diff options
| author | Pablo Tello <pablo.tello@arm.com> | 2019-02-05 16:16:19 +0000 |
|---|---|---|
| committer | Pablo Marquez <pablo.tello@arm.com> | 2019-02-08 11:15:30 +0000 |
| commit | 54e98d98fbe082b265b2c4a384eabe0144866bcc (patch) | |
| tree | f0367171d8dfcad282523efceb709ec523ef9edd /arm_compute | |
| parent | 03bb550b72ebd107a6cbd994008cdcc00597d822 (diff) | |
| download | ComputeLibrary-54e98d98fbe082b265b2c4a384eabe0144866bcc.tar.gz | |
COMPMID-1918: Different qinfos support in NEConcatLayer.
Added support in NEDepthConcatenateLayerKernel and NEWidthConcatenateLayer for
different quantization arguments both for the input and output.
If input's quantization infos are not homogeneous the input values are requantized using
the output's quantization info.
Change-Id: I2daa638361947eb3ec848d5425d0a5bbfea1936d
Reviewed-on: https://review.mlplatform.org/627
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-by: Isabella Gottardi <isabella.gottardi@arm.com>
Diffstat (limited to 'arm_compute')
| -rw-r--r-- | arm_compute/core/NEON/NEAsymm.h | 62 |
1 files changed, 61 insertions, 1 deletions
diff --git a/arm_compute/core/NEON/NEAsymm.h b/arm_compute/core/NEON/NEAsymm.h index faff59563b..c7f59e9eba 100644 --- a/arm_compute/core/NEON/NEAsymm.h +++ b/arm_compute/core/NEON/NEAsymm.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2017-2018 ARM Limited. + * Copyright (c) 2017-2019 ARM Limited. * * SPDX-License-Identifier: MIT * @@ -124,6 +124,66 @@ uint8x16_t finalize_quantization(int32x4x4_t &in_s32, return out_u8; } + +/** Dequantize a neon vector holding 16 quantized values. + * + * @param qv Input values to be dequantized. + * @param qi Quantization information to be used in the computation. + * + * @return Dequantized values in a neon vector + */ +inline float32x4x4_t vdequantize(const uint8x16_t &qv, const QuantizationInfo &qi) +{ + const float scale = qi.scale; + const int offset = qi.offset; + const int32x4_t voffset = vdupq_n_s32(offset); + const float32x4_t vscale = vdupq_n_f32(scale); + const float32x4x4_t vdequantized_input = + { + { + vmulq_f32(vcvtq_f32_s32(vsubq_s32(vreinterpretq_s32_u32(vmovl_u16(vget_low_u16(vmovl_u8(vget_low_u8(qv))))), voffset)), vscale), + vmulq_f32(vcvtq_f32_s32(vsubq_s32(vreinterpretq_s32_u32(vmovl_u16(vget_high_u16(vmovl_u8(vget_low_u8(qv))))), voffset)), vscale), + vmulq_f32(vcvtq_f32_s32(vsubq_s32(vreinterpretq_s32_u32(vmovl_u16(vget_low_u16(vmovl_u8(vget_high_u8(qv))))), voffset)), vscale), + vmulq_f32(vcvtq_f32_s32(vsubq_s32(vreinterpretq_s32_u32(vmovl_u16(vget_high_u16(vmovl_u8(vget_high_u8(qv))))), voffset)), vscale), + } + }; + return vdequantized_input; +} + +/** Quantize a neon vector holding 16 floating point values. + * + * @param qv Input values to be quantized. + * @param qi Quantization information to be used in the computation. + * + * @return A neon vector holding the quantized values + */ +inline uint8x16_t vquantize(const float32x4x4_t &qv, const QuantizationInfo &qi) +{ + const float scale = qi.scale; + const int offset = qi.offset; + const float32x4_t voffset = vdupq_n_f32(offset); + const float32x4_t vinvscale = vdupq_n_f32(1.f / scale); + const int32x4x4_t rf = + { + { +#ifdef __aarch64__ + vcvtnq_s32_f32(vmlaq_f32(voffset, qv.val[0], vinvscale)), + vcvtnq_s32_f32(vmlaq_f32(voffset, qv.val[1], vinvscale)), + vcvtnq_s32_f32(vmlaq_f32(voffset, qv.val[2], vinvscale)), + vcvtnq_s32_f32(vmlaq_f32(voffset, qv.val[3], vinvscale)), +#else //__aarch64__ + vcvtq_s32_f32(vmlaq_f32(voffset, qv.val[0], vinvscale)), + vcvtq_s32_f32(vmlaq_f32(voffset, qv.val[1], vinvscale)), + vcvtq_s32_f32(vmlaq_f32(voffset, qv.val[2], vinvscale)), + vcvtq_s32_f32(vmlaq_f32(voffset, qv.val[3], vinvscale)), +#endif //__aarch64__ + } + }; + const uint8x8_t pa = vqmovun_s16(vcombine_s16(vqmovn_s32(rf.val[0]), vqmovn_s32(rf.val[1]))); + const uint8x8_t pb = vqmovun_s16(vcombine_s16(vqmovn_s32(rf.val[2]), vqmovn_s32(rf.val[3]))); + return vcombine_u8(pa, pb); +} + } // namespace arm_compute #include "arm_compute/core/NEON/NEAsymm.inl" #endif // __ARM_COMPUTE_NEASYMM_H__ |