diff options
| author | morgolock <pablo.tello@arm.com> | 2020-09-08 16:00:56 +0100 |
|---|---|---|
| committer | Pablo Marquez <pablo.tello@arm.com> | 2020-09-09 12:17:34 +0000 |
| commit | fa269bb2f150b5f9f00cd8ad5c57ff02613cb628 (patch) | |
| tree | c9560ca157a6a861b27537e3fae45be47d5c6652 | |
| parent | bb88f89b7a12e83eea2fc701f1f82aabf7dfcf7a (diff) | |
| download | ComputeLibrary-fa269bb2f150b5f9f00cd8ad5c57ff02613cb628.tar.gz | |
MLCE-229: Fixed requantization per channel in asm kernel
Change-Id: Iaf1465f3144371e153ce123ac00da5cc092f77df
Signed-off-by: morgolock <pablo.tello@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3939
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
| -rw-r--r-- | src/core/NEON/kernels/arm_gemm/quantized.cpp | 1 | ||||
| -rw-r--r-- | src/runtime/NEON/functions/NEGEMMAssemblyDispatch.cpp | 2 |
2 files changed, 2 insertions, 1 deletions
diff --git a/src/core/NEON/kernels/arm_gemm/quantized.cpp b/src/core/NEON/kernels/arm_gemm/quantized.cpp index 201bd9dc2c..cac02cf28e 100644 --- a/src/core/NEON/kernels/arm_gemm/quantized.cpp +++ b/src/core/NEON/kernels/arm_gemm/quantized.cpp @@ -137,6 +137,7 @@ void requantize_block_32_int(const Requantize32 &qp, unsigned int width, unsigne v_shf1l = vld1q_s32(perch_shiftl_ptr + 4); v_shf2l = vld1q_s32(perch_shiftl_ptr + 8); v_shf3l = vld1q_s32(perch_shiftl_ptr + 12); + perch_shiftl_ptr += 16; } } else { v_mul0=v_mul1=v_mul2=v_mul3=v_mul; diff --git a/src/runtime/NEON/functions/NEGEMMAssemblyDispatch.cpp b/src/runtime/NEON/functions/NEGEMMAssemblyDispatch.cpp index eeea3a45ee..30232b4435 100644 --- a/src/runtime/NEON/functions/NEGEMMAssemblyDispatch.cpp +++ b/src/runtime/NEON/functions/NEGEMMAssemblyDispatch.cpp @@ -252,7 +252,7 @@ std::tuple<bool, const int32_t *, const int32_t *, const int32_t *> Fallback<Typ { left_shifts.push_back(std::max(-s, int32_t(0))); right_shifts.push_back(std::min(-s, int32_t(0))); - if(s > 0 && !need_left) + if(s < 0 && !need_left) { need_left = true; } |