// Auto-generated file. Do not edit! // Template: src/qs8-igemm/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/igemm.h" #include "xnnpack/math.h" void xnn_qd8_f16_qc8w_igemm_minmax_ukernel_4x32c4__neondotfp16arith( size_t mr, size_t nc, size_t kc, size_t ks, const int8_t** restrict a, const void* restrict w, xnn_float16* restrict c, size_t cm_stride, size_t cn_stride, size_t a_offset, const int8_t* zero, const int8_t* zero_data, const union xnn_f16_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 <= 4); assert(nc != 0); assert(kc != 0); assert(ks != 0); assert(ks % (4 * sizeof(void*)) == 0); assert(a_offset % sizeof(int8_t) == 0); assert(a != NULL); assert(w != NULL); assert(c != NULL); kc = round_up_po2(kc, 4 * sizeof(int8_t)); uint16_t* c0 = (uint16_t*) c; uint16_t* c1 = (uint16_t*) ((uintptr_t) c0 + cm_stride); if XNN_UNPREDICTABLE(mr < 2) { c1 = c0; } uint16_t* c2 = (uint16_t*) ((uintptr_t) c1 + cm_stride); if XNN_UNPREDICTABLE(mr <= 2) { c2 = c1; } uint16_t* c3 = (uint16_t*) ((uintptr_t) c2 + cm_stride); if XNN_UNPREDICTABLE(mr != 4) { c3 = c2; } do { const int32x4_t vinput_zero_point = vld1q_dup_s32(&quantization_params->zero_point); const int32x4_t vksum0123 = vld1q_s32(w); w = (const int32_t*) w + 4; int32x4_t vacc0x0123 = vmulq_s32(vksum0123, vinput_zero_point); const int32x4_t vksum4567 = vld1q_s32(w); w = (const int32_t*) w + 4; int32x4_t vacc0x4567 = vmulq_s32(vksum4567, vinput_zero_point); const int32x4_t vksum89AB = vld1q_s32(w); w = (const int32_t*) w + 4; int32x4_t vacc0x89AB = vmulq_s32(vksum89AB, vinput_zero_point); const int32x4_t vksumCDEF = vld1q_s32(w); w = (const int32_t*) w + 4; int32x4_t vacc0xCDEF = vmulq_s32(vksumCDEF, vinput_zero_point); const int32x4_t vksumGHIJ = vld1q_s32(w); w = (const int32_t*) w + 4; int32x4_t vacc0xGHIJ = vmulq_s32(vksumGHIJ, vinput_zero_point); const int32x4_t vksumKLMN = vld1q_s32(w); w = (const int32_t*) w + 4; int32x4_t vacc0xKLMN = vmulq_s32(vksumKLMN, vinput_zero_point); const int32x4_t vksumOPQR = vld1q_s32(w); w = (const int32_t*) w + 4; int32x4_t vacc0xOPQR = vmulq_s32(vksumOPQR, vinput_zero_point); const int32x4_t vksumSTUV = vld1q_s32(w); w = (const int32_t*) w + 4; int32x4_t vacc0xSTUV = vmulq_s32(vksumSTUV, vinput_zero_point); int32x4_t vacc1x0123 = vacc0x0123; int32x4_t vacc1x4567 = vacc0x4567; int32x4_t vacc1x89AB = vacc0x89AB; int32x4_t vacc1xCDEF = vacc0xCDEF; int32x4_t vacc1xGHIJ = vacc0xGHIJ; int32x4_t vacc1xKLMN = vacc0xKLMN; int32x4_t vacc1xOPQR = vacc0xOPQR; int32x4_t vacc1xSTUV = vacc0xSTUV; int32x4_t vacc2x0123 = vacc0x0123; int32x4_t vacc2x4567 = vacc0x4567; int32x4_t vacc2x89AB = vacc0x89AB; int32x4_t vacc2xCDEF = vacc0xCDEF; int32x4_t vacc2xGHIJ = vacc0xGHIJ; int32x4_t vacc2xKLMN = vacc0xKLMN; int32x4_t vacc2xOPQR = vacc0xOPQR; int32x4_t vacc2xSTUV = vacc0xSTUV; int32x4_t vacc3x0123 = vacc0x0123; int32x4_t vacc3x4567 = vacc0x4567; int32x4_t vacc3x89AB = vacc0x89AB; int32x4_t vacc3xCDEF = vacc0xCDEF; int32x4_t vacc3xGHIJ = vacc0xGHIJ; int32x4_t vacc3xKLMN = vacc0xKLMN; int32x4_t vacc3xOPQR = vacc0xOPQR; int32x4_t vacc3xSTUV = vacc0xSTUV; size_t p = ks; do { const int8_t* restrict a0 = a[0]; if XNN_UNPREDICTABLE(a0 != zero) { a0 = (const int8_t*) ((uintptr_t) a0 + a_offset); } else { a0 = zero_data; } const int8_t* restrict a1 = a[1]; if XNN_UNPREDICTABLE(a1 != zero) { a1 = (const int8_t*) ((uintptr_t) a1 + a_offset); } else { a1 = zero_data; } const int8_t* restrict a2 = a[2]; if XNN_UNPREDICTABLE(a2 != zero) { a2 = (const