// Auto-generated file. Do not edit! // Template: src/qs8-gemm/c4-neondot.c.in // Generator: tools/xngen // // Copyright 2020 Google LLC // // This source code is licensed under the BSD-style license found in the // LICENSE file in the root directory of this source tree. #include #include #include "xnnpack/gemm.h" #include "xnnpack/math.h" void xnn_qd8_f32_qb4w_gemm_minmax_ukernel_3x16c4__neondot( size_t mr, size_t nc, size_t kc, const int8_t* restrict a, size_t a_stride, const void* restrict w, float* restrict c, size_t cm_stride, size_t cn_stride, const struct xnn_f32_qb4w_minmax_params params[restrict XNN_MIN_ELEMENTS(1)], const struct xnn_qd8_quantization_params quantization_params[restrict XNN_MIN_ELEMENTS(1)]) XNN_OOB_READS { assert(mr != 0); assert(mr <= 3); assert(nc != 0); assert(kc != 0); assert(kc % sizeof(int8_t) == 0); assert(a != NULL); assert(w != NULL); assert(c != NULL); kc = round_up_po2(kc, 4 * sizeof(int8_t)); const int8_t* a0 = a; float* c0 = c; const int8_t* a1 = (const int8_t*) ((uintptr_t) a0 + a_stride); float* c1 = (float*) ((uintptr_t) c0 + cm_stride); if XNN_UNPREDICTABLE(mr < 2) { a1 = a0; c1 = c0; } const int8_t* a2 = (const int8_t*) ((uintptr_t) a1 + a_stride); float* c2 = (float*) ((uintptr_t) c1 + cm_stride); if XNN_UNPREDICTABLE(mr <= 2) { a2 = a1; c2 = c1; } size_t bl = params->scalar.blocksize; assert(bl <= kc); assert(bl != 0); assert(bl % 32 == 0); const int8x16_t vmask = vmovq_n_s8(INT8_C(0xF0)); // Loop over groups of 16 columns. do { // Initialize accumulators with bias. 16 bias values are loaded from the // weight matrix, at the start of the group of 16 columns. const float32x4_t vinput_zero_point01 = vcvtq_f32_s32(vld1q_s32(&quantization_params[0].zero_point)); const float32x4_t vksum0123 = vld1q_f32(w); w = (const float*) w + 4; float32x4_t vout0x0123 = vmulq_lane_f32(vksum0123, vget_low_f32(vinput_zero_point01), 0); float32x4_t vout1x0123 = vmulq_lane_f32(vksum0123, vget_high_f32(vinput_zero_point01), 0); const float32x4_t vksum4567 = vld1q_f32(w); w = (const float*) w + 4; float32x4_t vout0x4567 = vmulq_lane_f32(vksum4567, vget_low_f32(vinput_zero_point01), 0); float32x4_t vout1x4567 = vmulq_lane_f32(vksum4567, vget_high_f32(vinput_zero_point01), 0); const float32x4_t vksum89AB = vld1q_f32(w); w = (const float*) w + 4; float32x4_t vout0x89AB = vmulq_lane_f32(vksum89AB, vget_low_f32(vinput_zero_point01), 0); float32x4_t vout1x89AB = vmulq_lane_f32(vksum89AB, vget_high_f32(vinput_zero_point01), 0); const float32x4_t vksumCDEF = vld1q_f32(w); w = (const float*) w + 4; float32x4_t vout0xCDEF = vmulq_lane_f32(vksumCDEF, vget_low_f32(vinput_zero_point01), 0); float32x4_t vout1xCDEF = vmulq_lane_f32(vksumCDEF, vget_high_f32(vinput_zero_point01), 0); const float32x4_t vinput_zero_point2 = vcvtq_f32_s32(vld1q_dup_s32(&quantization_params[2].zero_point)); float32x4_t vout2x0123 = vmulq_f32(vksum0123, vinput_zero_point2); float32x4_t vout2x4567 = vmulq_f32(vksum4567, vinput_zero_point2); float32x4_t