From 448a81fcec04333364a1e3266d5081596d3a0477 Mon Sep 17 00:00:00 2001 From: Georgios Pinitas Date: Thu, 21 Nov 2019 14:10:25 +0000 Subject: COMPMID-2805: Add QASYMM8_SIGNED support in NEGEMMLowpOutputStage Add support from requantizing down from S32 to Int8 with fixed point requantization. This involves the following: - Compute fixed point multiplication between each entry of input by result_fixedpoint_multiplier - Add bias to final result if bias tensor is not a nullptr - Round to nearest division by a power-of-two using result_shift - Add offset to each result - Clamp the value between the specified min and max bounds - Cast to int8 data type Change-Id: I641b3fac0833c568d8565ccb859bbc561a24c17d Signed-off-by: Georgios Pinitas Reviewed-on: https://review.mlplatform.org/c/2340 Comments-Addressed: Arm Jenkins Reviewed-by: Michele Di Giorgio Tested-by: Arm Jenkins --- arm_compute/core/NEON/NEKernels.h | 1 + 1 file changed, 1 insertion(+) (limited to 'arm_compute/core/NEON/NEKernels.h') diff --git a/arm_compute/core/NEON/NEKernels.h b/arm_compute/core/NEON/NEKernels.h index aa46a346e9..05485d847a 100644 --- a/arm_compute/core/NEON/NEKernels.h +++ b/arm_compute/core/NEON/NEKernels.h @@ -80,6 +80,7 @@ #include "arm_compute/core/NEON/kernels/NEGEMMLowpOffsetContributionKernel.h" #include "arm_compute/core/NEON/kernels/NEGEMMLowpOffsetContributionOutputStageKernel.h" #include "arm_compute/core/NEON/kernels/NEGEMMLowpQuantizeDownInt32ToInt16ScaleByFixedPointKernel.h" +#include "arm_compute/core/NEON/kernels/NEGEMMLowpQuantizeDownInt32ToInt8ScaleByFixedPointKernel.h" #include "arm_compute/core/NEON/kernels/NEGEMMLowpQuantizeDownInt32ToUint8ScaleByFixedPointKernel.h" #include "arm_compute/core/NEON/kernels/NEGEMMLowpQuantizeDownInt32ToUint8ScaleKernel.h" #include "arm_compute/core/NEON/kernels/NEGEMMLowpReductionKernel.h" -- cgit v1.2.1