int8_t*) ((uintptr_t) a2 + a_offset); } else { a2 = zero_data; } const int8_t* restrict a3 = a[3]; if XNN_UNPREDICTABLE(a3 != zero) { a3 = (const int8_t*) ((uintptr_t) a3 + a_offset); } else { a3 = zero_data; } a += 4; // Inner accumulation loop along the 32 columns. size_t k = kc; // 2x partial unrolled loop to load 8 bytes at a time. while (k >= 8 * sizeof(int8_t)) { // Load a 4x8 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; const int8x8_t va3x01234567 = vld1_s8(a3); a3 += 8; // Load a 8x32 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; const int8x16_t vb0123xGHIJ = vld1q_s8(w); w = (const int8_t*) w + 16; const int8x16_t vb0123xKLMN = vld1q_s8(w); w = (const int8_t*) w + 16; const int8x16_t vb0123xOPQR = vld1q_s8(w); w = (const int8_t*) w + 16; const int8x16_t vb0123xSTUV = vld1q_s8(w); w = (const int8_t*) w + 16; const int8x16_t vb4567x0123 = vld1q_s8(w); w = (const int8_t*) w + 16; const int8x16_t vb4567x4567 = vld1q_s8(w); w = (const int8_t*) w + 16; const int8x16_t vb4567x89AB = vld1q_s8(w); w = (const int8_t*) w + 16; const int8x16_t vb4567xCDEF = vld1q_s8(w); w = (const int8_t*) w + 16; const int8x16_t vb4567xGHIJ = vld1q_s8(w); w = (const int8_t*) w + 16; const int8x16_t vb4567xKLMN = vld1q_s8(w); w = (const int8_t*) w + 16; const int8x16_t vb4567xOPQR = vld1q_s8(w); w = (const int8_t*) w + 16; const int8x16_t vb4567xSTUV = vld1q_s8(w); w = (const int8_t*) w + 16; // Multiply-accumulate: 4x8 * 8x32 --> 4x32. 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); vacc0xGHIJ = vdotq_lane_s32(vacc0xGHIJ, vb0123xGHIJ, va0x01234567, 0); vacc0xKLMN = vdotq_lane_s32(vacc0xKLMN, vb0123xKLMN, va0x01234567, 0); vacc0xOPQR = vdotq_lane_s32(vacc0xOPQR, vb0123xOPQR, va0x01234567, 0); vacc0xSTUV = vdotq_lane_s32(vacc0xSTUV, vb0123xSTUV, 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); vacc1xGHIJ = vdotq_lane_s32(vacc1xGHIJ, vb0123xGHIJ, va1x01234567, 0); vacc1xKLMN = vdotq_lane_s32(vacc1xKLMN, vb0123xKLMN, va1x01234567, 0); vacc1xOPQR = vdotq_lane_s32(vacc1xOPQR, vb0123xOPQR, va1x01234567, 0); vacc1xSTUV = vdotq_lane_s32(vacc1xSTUV, vb0123xSTUV, 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); vacc2xGHIJ = vdotq_lane_s32(vacc2xGHIJ, vb0123xGHIJ, va2x01234567, 0); vacc2xKLMN = vdotq_lane_s32(vacc2xKLMN, vb0123xKLMN, va2x01234567, 0); vacc2xOPQR = vdotq_lane_s32(vacc2xOPQR, vb0123xOPQR, va2x01234567, 0); vacc2xSTUV = vdotq_lane_s32(vacc2xSTUV, vb0123xSTUV, va2x01234567, 0); vacc3x0123 = vdotq_lane_s32(vacc3x0123, vb0123x0123, va3x01234567, 0); vacc3x4567 = vdotq_lane_s32(vacc3x4567, vb0123x4567, va3x01234567, 0); vacc3x89AB = vdotq_lane_s32(vacc3x89AB, vb0123x89AB, va3x01234567, 0); vacc3xCDEF = vdotq_lane_s32(vacc3xCDEF, vb0123xCDEF, va3x01234567, 0); vacc3xGHIJ = vdotq_lane_s32(vacc3xGHIJ, vb0123xGHIJ, va3x01234567, 0); vacc3xKLMN = vdotq_lane_s32(vacc3xKLMN, vb0123xKLMN, va3x01234567, 0); vacc3xOPQR = vdotq_lane_s32(vacc3xOPQR, vb0123xOPQR, va3x01234567, 0); vacc3xSTUV = vdotq_lane_s32(vacc3xSTUV, vb0123xSTUV, va3x01234567, 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); vacc0xGHIJ = vdotq_lane_s32(vacc0xGHIJ, vb4567xGHIJ, va0x01234567, 1); vacc0xKLMN = vdotq_lane_s32(vacc0xKLMN, vb4567xKLMN, va0x01234567, 1); vacc0xOPQR = vdotq_lane_s32(vacc0xOPQR, vb4567xOPQR, va0x01234567, 1); vacc0xSTUV = vdotq_lane_s32(vacc0xSTUV, vb4567xSTUV, 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); vacc1xGHIJ = vdotq_lane_s32(vacc1xGHIJ, vb4567xGHIJ, va1x01234567, 1); vacc1xKLMN = vdotq_lane_s32(vacc1xKLMN, vb4567xKLMN, va1x01234567, 