diff options
| author | Felix Thomasmathibalan <felixjohnny.thomasmathibalan@arm.com> | 2023-09-27 17:46:17 +0100 |
|---|---|---|
| committer | felixjohnny.thomasmathibalan <felixjohnny.thomasmathibalan@arm.com> | 2023-09-28 12:08:05 +0000 |
| commit | afd38f0c617d6f89b2b4532c6c44f116617e2b6f (patch) | |
| tree | 03bc7d5a762099989b16a656fa8d397b490ed70e /src/core/NEON/SVESymm.h | |
| parent | bdcb4c148ee2fdeaaddf4cf1e57bbb0de02bb894 (diff) | |
| download | ComputeLibrary-afd38f0c617d6f89b2b4532c6c44f116617e2b6f.tar.gz | |
Apply clang-format on repository
Code is formatted as per a revised clang format configuration
file(not part of this delivery). Version 14.0.6 is used.
Exclusion List:
- files with .cl extension
- files that are not strictly C/C++ (e.g. Android.bp, Sconscript ...)
And the following directories
- compute_kernel_writer/validation/
- tests/
- include/
- src/core/NEON/kernels/convolution/
- src/core/NEON/kernels/arm_gemm/
- src/core/NEON/kernels/arm_conv/
- data/
There will be a follow up for formatting of .cl files and the
files under tests/ and compute_kernel_writer/validation/.
Signed-off-by: Felix Thomasmathibalan <felixjohnny.thomasmathibalan@arm.com>
Change-Id: Ib7eb1fcf4e7537b9feaefcfc15098a804a3fde0a
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/10391
Benchmark: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Gunes Bayir <gunes.bayir@arm.com>
Diffstat (limited to 'src/core/NEON/SVESymm.h')
| -rw-r--r-- | src/core/NEON/SVESymm.h | 23 |
1 files changed, 13 insertions, 10 deletions
diff --git a/src/core/NEON/SVESymm.h b/src/core/NEON/SVESymm.h index 6808577681..288d45d979 100644 --- a/src/core/NEON/SVESymm.h +++ b/src/core/NEON/SVESymm.h @@ -28,6 +28,7 @@ #if defined(ARM_COMPUTE_ENABLE_SVE2) #include "src/core/NEON/SVEMath.h" + #include <arm_sve.h> namespace arm_compute @@ -42,8 +43,10 @@ namespace arm_compute */ inline svfloat32x2_t svdequantize_qsymm16_z(svbool_t pg, const svint16_t &qv, float scale) { - const auto vscale = svdup_n_f32(scale); - const svfloat32x2_t vdequantized_input = svcreate2_f32(svmul_f32_z(pg, svcvt_f32_s32_z(pg, svmovlb_s32(qv)), vscale), svmul_f32_z(pg, svcvt_f32_s32_z(pg, svmovlt_s32(qv)), vscale)); + const auto vscale = svdup_n_f32(scale); + const svfloat32x2_t vdequantized_input = + svcreate2_f32(svmul_f32_z(pg, svcvt_f32_s32_z(pg, svmovlb_s32(qv)), vscale), + svmul_f32_z(pg, svcvt_f32_s32_z(pg, svmovlt_s32(qv)), vscale)); return vdequantized_input; } @@ -76,13 +79,13 @@ inline svint16_t svquantize_qsymm16_z(svbool_t pg, const svfloat32x2_t qv, float */ inline svfloat32x4_t svdequantize_z(svbool_t pg, const svint16x2_t qv, const UniformQuantizationInfo &qi) { - const float scale = qi.scale; - const auto vscale = svdup_n_f32(scale); - const svfloat32x4_t vdequantized_input = svcreate4_f32( - svmul_f32_z(pg, svcvt_f32_s32_z(pg, svmovlb_s32(svget2_s16(qv, 0))), vscale), - svmul_f32_z(pg, svcvt_f32_s32_z(pg, svmovlt_s32(svget2_s16(qv, 0))), vscale), - svmul_f32_z(pg, svcvt_f32_s32_z(pg, svmovlb_s32(svget2_s16(qv, 1))), vscale), - svmul_f32_z(pg, svcvt_f32_s32_z(pg, svmovlt_s32(svget2_s16(qv, 1))), vscale)); + const float scale = qi.scale; + const auto vscale = svdup_n_f32(scale); + const svfloat32x4_t vdequantized_input = + svcreate4_f32(svmul_f32_z(pg, svcvt_f32_s32_z(pg, svmovlb_s32(svget2_s16(qv, 0))), vscale), + svmul_f32_z(pg, svcvt_f32_s32_z(pg, svmovlt_s32(svget2_s16(qv, 0))), vscale), + svmul_f32_z(pg, svcvt_f32_s32_z(pg, svmovlb_s32(svget2_s16(qv, 1))), vscale), + svmul_f32_z(pg, svcvt_f32_s32_z(pg, svmovlt_s32(svget2_s16(qv, 1))), vscale)); return vdequantized_input; } @@ -112,4 +115,4 @@ inline svint16x2_t svquantize_qsymm16_z(svbool_t pg, const svfloat32x4_t qv, con } // namespace arm_compute #endif /* defined(ARM_COMPUTE_ENABLE_SVE2) */ -#endif // ARM_COMPUTE_NESYMM_H
\ No newline at end of file +#endif // ARM_COMPUTE_NESYMM_H |