vout2x89AB = vmulq_f32(vksum89AB, vinput_zero_point2); float32x4_t vout2xCDEF = vmulq_f32(vksumCDEF, vinput_zero_point2); for (size_t kb=0; kb < kc; kb += bl) { int32x4_t vacc0x0123 = vdupq_n_s32(0); int32x4_t vacc1x0123 = vdupq_n_s32(0); int32x4_t vacc0x4567 = vdupq_n_s32(0); int32x4_t vacc1x4567 = vdupq_n_s32(0); int32x4_t vacc0x89AB = vdupq_n_s32(0); int32x4_t vacc1x89AB = vdupq_n_s32(0); int32x4_t vacc0xCDEF = vdupq_n_s32(0); int32x4_t vacc1xCDEF = vdupq_n_s32(0); int32x4_t vacc2x0123 = vdupq_n_s32(0); int32x4_t vacc2x4567 = vdupq_n_s32(0); int32x4_t vacc2x89AB = vdupq_n_s32(0); int32x4_t vacc2xCDEF = vdupq_n_s32(0); // Inner accumulation loop along the 16 columns. size_t k = bl; // 2x partial unrolled loop to load 8 bytes at a time. while (k >= 8 * sizeof(int8_t)) { // Load a 3x8 block of activations. const int8x8_t va0x01234567 = vld1_s8(a0); a0 += 8; const int8x8_t va1x01234567 = vld1_s8(a1); a1 += 8; const int8x8_t va2x01234567 = vld1_s8(a2); a2 += 8; // Load a 8x16 block of weights. const int8x16_t vb01234567x0123 = vld1q_s8(w); w = (const int8_t*) w + 16; const int8x16_t vb01234567x4567 = vld1q_s8(w); w = (const int8_t*) w + 16; const int8x16_t vb01234567x89AB = vld1q_s8(w); w = (const int8_t*) w + 16; const int8x16_t vb01234567xCDEF = vld1q_s8(w); w = (const int8_t*) w + 16; const int8x16_t vb0123x0123 = vshlq_n_s8(vb01234567x0123, 4); const int8x16_t vb0123x4567 = vshlq_n_s8(vb01234567x4567, 4); const int8x16_t vb0123x89AB = vshlq_n_s8(vb01234567x89AB, 4); const int8x16_t vb0123xCDEF = vshlq_n_s8(vb01234567xCDEF, 4); const int8x16_t vb4567x0123 = vandq_s8(vb01234567x0123, vmask); const int8x16_t vb4567x4567 = vandq_s8(vb01234567x4567, vmask); const int8x16_t vb4567x89AB = vandq_s8(vb01234567x89AB, vmask); const int8x16_t vb4567xCDEF = vandq_s8(vb01234567xCDEF, vmask); // Multiply-accumulate: 3x8 * 8x16 --> 3x16. vacc0x0123 = vdotq_lane_s32(vacc0x0123, vb0123x0123, va0x01234567, 0); vacc0x4567 = vdotq_lane_s32(vacc0x4567, vb0123x4567, va0x01234567, 0); vacc0x89AB = vdotq_lane_s32(vacc0x89AB, vb0123x89AB, va0x01234567, 0); vacc0xCDEF = vdotq_lane_s32(vacc0xCDEF, vb0123xCDEF, va0x01234567, 0); vacc1x0123 = vdotq_lane_s32(vacc1x0123, vb0123x0123, va1x01234567, 0); vacc1x4567 = vdotq_lane_s32(vacc1x4567, vb0123x4567, va1x01234567, 0); vacc1x89AB = vdotq_lane_s32(vacc1x89AB, vb0123x89AB, va1x01234567, 0); vacc1xCDEF = vdotq_lane_s32(vacc1xCDEF, vb0123xCDEF, va1x01234567, 0); vacc2x0123 = vdotq_lane_s32(vacc2x0123, vb0123x0123, va2x01234567, 0); vacc2x4567 = vdotq_lane_s32(vacc2x4567, vb0123x4567, va2x01234567, 0); vacc2x89AB = vdotq_lane_s32(vacc2x89AB, vb0123x89AB, va2x01234567, 0); vacc2xCDEF = vdotq_lane_s32(vacc2xCDEF, vb0123xCDEF, va2x01234567, 0); vacc0x0123 = vdotq_lane_s32(vacc0x0123, vb4567x0123, va0x01234567, 1); vacc0x4567 = vdotq_lane_s32(vacc0x4567, vb4567x4567, va0x01234567, 1); vacc0x89AB = vdotq_lane_s32(vacc0x89AB, vb4567x89AB, va0x01234567, 1); vacc0xCDEF = vdotq_lane_s32(vacc0xCDEF, vb4567xCDEF, va0x01234567, 