1); vacc1xOPQR = vdotq_lane_s32(vacc1xOPQR, vb4567xOPQR, va1x01234567, 1); vacc1xSTUV = vdotq_lane_s32(vacc1xSTUV, vb4567xSTUV, 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); vacc2xGHIJ = vdotq_lane_s32(vacc2xGHIJ, vb4567xGHIJ, va2x01234567, 1); vacc2xKLMN = vdotq_lane_s32(vacc2xKLMN, vb4567xKLMN, va2x01234567, 1); vacc2xOPQR = vdotq_lane_s32(vacc2xOPQR, vb4567xOPQR, va2x01234567, 1); vacc2xSTUV = vdotq_lane_s32(vacc2xSTUV, vb4567xSTUV, va2x01234567, 1); vacc3x0123 = vdotq_lane_s32(vacc3x0123, vb4567x0123, va3x01234567, 1); vacc3x4567 = vdotq_lane_s32(vacc3x4567, vb4567x4567, va3x01234567, 1); vacc3x89AB = vdotq_lane_s32(vacc3x89AB, vb4567x89AB, va3x01234567, 1); vacc3xCDEF = vdotq_lane_s32(vacc3xCDEF, vb4567xCDEF, va3x01234567, 1); vacc3xGHIJ = vdotq_lane_s32(vacc3xGHIJ, vb4567xGHIJ, va3x01234567, 1); vacc3xKLMN = vdotq_lane_s32(vacc3xKLMN, vb4567xKLMN, va3x01234567, 1); vacc3xOPQR = vdotq_lane_s32(vacc3xOPQR, vb4567xOPQR, va3x01234567, 1); vacc3xSTUV = vdotq_lane_s32(vacc3xSTUV, vb4567xSTUV, va3x01234567, 1); k -= 8 * sizeof(int8_t); } // Handle up to 4 final positions of `k` if XNN_UNLIKELY(k != 0) { // Load a 4x4 block of activations. const int8x8_t va0x01234567 = vld1_s8(a0); const int8x8_t va1x01234567 = vld1_s8(a1); const int8x8_t va2x01234567 = vld1_s8(a2); const int8x8_t va3x01234567 = vld1_s8(a3); // Load a 4x32 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; const int8x16_t vb0123xGHIJ = vld1q_s8(w); w = (const int8_t*) w + 16; const int8x16_t vb0123xKLMN = vld1q_s8(w); w = (const int8_t*) w + 16; const int8x16_t vb0123xOPQR = vld1q_s8(w); w = (const int8_t*) w + 16; const int8x16_t vb0123xSTUV = vld1q_s8(w); w = (const int8_t*) w + 16; // Multiply-accumulate: 4x4 * 4x32 --> 4x32. 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); vacc0xGHIJ = vdotq_lane_s32(vacc0xGHIJ, vb0123xGHIJ, va0x01234567, 0); vacc0xKLMN = vdotq_lane_s32(vacc0xKLMN, vb0123xKLMN, va0x01234567, 0); vacc0xOPQR = vdotq_lane_s32(vacc0xOPQR, vb0123xOPQR, va0x01234567, 0); vacc0xSTUV = vdotq_lane_s32(vacc0xSTUV, vb0123xSTUV, 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); vacc1xGHIJ = vdotq_lane_s32(vacc1xGHIJ, vb0123xGHIJ, va1x01234567, 0); vacc1xKLMN = vdotq_lane_s32(vacc1xKLMN, vb0123xKLMN, va1x01234567, 0); vacc1xOPQR = vdotq_lane_s32(vacc1xOPQR, vb0123xOPQR, va1x01234567, 0); vacc1xSTUV = vdotq_lane_s32(vacc1xSTUV, vb0123xSTUV, 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); vacc2xGHIJ = vdotq_lane_s32(vacc2xGHIJ, vb0123xGHIJ, va2x01234567, 0); vacc2xKLMN = vdotq_lane_s32(vacc2xKLMN, vb0123xKLMN, va2x01234567, 0); vacc2xOPQR = vdotq_lane_s32(vacc2xOPQR, vb0123xOPQR, va2x01234567, 0); vacc2xSTUV = vdotq_lane_s32(vacc2xSTUV, vb0123xSTUV, va2x01234567, 0); vacc3x0123 = vdotq_lane_s32(vacc3x0123, vb0123x0123, va3x01234567, 0); vacc3x4567 = vdotq_lane_s32(vacc3x4567, vb0123x4567, va3x01234567, 0); vacc3x89AB = vdotq_lane_s32(vacc3x89AB, vb0123x89AB, va3x01234567, 0); vacc3xCDEF = vdotq_lane_s32(vacc3xCDEF, vb0123xCDEF, va3x01234567, 0); vacc3xGHIJ = vdotq_lane_s32(vacc3xGHIJ, vb0123xGHIJ, va3x01234567, 0); vacc3xKLMN = vdotq_lane_s32(vacc3xKLMN, vb0123xKLMN, va3x01234567, 0); vacc3xOPQR = vdotq_lane_s32(vacc3xOPQR, vb0123xOPQR, va3x01234567, 0); vacc3xSTUV = vdotq_lane_s32(vacc3xSTUV, vb0123xSTUV, va3x01234567, 0); } p -= 4 * sizeof(void*); } while (p != 0); float32x4_t vout0x0123 = vcvtq_f32_s32(vacc0x0123); float32x4_t vout0x4567 = vcvtq_f32_s32(vacc0x4567); float32x4_t vout0x89AB = vcvtq_f32_s32(vacc0x89AB); float32x4_t vout0xCDEF = vcvtq_f32_s32(vacc0xCDEF); float32x4_t vout0xGHIJ = vcvtq_f32_s32(vacc0xGHIJ); float32x4_t vout0xKLMN = vcvtq_f32_s32(vacc0xKLMN); float32x4_t vout0xOPQR = vcvtq_f32_s32(vacc0xOPQR); float32x4_t vout0xSTUV = vcvtq_f32_s32(vacc0xSTUV); float32x4_t vout1x0123 = vcvtq_f32_s32(vacc1x0123); float32x4_t vout1x4567 = vcvtq_f32_s32(vacc1x4567); float32x4_t vout1x89AB = vcvtq_f32_s32(vacc1x89AB); float32x4_t vout1xCDEF = vcvtq_f32_s32(vacc1xCDEF); float32x4_t vout1xGHIJ = vcvtq_f32_s32(vacc1xGHIJ); float32x4_t vout1xKLMN = vcvtq_f32_s32(vacc1xKLMN); float32x4_t vout1xOPQR = vcvtq_f32_s32(vacc1xOPQR); float32x4_t vout1xSTUV = vcvtq_f32_s32(vacc1xSTUV); float32x4_t vout2x0123 = vcvtq_f32_s32(vacc2x0123); float32x4_t vout2x4567 = vcvtq_f32_s32(vacc2x4567); float32x4_t vout2x89AB = vcvtq_f32_s32(vacc2x89AB); float32x4_t vout2xCDEF = vcvtq_f32_s32(vacc2xCDEF); float32x4_t vout2xGHIJ = vcvtq_f32_s32(vacc2xGHIJ); float32x4_t vout2xKLMN = vcvtq_f32_s32(vacc2xKLMN); float32x4_t vout2xOPQR = vcvtq_f32_s32(vacc2xOPQR); float32x4_t vout2xSTUV = vcvtq_f32_s32(vacc2xSTUV); float32x4_t vout3x0123 = vcvtq_f32_s32(vacc3x0123); float32x4_t vout3x4567 = vcvtq_f32_s32(vacc3x4567); float32x4_t vout3x89AB = vcvtq_f32_s32(vacc3x89AB); float32x4_t vout3xCDEF = vcvtq_f32_s32(vacc3xCDEF); float32x4_t vout3xGHIJ = vcvtq_f32_s32(vacc3xGHIJ); float32x4_t vout3xKLMN = vcvtq_f32_s32(vacc3xKLMN); float32x4_t vout3xOPQR = vcvtq_f32_s32(vacc3xOPQR); float32x4_t vout3xSTUV = vcvtq_f32_s32(vacc3xSTUV); const float32x4_t vinput_scale = vld1q_dup_f32(&quantization_params->inv_scale); vout0x0123 = vmulq_f32(vout0x0123, vinput_scale); vout0x4567 = vmulq_f32(vout0x4567, vinput_scale); vout0x89AB = vmulq_f32(vout0x89AB, vinput_scale); vout0xCDEF = vmulq_f32(vout0xCDEF, vinput_scale); vout0xGHIJ = vmulq_f32(vout0xGHIJ, vinput_scale); vout0xKLMN = vmulq_f32(vout0xKLMN, vinput_scale); vout0xOPQR = vmulq_f32(vout0xOPQR, vinput_scale); vout0xSTUV = vmulq_f32(vout0xSTUV, vinput_scale); vout1x0123 = vmulq_f32(vout1x0123, vinput_scale); vout1x4567 = vmulq_f32(vout1x4567, vinput_scale); vout1x89AB = vmulq_f32(vout1x89AB, vinput_scale); vout1xCDEF = vmulq_f32(vout1xCDEF, vinput_scale); vout1xGHIJ = vmulq_f32(vout1xGHIJ, vinput_scale); vout1xKLMN = vmulq_f32(vout1xKLMN, vinput_scale); vout1xOPQR = vmulq_f32(vout1xOPQR, vinput_scale); vout1xSTUV = vmulq_f32(vout1xSTUV, vinput_scale); vout2x0123 = vmulq_f32(vout2x0123, vinput_scale); vout2x4567 = vmulq_f32(vout2x4567, vinput_scale); vout2x89AB = vmulq_f32(vout2x89AB, vinput_scale); vout2xCDEF = vmulq_f32(vout2xCDEF, vinput_scale); vout2xGHIJ = vmulq_f32(vout2xGHIJ, vinput_scale); vout2xKLMN = vmulq_f32(vout2xKLMN, vinput_scale); vout2xOPQR = vmulq_f32(vout2xOPQR, vinput_scale); vout2xSTUV = vmulq_f32(vout2xSTUV, vinput_scale); vout3x0123 = vmulq_f32(vout3x0123, vinput_scale); vout3x4567 = vmulq_f32(vout3x4567, vinput_scale); vout3x89AB = vmulq_f32(vout3x89AB, vinput_scale); vout3xCDEF = vmulq_f32(vout3xCDEF, vinput_scale); vout3xGHIJ = vmulq_f32(vout3xGHIJ, vinput_scale); vout3xKLMN = vmulq_f32(vout3xKLMN, vinput_scale); vout3xOPQR = vmulq_f32(vout3xOPQR, vinput_scale); vout3xSTUV = vmulq_f32(vout3xSTUV, vinput_scale); const float32x4_t vfilter_output_scale0123 = vld1q_f32(w); w = (const float*) w + 4; const float32x4_t vfilter_output_scale4567 = vld1q_f32(w); w = (const float*) w + 4; const float32x4_t vfilter_output_scale89AB = vld1q_f32(w); w = (const float*) w + 4; const float32x4_t