diff options
| author | SiCong Li <sicong.li@arm.com> | 2023-05-17 13:46:13 +0100 |
|---|---|---|
| committer | SiCong Li <sicong.li@arm.com> | 2023-05-17 16:21:31 +0100 |
| commit | 3f70cd053573cb6140990ab619ead865f55f3139 (patch) | |
| tree | 6ed9cf0222c4b0ab1def1c9276bef41767accd36 /src/cpu/kernels/elementwise_unary | |
| parent | 81ca48606cea2220a83ae8d736d3935bcc17f854 (diff) | |
| download | ComputeLibrary-3f70cd053573cb6140990ab619ead865f55f3139.tar.gz | |
Move lut kernel to sve2 categoryv23.05branches/arm_compute_23_05
This specific Lut kernel uses sve2 instructions
Resolves: COMPMID-6268
Signed-off-by: SiCong Li <sicong.li@arm.com>
Change-Id: I44fa3812e96fa79b3d1e1e3a31d587581f59f0e1
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/9675
Reviewed-by: Viet-Hoa Do <viet-hoa.do@arm.com>
Reviewed-by: Jakub Sujak <jakub.sujak@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Benchmark: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src/cpu/kernels/elementwise_unary')
| -rw-r--r-- | src/cpu/kernels/elementwise_unary/generic/sve2/q8.cpp (renamed from src/cpu/kernels/elementwise_unary/generic/sve/q8.cpp) | 12 | ||||
| -rw-r--r-- | src/cpu/kernels/elementwise_unary/list.h | 2 |
2 files changed, 7 insertions, 7 deletions
diff --git a/src/cpu/kernels/elementwise_unary/generic/sve/q8.cpp b/src/cpu/kernels/elementwise_unary/generic/sve2/q8.cpp index b68f691086..7e32f50132 100644 --- a/src/cpu/kernels/elementwise_unary/generic/sve/q8.cpp +++ b/src/cpu/kernels/elementwise_unary/generic/sve2/q8.cpp @@ -29,23 +29,23 @@ namespace arm_compute { namespace cpu { - -void sve_q8_elementwise_unary(const ITensor *in, ITensor *out, const Window &window, ElementWiseUnary op, const uint8_t *lut) +void sve2_q8_elementwise_unary(const ITensor *in, ITensor *out, const Window &window, ElementWiseUnary op, const uint8_t *lut) { ARM_COMPUTE_UNUSED(op); - auto win = window; + auto win = window; const auto window_end_x = window.x().end(); win.set(0, Window::Dimension(0, 1, 1)); Iterator src_it(in, win); Iterator dst_it(out, win); - execute_window_loop(win, [&](const Coordinates &) { + execute_window_loop(win, [&](const Coordinates &) + { const auto src_ptr = src_it.ptr(); - auto dst_ptr = dst_it.ptr(); + auto dst_ptr = dst_it.ptr(); - lut_u8_sve(lut, 1, window_end_x, &src_ptr, &dst_ptr); + lut_u8_sve2(lut, 1, window_end_x, &src_ptr, &dst_ptr); }, src_it, dst_it); } diff --git a/src/cpu/kernels/elementwise_unary/list.h b/src/cpu/kernels/elementwise_unary/list.h index 432fabf4af..a9701afdd8 100644 --- a/src/cpu/kernels/elementwise_unary/list.h +++ b/src/cpu/kernels/elementwise_unary/list.h @@ -37,7 +37,7 @@ namespace cpu DECLARE_ELEMETWISE_UNARY_KERNEL(sve_fp32_elementwise_unary); DECLARE_ELEMETWISE_UNARY_KERNEL(sve_fp16_elementwise_unary); DECLARE_ELEMETWISE_UNARY_KERNEL(sve_s32_elementwise_unary); -DECLARE_ELEMETWISE_UNARY_KERNEL(sve_q8_elementwise_unary); +DECLARE_ELEMETWISE_UNARY_KERNEL(sve2_q8_elementwise_unary); DECLARE_ELEMETWISE_UNARY_KERNEL(neon_fp32_elementwise_unary); DECLARE_ELEMETWISE_UNARY_KERNEL(neon_fp16_elementwise_unary); DECLARE_ELEMETWISE_UNARY_KERNEL(neon_s32_elementwise_unary); |