
kernels to be added later on.
Change-Id: Ide5bccac2f47278e93fff3d648231aee2d5f8c2e
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/139070
Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
Tested-by: Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src/core/NEON/kernels/arm_gemm/transforms/a64_transpose_interleave_12way_16bit.hpp')
| -rw-r--r-- | src/core/NEON/kernels/arm_gemm/transforms/a64_transpose_interleave_12way_16bit.hpp | 8 |
1 files changed, 4 insertions, 4 deletions
diff --git a/src/core/NEON/kernels/arm_gemm/transforms/a64_transpose_interleave_12way_16bit.hpp b/src/core/NEON/kernels/arm_gemm/transforms/a64_transpose_interleave_12way_16bit.hpp index 6e07064a0c..16fa31eb67 100644 --- a/src/core/NEON/kernels/arm_gemm/transforms/a64_transpose_interleave_12way_16bit.hpp +++ b/src/core/NEON/kernels/arm_gemm/transforms/a64_transpose_interleave_12way_16bit.hpp @@ -30,12 +30,12 @@ // Generic unblocked transposed 6x32-bit sized specialisation template <> template <typename T> -inline void TransformImpl<6, 1, true, 4, 4>::Transform( +inline void TransformImpl<6, 1, true, 4, 4, false>::Transform( T* out, const T* const in, const int stride, const int x0, const int xmax, const int k0, const int kmax ) { // Redirect to a 12 x uint16_t specialisation - TransformImpl<12, 1, true, 2, 2>::Transform( + TransformImpl<12, 1, true, 2, 2, false>::Transform( reinterpret_cast<uint16_t *>(out), reinterpret_cast<const uint16_t * const>(in), stride*2, x0*2, xmax*2, k0, kmax @@ -45,7 +45,7 @@ inline void TransformImpl<6, 1, true, 4, 4>::Transform( // Generic 12x16-bit sized specialisation template <> template <typename T> -inline void TransformImpl<12, 1, true, 2, 2>::Transform( +inline void TransformImpl<12, 1, true, 2, 2, false>::Transform( T* out, const T* const in, const int stride, const int x0, const int xmax, const int k0, const int kmax ) { @@ -135,7 +135,7 @@ inline void TransposeInterleaveCommon<12, uint16_t, uint16_t>::moveblock_1x4(con template <> template <> -inline void TransformImpl<12, 1, true, 2, 2>::Transform( +inline void TransformImpl<12, 1, true, 2, 2, false>::Transform( uint16_t* out, const uint16_t* const in, const int stride, const int x0, const int xmax, const int k0, const int kmax ) { |