vfilter_output_scaleCDEF = vld1q_f32(w); w = (const float*) w + 4; const float32x4_t vfilter_output_scaleGHIJ = vld1q_f32(w); w = (const float*) w + 4; const float32x4_t vfilter_output_scaleKLMN = vld1q_f32(w); w = (const float*) w + 4; const float32x4_t vfilter_output_scaleOPQR = vld1q_f32(w); w = (const float*) w + 4; const float32x4_t vfilter_output_scaleSTUV = vld1q_f32(w); w = (const float*) w + 4; const float32x4_t vbias0123 = vld1q_f32(w); w = (const float*) w + 4; #if XNN_ARCH_ARM64 vout0x0123 = vfmaq_f32(vbias0123, vout0x0123, vfilter_output_scale0123); vout1x0123 = vfmaq_f32(vbias0123, vout1x0123, vfilter_output_scale0123); vout2x0123 = vfmaq_f32(vbias0123, vout2x0123, vfilter_output_scale0123); vout3x0123 = vfmaq_f32(vbias0123, vout3x0123, vfilter_output_scale0123); #else vout0x0123 = vmlaq_f32(vbias0123, vout0x0123, vfilter_output_scale0123); vout1x0123 = vmlaq_f32(vbias0123, vout1x0123, vfilter_output_scale0123); vout2x0123 = vmlaq_f32(vbias0123, vout2x0123, vfilter_output_scale0123); vout3x0123 = vmlaq_f32(vbias0123, vout3x0123, vfilter_output_scale0123); #endif const float32x4_t vbias4567 = vld1q_f32(w); w = (const float*) w + 4; #if XNN_ARCH_ARM64 vout0x4567 = vfmaq_f32(vbias4567, vout0x4567, vfilter_output_scale4567); vout1x4567 = vfmaq_f32(vbias4567, vout1x4567, vfilter_output_scale4567); vout2x4567 = vfmaq_f32(vbias4567, vout2x4567, vfilter_output_scale4567); vout3x4567 = vfmaq_f32(vbias4567, vout3x4567, vfilter_output_scale4567); #else vout0x4567 = vmlaq_f32(vbias4567, vout0x4567, vfilter_output_scale4567); vout1x4567 = vmlaq_f32(vbias4567, vout1x4567, vfilter_output_scale4567); vout2x4567 = vmlaq_f32(vbias4567, vout2x4567, vfilter_output_scale4567); vout3x4567 = vmlaq_f32(vbias4567, vout3x4567, vfilter_output_scale4567); #endif const float32x4_t vbias89AB = vld1q_f32(w); w = (const float*) w + 4; #if XNN_ARCH_ARM64 vout0x89AB = vfmaq_f32(vbias89AB, vout0x89AB, vfilter_output_scale89AB); vout1x89AB = vfmaq_f32(vbias89AB, vout1x89AB, vfilter_output_scale89AB); vout2x89AB = vfmaq_f32(vbias89AB, vout2x89AB, vfilter_output_scale89AB); vout3x89AB = vfmaq_f32(vbias89AB, vout3x89AB, vfilter_output_scale89AB); #else vout0x89AB = vmlaq_f32(vbias89AB, vout0x89AB, vfilter_output_scale89AB); vout1x89AB = vmlaq_f32(vbias89AB, vout1x89AB, vfilter_output_scale89AB); vout2x89AB = vmlaq_f32(vbias89AB, vout2x89AB, vfilter_output_scale89AB); vout3x89AB = vmlaq_f32(vbias89AB, vout3x89AB, vfilter_output_scale89AB); #endif const float32x4_t vbiasCDEF = vld1q_f32(w); w = (const float*) w + 4; #if XNN_ARCH_ARM64 vout0xCDEF = vfmaq_f32(vbiasCDEF, vout0xCDEF, vfilter_output_scaleCDEF); vout1xCDEF = vfmaq_f32(vbiasCDEF, vout1xCDEF, vfilter_output_scaleCDEF); vout2xCDEF = vfmaq_f32(vbiasCDEF, vout2xCDEF, vfilter_output_scaleCDEF); vout3xCDEF = vfmaq_f32(vbiasCDEF, vout3xCDEF, vfilter_output_scaleCDEF); #else vout0xCDEF = vmlaq_f32(vbiasCDEF, vout0xCDEF, vfilter_output_scaleCDEF); vout1xCDEF = vmlaq_f32(vbiasCDEF, vout1xCDEF, vfilter_output_scaleCDEF); vout2xCDEF = vmlaq_f32(vbiasCDEF, vout2xCDEF, vfilter_output_scaleCDEF); vout3xCDEF = vmlaq_f32(vbiasCDEF, vout3xCDEF, vfilter_output_scaleCDEF); #endif const float32x4_t vbiasGHIJ = vld1q_f32(w); w = (const float*) w + 4; #if XNN_ARCH_ARM64 vout0xGHIJ = vfmaq_f32(vbiasGHIJ, vout0xGHIJ, vfilter_output_scaleGHIJ); vout1xGHIJ = vfmaq_f32(vbiasGHIJ, vout1xGHIJ, vfilter_output_scaleGHIJ); vout2xGHIJ = vfmaq_f32(vbiasGHIJ, vout2xGHIJ, vfilter_output_scaleGHIJ); vout3xGHIJ = vfmaq_f32(vbiasGHIJ, vout3xGHIJ, vfilter_output_scaleGHIJ); #else vout0xGHIJ = vmlaq_f32(vbiasGHIJ, vout0xGHIJ, vfilter_output_scaleGHIJ); vout1xGHIJ = vmlaq_f32(vbiasGHIJ, vout1xGHIJ, vfilter_output_scaleGHIJ); vout2xGHIJ = vmlaq_f32(vbiasGHIJ, vout2xGHIJ, vfilter_output_scaleGHIJ); vout3xGHIJ = vmlaq_f32(vbiasGHIJ, vout3xGHIJ, vfilter_output_scaleGHIJ); #endif const float32x4_t vbiasKLMN = vld1q_f32(w); w = (const float*) w + 4; #if XNN_ARCH_ARM64 vout0xKLMN = vfmaq_f32(vbiasKLMN, vout0xKLMN, vfilter_output_scaleKLMN); vout1xKLMN = vfmaq_f32(vbiasKLMN, vout1xKLMN, vfilter_output_scaleKLMN); vout2xKLMN = vfmaq_f32(vbiasKLMN, vout2xKLMN, vfilter_output_scaleKLMN); vout3xKLMN = vfmaq_f32(vbiasKLMN, vout3xKLMN, vfilter_output_scaleKLMN); #else vout0xKLMN = vmlaq_f32(vbiasKLMN, vout0xKLMN, vfilter_output_scaleKLMN); vout1xKLMN = vmlaq_f32(vbiasKLMN, vout1xKLMN, vfilter_output_scaleKLMN); vout2xKLMN = vmlaq_f32(vbiasKLMN, vout2xKLMN, vfilter_output_scaleKLMN); vout3xKLMN = vmlaq_f32(vbiasKLMN, vout3xKLMN, vfilter_output_scaleKLMN); #endif const float32x4_t vbiasOPQR = vld1q_f32(w); w = (const float*) w + 4; #if XNN_ARCH_ARM64 vout0xOPQR = vfmaq_f32(vbiasOPQR, vout0xOPQR, vfilter_output_scaleOPQR); vout1xOPQR = vfmaq_f32(vbiasOPQR, vout1xOPQR, vfilter_output_scaleOPQR); vout2xOPQR = vfmaq_f32(vbiasOPQR, vout2xOPQR, vfilter_output_scaleOPQR); vout3xOPQR = vfmaq_f32(vbiasOPQR, vout3xOPQR, vfilter_output_scaleOPQR); #else vout0xOPQR = vmlaq_f32(vbiasOPQR, vout0xOPQR, vfilter_output_scaleOPQR); vout1xOPQR = vmlaq_f32(vbiasOPQR, vout1xOPQR, vfilter_output_scaleOPQR); vout2xOPQR = vmlaq_f32(vbiasOPQR, vout2xOPQR, vfilter_output_scaleOPQR); vout3xOPQR = vmlaq_f32(vbiasOPQR, vout3xOPQR, vfilter_output_scaleOPQR); #endif const float32x4_t vbiasSTUV = vld1q_f32(w); w = (const float*) w + 4; #if XNN_ARCH_ARM64 vout0xSTUV = vfmaq_f32(vbiasSTUV, vout0xSTUV, vfilter_output_scaleSTUV); vout1xSTUV = vfmaq_f32(vbiasSTUV, vout1xSTUV, vfilter_output_scaleSTUV); vout2xSTUV = vfmaq_f32(vbiasSTUV, vout2xSTUV, vfilter_output_scaleSTUV); vout3xSTUV = vfmaq_f32(vbiasSTUV, vout3xSTUV, vfilter_output_scaleSTUV); #else vout0xSTUV = vmlaq_f32(vbiasSTUV, vout0xSTUV, vfilter_output_scaleSTUV); vout1xSTUV = vmlaq_f32(vbiasSTUV, vout1xSTUV, vfilter_output_scaleSTUV); vout2xSTUV = vmlaq_f32(vbiasSTUV, vout2xSTUV, vfilter_output_scaleSTUV); vout3xSTUV = vmlaq_f32(vbiasSTUV, vout3xSTUV, vfilter_output_scaleSTUV); #endif float16x8_t vfp16out0x01234567 = vcombine_f16(vcvt_f16_f32(vout0x0123), vcvt_f16_f32(vout0x4567)); float16x8_t vfp16out0x89ABCDEF = vcombine_f16(vcvt_f16_f32(vout0x89AB), vcvt_f16_f32(vout0xCDEF)); float16x8_t vfp16out0xGHIJKLMN = vcombine_f16(vcvt_f16_f32(vout0xGHIJ), vcvt_f16_f32(vout0xKLMN)); float16x8_t vfp16out0xOPQRSTUV = vcombine_f16(vcvt_f16_f32(vout0xOPQR), vcvt_f16_f32(vout0xSTUV)); float16x8_t vfp16out1x01234567 = vcombine_f16(vcvt_f16_f32(vout1x0123), vcvt_f16_f32(vout1x4567)); float16x8_t vfp16out1x89ABCDEF = vcombine_f16(vcvt_f16_f32(vout1x89AB), vcvt_f16_f32(vout1xCDEF)); float16x8_t vfp16out1xGHIJKLMN = vcombine_f16(vcvt_f16_f32(vout1xGHIJ), vcvt_f16_f32(vout1xKLMN)); float16x8_t vfp16out1xOPQRSTUV = vcombine_f16(vcvt_f16_f32(vout1xOPQR), vcvt_f16_f32(vout1xSTUV)); float16x8_t vfp16out2x01234567 = vcombine_f16(vcvt_f16_f32(vout2x0123), vcvt_f16_f32(vout2x4567)); float16x8_t vfp16out2x89ABCDEF = vcombine_f16(vcvt_f16_f32(vout2x89AB), vcvt_f16_f32(vout2xCDEF)); float16x8_t vfp16out2xGHIJKLMN = vcombine_f16(vcvt_f16_f32(vout2xGHIJ), vcvt_f16_f32(vout2xKLMN)); float16x8_t