1); vacc1x0123 = vdotq_lane_s32(vacc1x0123, vb4567x0123, va1x01234567, 1); vacc1x4567 = vdotq_lane_s32(vacc1x4567, vb4567x4567, va1x01234567, 1); vacc1x89AB = vdotq_lane_s32(vacc1x89AB, vb4567x89AB, va1x01234567, 1); vacc1xCDEF = vdotq_lane_s32(vacc1xCDEF, vb4567xCDEF, va1x01234567, 1); vacc2x0123 = vdotq_lane_s32(vacc2x0123, vb4567x0123, va2x01234567, 1); vacc2x4567 = vdotq_lane_s32(vacc2x4567, vb4567x4567, va2x01234567, 1); vacc2x89AB = vdotq_lane_s32(vacc2x89AB, vb4567x89AB, va2x01234567, 1); vacc2xCDEF = vdotq_lane_s32(vacc2xCDEF, vb4567xCDEF, va2x01234567, 1); k -= 8 * sizeof(int8_t); } // Handle up to 4 final positions of `k` if XNN_UNLIKELY(k != 0) { // Load a 3x4 block of activations. const int8x8_t va0x01234567 = vld1_s8(a0); a0 += 4; const int8x8_t va1x01234567 = vld1_s8(a1); a1 += 4; const int8x8_t va2x01234567 = vld1_s8(a2); a2 += 4; // Load a 4x16 block of weights. const int8x16_t vb0123x0123 = vld1q_s8(w); w = (const int8_t*) w + 16; const int8x16_t vb0123x4567 = vld1q_s8(w); w = (const int8_t*) w + 16; const int8x16_t vb0123x89AB = vld1q_s8(w); w = (const int8_t*) w + 16; const int8x16_t vb0123xCDEF = vld1q_s8(w); w = (const int8_t*) w + 16; // Multiply-accumulate: 3x4 * 4x16 --> 3x16. vacc0x0123 = vdotq_lane_s32(vacc0x0123, vb0123x0123, va0x01234567, 0); vacc0x4567 = vdotq_lane_s32(vacc0x4567, vb0123x4567, va0x01234567, 0); vacc0x89AB = vdotq_lane_s32(vacc0x89AB, vb0123x89AB, va0x01234567, 0); vacc0xCDEF = vdotq_lane_s32(vacc0xCDEF, vb0123xCDEF, va0x01234567, 0); vacc1x0123 = vdotq_lane_s32(vacc1x0123, vb0123x0123, va1x01234567, 0); vacc1x4567 = vdotq_lane_s32(vacc1x4567, vb0123x4567, va1x01234567, 0); vacc1x89AB = vdotq_lane_s32(vacc1x89AB, vb0123x89AB, va1x01234567, 0); vacc1xCDEF = vdotq_lane_s32(vacc1xCDEF, vb0123xCDEF, va1x01234567, 0); vacc2x0123 = vdotq_lane_s32(vacc2x0123, vb0123x0123, va2x01234567, 0); vacc2x4567 = vdotq_lane_s32(vacc2x4567, vb0123x4567, va2x01234567, 0); vacc2x89AB = vdotq_lane_s32(vacc2x89AB, vb0123x89AB, va2x01234567, 0); vacc2xCDEF = vdotq_lane_s32(vacc2xCDEF, vb0123xCDEF, va2x01234567, 0); } const float32x4_t vfilter_output_scale0123 = vreinterpretq_f32_u32(vshll_n_u16(vld1_u16(w), 16)); w = (const uint16_t*) w + 4; const float32x4_t vfilter_output_scale4567 = vreinterpretq_f32_u32(vshll_n_u16(vld1_u16(w), 16)); w = (const uint16_t*) w + 4; const float32x4_t vfilter_output_scale89AB = vreinterpretq_f32_u32(vshll_n_u16(vld1_u16(w), 16)); w = (const uint16_t*) w + 4; const float32x4_t vfilter_output_scaleCDEF = vreinterpretq_f32_u32(vshll_n_u16(vld1_u16(w), 16)); w = (const uint16_t*) w + 4; float32x4_t vf0x0123 = vcvtq_f32_s32(vacc0x0123); float32x4_t vf1x0123 = vcvtq_f32_s32(vacc1x0123); vout0x0123 = vfmaq_f32(vout0x0123, vf0x0123, vfilter_output_scale0123); vout1x0123 = vfmaq_f32(vout1x0123, vf1x0123, vfilter_output_scale0123); float32x4_t vf0x4567 = vcvtq_f32_s32(vacc0x4567); float32x4_t vf1x4567 = vcvtq_f32_s32(vacc1x4567); vout0x4567 = vfmaq_f32(vout0x4567, vf0x4567, vfilter_output_scale4567); vout1x4567 = vfmaq_f32(vout1x4567, vf1x4567, vfilter_output_scale4567); float32x4_t vf0x89AB = vcvtq_f32_s32(vacc0x89AB); float32x4_t vf1x89AB = vcvtq_f32_s32(vacc1x89AB); vout0x89AB = vfmaq_f32(vout0x89AB, vf0x89AB, vfilter_output_scale89AB); vout1x89AB = vfmaq_f32(vout1x89AB, vf1x89AB, vfilter_output_scale89AB); float32x4_t vf0xCDEF = vcvtq_f32_s32(vacc0xCDEF); float32x4_t vf1xCDEF = vcvtq_f32_s32(vacc1xCDEF); vout0xCDEF = vfmaq_f32(vout0xCDEF, vf0xCDEF, vfilter_output_scaleCDEF); vout1xCDEF = vfmaq_f32(vout1xCDEF, vf1xCDEF, vfilter_output_scaleCDEF); float32x4_t vf2x0123 = vcvtq_f32_s32(vacc2x0123); vout2x0123 = vfmaq_f32(vout2x0123, vf2x0123, vfilter_output_scale0123); float32x4_t vf2x4567 = vcvtq_f32_s32(vacc2x4567); vout2x4567 = vfmaq_f32(vout2x4567, vf2x4567, vfilter_output_scale4567); float32x4_t vf2x89AB = vcvtq_f32_s32(vacc2x89AB); vout2x89AB = vfmaq_f32(vout2x89AB, vf2x89AB, vfilter_output_scale89AB); float32x4_t vf2xCDEF = vcvtq_f32_s32(vacc2xCDEF); vout2xCDEF = vfmaq_f32(vout2xCDEF, vf2xCDEF, vfilter_output_scaleCDEF); } const float32x4_t vinput_scale01 = vreinterpretq_f32_s32(vld1q_s32(&quantization_params[0].zero_point)); vout0x0123 = vmulq_lane_f32(vout0x0123, vget_low_f32(vinput_scale01), 1); vout1x0123 = vmulq_lane_f32(vout1x0123, vget_high_f32(vinput_scale01), 1); vout0x4567 = vmulq_lane_f32(vout0x4567, vget_low_f32(vinput_scale01), 1); vout1x4567 = vmulq_lane_f32(vout1x4567, vget_high_f32(vinput_scale01), 1); vout0x89AB = vmulq_lane_f32(vout0x89AB, vget_low_f32(vinput_scale01), 1); vout1x89AB = vmulq_lane_f32(vout1x89AB, vget_high_f32(vinput_scale01), 1); vout0xCDEF = vmulq_lane_f32(vout0xCDEF, vget_low_f32(vinput_scale01), 1); vout1xCDEF = vmulq_lane_f32(vout1xCDEF, vget_high_f32(vinput_scale01), 1); const float32x4_t vinput_scale2 = vld1q_dup_f32(&quantization_params[2].inv_scale); vout2x0123 = vmulq_f32(vout2x0123, vinput_scale2); vout2x4567 = vmulq_f32(vout2x4567, vinput_scale2); vout2x89AB = vmulq_f32(vout2x89AB, vinput_scale2); vout2xCDEF = vmulq_f32(vout2xCDEF, vinput_scale2); const float32x4_t vbias0123 = vld1q_f32(w); w = (const float*) w + 4; vout0x0123 = vaddq_f32(vbias0123, vout0x0123); vout1x0123 = vaddq_f32(vbias0123, vout1x0123); vout2x0123 = vaddq_f32(vbias0123, vout2x0123); const float32x4_t vbias4567 = vld1q_f32(w); w = (const float*) w + 4; vout0x4567 = vaddq_f32(vbias4567, vout0x4567); vout1x4567 = vaddq_f32(vbias4567, vout1x4567); vout2x4567 = vaddq_f32(vbias4567, vout2x4567); const float32x4_t vbias89AB = vld1q_f32(w); w = (const float*) w + 4; vout0x89AB = vaddq_f32(vbias89AB, vout0x89AB); vout1x89AB = vaddq_f32(vbias89AB, vout1x89AB); vout2x89AB = vaddq_f32(vbias89AB, vout2x89AB); const float32x4_t vbiasCDEF = vld1q_f32(w); w = (const float*) w + 4; vout0xCDEF = vaddq_f32(vbiasCDEF, vout0xCDEF); vout1xCDEF = vaddq_f32(vbiasCDEF, vout1xCDEF); vout2xCDEF = vaddq_f32(vbiasCDEF, vout2xCDEF); const float32x4_t voutput_min = vld1q_dup_f32(¶ms->scalar.min); vout0x0123 = vmaxq_f32(vout0x0123, voutput_min); vout0x4567 = vmaxq_f32(vout0x4567, voutput_min); vout0x89AB = vmaxq_f32(vout0x89AB, voutput_min); vout0xCDEF = vmaxq_f32(vout0xCDEF, voutput_min); vout1x0123 = vmaxq_f32(vout1x0123, voutput_min); vout1x4567 = vmaxq_f32(vout1x4567, voutput_min); vout1x89AB = vmaxq_f32(vout1x89AB, voutput_min); vout1xCDEF = vmaxq_f32(vout1xCDEF, voutput_min); vout2x0123 = vmaxq_f32(vout2x0123, voutput_min); vout2x4567 = vmaxq_f32(vout2x4567, voutput_min); vout2x89AB = vmaxq_f32(vout2x89AB, voutput_min); vout2xCDEF = vmaxq_f32(vout2xCDEF, voutput_min); const float32x4_t voutput_max = vld1q_dup_f32(¶ms->scalar.max); vout0x0123 = vminq_f32(vout0x0123, voutput_max); vout0x4567 = vminq_f32(vout0x4567, voutput_max); vout0x89AB = vminq_f32(vout0x89AB, voutput_max); vout0xCDEF = vminq_f32(vout0xCDEF, voutput_max); vout1x0123 = vminq_f32(vout1x0123, voutput_max); vout1x4567 = vminq_f32(vout1x4567, voutput_max); vout1x89AB = vminq_f32(vout1x89AB, voutput_max); vout1xCDEF = vminq_f32(vout1xCDEF, voutput_max); vout2x0123 = vminq_f32(vout2x0123, voutput_max); vout2x4567 = vminq_f32(vout2x4567, voutput_max); vout2x89AB = vminq_f32(vout2x89AB, voutput_max); vout2xCDEF = vminq_f32(vout2xCDEF, voutput_max); if XNN_LIKELY(nc >= 16) { vst1q_f32(c0, vout0x0123); vst1q_f32(c0 + 4, vout0x4567); vst1q_f32(c0 + 8, vout0x89AB); vst1q_f32(c0 + 12, vout0xCDEF); vst1q_f32(c1, vout1x0123); vst1q_f32(c1 + 4, vout1x4567); vst1q_f32(c1 + 8, vout1x89AB); vst1q_f32(c1 + 12, vout1xCDEF); vst1q_f32(c2, vout2x0123); vst1q_f32(c2 + 4, vout2x4567); vst1q_f32(c2 + 8, vout2x89AB); vst1q_f32(c2 + 12, vout2xCDEF); a0 = (const int8_t*) ((uintptr_t) a0 - kc); a1 = (const int8_t*) ((uintptr_t) a1 - kc); a2 = (const int8_t*) ((uintptr_t) a2 - kc); c0 = (float*) ((uintptr_t) c0 + cn_stride); c1 = (float*) ((uintptr_t) c1 + cn_stride); c2 = (float*) ((uintptr_t) c2 + cn_stride); nc -= 16; } else { if (nc & 8) { vst1q_f32(c0, vout0x0123); c0 += 4; vout0x0123 = vout0x89AB; vst1q_f32(c1, vout1x0123); c1 += 4; vout1x0123 = vout1x89AB; vst1q_f32(c2, vout2x0123); c2 += 4; vout2x0123 = vout2x89AB; vst1q_f32(c0, vout0x4567); c0 += 4; vout0x4567 = vout0xCDEF; vst1q_f32(c1, vout1x4567); c1 += 4; vout1x4567 = vout1xCDEF; vst1q_f32(c2, vout2x4567); c2 += 4; vout2x4567 = vout2xCDEF; } if (nc & 4) { vst1q_f32(c0, vout0x0123); c0 += 4; vout0x0123 = vout0x4567; vst1q_f32(c1, vout1x0123); c1 += 4; vout1x0123 = vout1x4567; vst1q_f32(c2, vout2x0123); c2 += 4; vout2x0123 = vout2x4567; } float32x2_t vout0x01 = vget_low_f32(vout0x0123); float32x2_t vout1x01 = vget_low_f32(vout1x0123); float32x2_t vout2x01 = vget_low_f32(vout2x0123); if (nc & 2) { vst1_f32(c0, vout0x01); c0 += 2; vst1_f32(c1, vout1x01); c1 += 2; vst1_f32(c2, vout2x01); c2 += 2; vout0x01 = vget_high_f32(vout0x0123); vout1x01 = vget_high_f32(vout1x0123); vout2x01 = vget_high_f32(vout2x0123); } if (nc & 1) { vst1_lane_f32(c0, vout0x01, 0); vst1_lane_f32(c1, vout1x01, 0); vst1_lane_f32(c2, vout2x01, 0); } nc = 0; } } while (nc != 0); }