diff options
Diffstat (limited to 'arm_compute/core/NEON/NESymm.h')
| -rw-r--r-- | arm_compute/core/NEON/NESymm.h | 68 |
1 files changed, 67 insertions, 1 deletions
diff --git a/arm_compute/core/NEON/NESymm.h b/arm_compute/core/NEON/NESymm.h index 364a317bc7..a60d5d0fde 100644 --- a/arm_compute/core/NEON/NESymm.h +++ b/arm_compute/core/NEON/NESymm.h @@ -24,11 +24,17 @@ #ifndef __ARM_COMPUTE_NESYMM_H__ #define __ARM_COMPUTE_NESYMM_H__ -#include "NEAsymm.h" +#include "arm_compute/core/NEON/NEMath.h" #include <arm_neon.h> namespace arm_compute { +using qsymm8_t = int8_t; /**< 8 bit quantized symmetric scalar value */ +using qsymm16_t = int16_t; /**< 16 bit quantized symmetric scalar value */ + +using qsymm16x8_t = int16x8_t; /**< 16 bit quantized symmetric vector with 8 elements */ +using qsymm16x8x2_t = int16x8x2_t; /**< 16 bit quantized symmetric vector with 16 elements */ + /** Performs final quantization step on 8 signed 16-bit elements * * @tparam is_bounded_relu Specified if a fused bounded relu should be applied @@ -149,5 +155,65 @@ inline int16x8_t vquantize_int16(const float32x4x2_t &qv, float scale) return vcombine_s16(vqmovn_s32(rf.val[0]), vqmovn_s32(rf.val[1])); } +/** Dequantize a neon vector holding 16 16-bit quantized values. + * + * @param[in] qv Input values to be dequantized. + * @param[in] qi Quantization information to be used in the computation. + * + * @return Dequantized values in a neon vector + */ +inline float32x4x4_t vdequantize(const int16x8x2_t &qv, const UniformQuantizationInfo &qi) +{ + const float scale = qi.scale; + const float32x4_t vscale = vdupq_n_f32(scale); + const float32x4x4_t vdequantized_input = + { + { + vmulq_f32(vcvtq_f32_s32(vmovl_s16(vget_low_s16(qv.val[0]))), vscale), + vmulq_f32(vcvtq_f32_s32(vmovl_s16(vget_high_s16(qv.val[0]))), vscale), + vmulq_f32(vcvtq_f32_s32(vmovl_s16(vget_low_s16(qv.val[1]))), vscale), + vmulq_f32(vcvtq_f32_s32(vmovl_s16(vget_high_s16(qv.val[1]))), vscale), + } + }; + return vdequantized_input; +} + +/** Quantize a neon vector holding 16 floating point values. + * + * @param[in] qv Input values to be quantized. + * @param[in] qi Quantization information to be used in the computation. + * + * @return A neon vector holding the quantized values + */ +inline qsymm16x8x2_t vquantize_qsymm16(const float32x4x4_t &qv, const UniformQuantizationInfo &qi) +{ + const float scale = qi.scale; + ARM_COMPUTE_ERROR_ON(scale == 0.f); + const float32x4_t vinvscale = vdupq_n_f32(1.f / scale); + const int32x4x4_t rf = + { + { +#ifdef __aarch64__ + vcvtnq_s32_f32(vmulq_f32(qv.val[0], vinvscale)), + vcvtnq_s32_f32(vmulq_f32(qv.val[1], vinvscale)), + vcvtnq_s32_f32(vmulq_f32(qv.val[2], vinvscale)), + vcvtnq_s32_f32(vmulq_f32(qv.val[3], vinvscale)), +#else //__aarch64__ + vcvtq_s32_f32(vmulq_f32(qv.val[0], vinvscale)), + vcvtq_s32_f32(vmulq_f32(qv.val[1], vinvscale)), + vcvtq_s32_f32(vmulq_f32(qv.val[2], vinvscale)), + vcvtq_s32_f32(vmulq_f32(qv.val[3], vinvscale)), +#endif //__aarch64__ + } + }; + const qsymm16x8x2_t res = + { + vcombine_s16(vqmovn_s32(rf.val[0]), vqmovn_s32(rf.val[1])), + vcombine_s16(vqmovn_s32(rf.val[2]), vqmovn_s32(rf.val[3])), + }; + + return res; +} + } // namespace arm_compute #endif // __ARM_COMPUTE_NESYMM_H__ |