vfp16out2xOPQRSTUV = vcombine_f16(vcvt_f16_f32(vout2xOPQR), vcvt_f16_f32(vout2xSTUV)); float16x8_t vfp16out3x01234567 = vcombine_f16(vcvt_f16_f32(vout3x0123), vcvt_f16_f32(vout3x4567)); float16x8_t vfp16out3x89ABCDEF = vcombine_f16(vcvt_f16_f32(vout3x89AB), vcvt_f16_f32(vout3xCDEF)); float16x8_t vfp16out3xGHIJKLMN = vcombine_f16(vcvt_f16_f32(vout3xGHIJ), vcvt_f16_f32(vout3xKLMN)); float16x8_t vfp16out3xOPQRSTUV = vcombine_f16(vcvt_f16_f32(vout3xOPQR), vcvt_f16_f32(vout3xSTUV)); const float16x8_t voutput_min = vreinterpretq_f16_u16(vld1q_dup_u16((const uint16_t*) ¶ms->scalar.min)); vfp16out0x01234567 = vmaxq_f16(vfp16out0x01234567, voutput_min); vfp16out0x89ABCDEF = vmaxq_f16(vfp16out0x89ABCDEF, voutput_min); vfp16out0xGHIJKLMN = vmaxq_f16(vfp16out0xGHIJKLMN, voutput_min); vfp16out0xOPQRSTUV = vmaxq_f16(vfp16out0xOPQRSTUV, voutput_min); vfp16out1x01234567 = vmaxq_f16(vfp16out1x01234567, voutput_min); vfp16out1x89ABCDEF = vmaxq_f16(vfp16out1x89ABCDEF, voutput_min); vfp16out1xGHIJKLMN = vmaxq_f16(vfp16out1xGHIJKLMN, voutput_min); vfp16out1xOPQRSTUV = vmaxq_f16(vfp16out1xOPQRSTUV, voutput_min); vfp16out2x01234567 = vmaxq_f16(vfp16out2x01234567, voutput_min); vfp16out2x89ABCDEF = vmaxq_f16(vfp16out2x89ABCDEF, voutput_min); vfp16out2xGHIJKLMN = vmaxq_f16(vfp16out2xGHIJKLMN, voutput_min); vfp16out2xOPQRSTUV = vmaxq_f16(vfp16out2xOPQRSTUV, voutput_min); vfp16out3x01234567 = vmaxq_f16(vfp16out3x01234567, voutput_min); vfp16out3x89ABCDEF = vmaxq_f16(vfp16out3x89ABCDEF, voutput_min); vfp16out3xGHIJKLMN = vmaxq_f16(vfp16out3xGHIJKLMN, voutput_min); vfp16out3xOPQRSTUV = vmaxq_f16(vfp16out3xOPQRSTUV, voutput_min); const float16x8_t voutput_max = vreinterpretq_f16_u16(vld1q_dup_u16((const uint16_t*) ¶ms->scalar.max)); vfp16out0x01234567 = vminq_f16(vfp16out0x01234567, voutput_max); vfp16out0x89ABCDEF = vminq_f16(vfp16out0x89ABCDEF, voutput_max); vfp16out0xGHIJKLMN = vminq_f16(vfp16out0xGHIJKLMN, voutput_max); vfp16out0xOPQRSTUV = vminq_f16(vfp16out0xOPQRSTUV, voutput_max); vfp16out1x01234567 = vminq_f16(vfp16out1x01234567, voutput_max); vfp16out1x89ABCDEF = vminq_f16(vfp16out1x89ABCDEF, voutput_max); vfp16out1xGHIJKLMN = vminq_f16(vfp16out1xGHIJKLMN, voutput_max); vfp16out1xOPQRSTUV = vminq_f16(vfp16out1xOPQRSTUV, voutput_max); vfp16out2x01234567 = vminq_f16(vfp16out2x01234567, voutput_max); vfp16out2x89ABCDEF = vminq_f16(vfp16out2x89ABCDEF, voutput_max); vfp16out2xGHIJKLMN = vminq_f16(vfp16out2xGHIJKLMN, voutput_max); vfp16out2xOPQRSTUV = vminq_f16(vfp16out2xOPQRSTUV, voutput_max); vfp16out3x01234567 = vminq_f16(vfp16out3x01234567, voutput_max); vfp16out3x89ABCDEF = vminq_f16(vfp16out3x89ABCDEF, voutput_max); vfp16out3xGHIJKLMN = vminq_f16(vfp16out3xGHIJKLMN, voutput_max); vfp16out3xOPQRSTUV = vminq_f16(vfp16out3xOPQRSTUV, voutput_max); if XNN_LIKELY(nc >= 32) { vst1q_u16(c3, vreinterpretq_u16_f16(vfp16out3x01234567)); vst1q_u16(c3 + 8, vreinterpretq_u16_f16(vfp16out3x89ABCDEF)); vst1q_u16(c3 + 16, vreinterpretq_u16_f16(vfp16out3xGHIJKLMN)); vst1q_u16(c3 + 24, vreinterpretq_u16_f16(vfp16out3xOPQRSTUV)); vst1q_u16(c2, vreinterpretq_u16_f16(vfp16out2x01234567)); vst1q_u16(c2 + 8, vreinterpretq_u16_f16(vfp16out2x89ABCDEF)); vst1q_u16(c2 + 16, vreinterpretq_u16_f16(vfp16out2xGHIJKLMN)); vst1q_u16(c2 + 24, vreinterpretq_u16_f16(vfp16out2xOPQRSTUV)); vst1q_u16(c1, vreinterpretq_u16_f16(vfp16out1x01234567)); vst1q_u16(c1 + 8, vreinterpretq_u16_f16(vfp16out1x89ABCDEF)); vst1q_u16(c1 + 16, vreinterpretq_u16_f16(vfp16out1xGHIJKLMN)); vst1q_u16(c1 + 24, vreinterpretq_u16_f16(vfp16out1xOPQRSTUV)); vst1q_u16(c0, vreinterpretq_u16_f16(vfp16out0x01234567)); vst1q_u16(c0 + 8, vreinterpretq_u16_f16(vfp16out0x89ABCDEF)); vst1q_u16(c0 + 16, vreinterpretq_u16_f16(vfp16out0xGHIJKLMN)); vst1q_u16(c0 + 24, vreinterpretq_u16_f16(vfp16out0xOPQRSTUV)); a = (const int8_t**restrict) ((uintptr_t) a - ks); c0 = (uint16_t*) ((uintptr_t) c0 + cn_stride); c1 = (uint16_t*) ((uintptr_t) c1 + cn_stride); c2 = (uint16_t*) ((uintptr_t) c2 + cn_stride); c3 = (uint16_t*) ((uintptr_t) c3 + cn_stride); nc -= 32; } else { if (nc & 16) { vst1q_u16(c3, vreinterpretq_u16_f16(vfp16out3x01234567)); c3 += 8; vfp16out3x01234567 = vfp16out3xGHIJKLMN; vst1q_u16(c2, vreinterpretq_u16_f16(vfp16out2x01234567)); c2 += 8; vfp16out2x01234567 = vfp16out2xGHIJKLMN; vst1q_u16(c1, vreinterpretq_u16_f16(vfp16out1x01234567)); c1 += 8; vfp16out1x01234567 = vfp16out1xGHIJKLMN; vst1q_u16(c0, vreinterpretq_u16_f16(vfp16out0x01234567)); c0 += 8; vfp16out0x01234567 = vfp16out0xGHIJKLMN; vst1q_u16(c3, vreinterpretq_u16_f16(vfp16out3x89ABCDEF)); c3 += 8; vfp16out3x89ABCDEF = vfp16out3xOPQRSTUV; vst1q_u16(c2, vreinterpretq_u16_f16(vfp16out2x89ABCDEF)); c2 += 8; vfp16out2x89ABCDEF = vfp16out2xOPQRSTUV; vst1q_u16(c1, vreinterpretq_u16_f16(vfp16out1x89ABCDEF)); c1 += 8; vfp16out1x89ABCDEF = vfp16out1xOPQRSTUV; vst1q_u16(c0, vreinterpretq_u16_f16(vfp16out0x89ABCDEF)); c0 += 8; vfp16out0x89ABCDEF = vfp16out0xOPQRSTUV; } if (nc & 8) { vst1q_u16(c3, vreinterpretq_u16_f16(vfp16out3x01234567)); c3 += 8; vfp16out3x01234567 = vfp16out3x89ABCDEF; vst1q_u16(c2, vreinterpretq_u16_f16(vfp16out2x01234567)); c2 += 8; vfp16out2x01234567 = vfp16out2x89ABCDEF; vst1q_u16(c1, vreinterpretq_u16_f16(vfp16out1x01234567)); c1 += 8; vfp16out1x01234567 = vfp16out1x89ABCDEF; vst1q_u16(c0, vreinterpretq_u16_f16(vfp16out0x01234567)); c0 += 8; vfp16out0x01234567 = vfp16out0x89ABCDEF; } float16x4_t vfp16out3x0123 = vget_low_f16(vfp16out3x01234567); float16x4_t vfp16out2x0123 = vget_low_f16(vfp16out2x01234567); float16x4_t vfp16out1x0123 = vget_low_f16(vfp16out1x01234567); float16x4_t vfp16out0x0123 = vget_low_f16(vfp16out0x01234567); if (nc & 4) { vst1_u16(c3, vreinterpret_u16_f16(vfp16out3x0123)); c3 += 4; vst1_u16(c2, vreinterpret_u16_f16(vfp16out2x0123)); c2 += 4; vst1_u16(c1, vreinterpret_u16_f16(vfp16out1x0123)); c1 += 4; vst1_u16(c0, vreinterpret_u16_f16(vfp16out0x0123)); c0 += 4; vfp16out3x0123 = vget_high_f16(vfp16out3x01234567); vfp16out2x0123 = vget_high_f16(vfp16out2x01234567); vfp16out1x0123 = vget_high_f16(vfp16out1x01234567); vfp16out0x0123 = vget_high_f16(vfp16out0x01234567); } if (nc & 2) { vst1_lane_u32((void*) c3, vreinterpret_u32_f16(vfp16out3x0123), 0); c3 += 2; vst1_lane_u32((void*) c2, vreinterpret_u32_f16(vfp16out2x0123), 0); c2 += 2; vst1_lane_u32((void*) c1, vreinterpret_u32_f16(vfp16out1x0123), 0); c1 += 2; vst1_lane_u32((void*) c0, vreinterpret_u32_f16(vfp16out0x0123), 0); c0 += 2; vfp16out3x0123 = vext_f16(vfp16out3x0123, vfp16out3x0123, 2); vfp16out2x0123 = vext_f16(vfp16out2x0123, vfp16out2x0123, 2); vfp16out1x0123 = vext_f16(vfp16out1x0123, vfp16out1x0123, 2); vfp16out0x0123 = vext_f16(vfp16out0x0123, vfp16out0x0123, 2); } if (nc & 1) { vst1_lane_u16(c3, vreinterpret_u16_f16(vfp16out3x0123), 0); vst1_lane_u16(c2, vreinterpret_u16_f16(vfp16out2x0123), 0); vst1_lane_u16(c1, vreinterpret_u16_f16(vfp16out1x0123), 0); vst1_lane_u16(c0, vreinterpret_u16_f16(vfp16out0x0123), 0); } nc = 0; } } while (nc != 0); }