diff options
| author | Murray Kornelsen <murray.kornelsen@mail.mcgill.ca> | 2022-07-13 21:40:26 -0400 |
|---|---|---|
| committer | Pablo Marquez Tello <pablo.tello@arm.com> | 2022-09-14 06:48:39 +0000 |
| commit | 6e09e1404c635d948cf20eb6b4b5747dfb6656f2 (patch) | |
| tree | 006199bd21b8a1330e1f1c86be60084bfb466706 /scripts | |
| parent | a4814e8394ffdd7e268614d54cc22e30648f48ff (diff) | |
| download | ComputeLibrary-6e09e1404c635d948cf20eb6b4b5747dfb6656f2.tar.gz | |
INT8 Quantized MeanStdDevNorm (LayerNorm)
Implements LayerNorm for qasymm8 tensors.
Uses uint8x16 loads and stores.
Summation is performed in integer arithmetic (vpaddl)
Normalization is performed in float32 before requantizing back to int8.
Signed-off-by: Murray Kornelsen <murray.kornelsen@mail.mcgill.ca>
Change-Id: I2407c8b34717fb47adab98791bd76fb8a3c62f4a
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/7922
Comments-Addressed: Pablo Marquez Tello <pablo.tello@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Viet-Hoa Do <viet-hoa.do@arm.com>
Reviewed-by: Pablo Marquez Tello <pablo.tello@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Benchmark: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'scripts')
| -rw-r--r-- | scripts/clang-tidy.h | 7 |
1 files changed, 7 insertions, 0 deletions
diff --git a/scripts/clang-tidy.h b/scripts/clang-tidy.h index b3705122c6..24e4b15c6f 100644 --- a/scripts/clang-tidy.h +++ b/scripts/clang-tidy.h @@ -1,5 +1,12 @@ #include <arm_neon.h> +#if __arm__ +inline uint32x4_t vpaddq_u32(uint32x4_t, uint32x4_t) +{ + return vdupq_n_u32(0); +} +#endif + inline float16x4_t vrsqrts_f16 (float16x4_t, float16x4_t) { return vdup_n_f16(0); |