| /external/XNNPACK/src/qs8-gemm/gen/ |
| D | 1x8c8-minmax-neon-mlal-padal.c | 50 …int32x4_t vacc0x6 = vld1q_lane_s32(w, vmovq_n_s32(0), 0); w = (const void*) ((uintptr_t) w + sizeo… in xnn_qs8_gemm_minmax_ukernel_1x8c8__neon_mlal_padal() local
95 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_gemm_minmax_ukernel_1x8c8__neon_mlal_padal()
128 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_gemm_minmax_ukernel_1x8c8__neon_mlal_padal()
140 const int32x4_t vsum0x67 = vpaddq_s32(vacc0x6, vacc0x7); in xnn_qs8_gemm_minmax_ukernel_1x8c8__neon_mlal_padal()
153 const int32x2_t vpsum0x6 = vadd_s32(vget_low_s32(vacc0x6), vget_high_s32(vacc0x6)); in xnn_qs8_gemm_minmax_ukernel_1x8c8__neon_mlal_padal()
|
| D | 1x8c8-minmax-neon-mull-padal.c | 50 …int32x4_t vacc0x6 = vld1q_lane_s32(w, vmovq_n_s32(0), 0); w = (const void*) ((uintptr_t) w + sizeo… in xnn_qs8_gemm_minmax_ukernel_1x8c8__neon_mull_padal() local
79 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_gemm_minmax_ukernel_1x8c8__neon_mull_padal()
91 const int32x4_t vsum0x67 = vpaddq_s32(vacc0x6, vacc0x7); in xnn_qs8_gemm_minmax_ukernel_1x8c8__neon_mull_padal()
104 const int32x2_t vpsum0x6 = vadd_s32(vget_low_s32(vacc0x6), vget_high_s32(vacc0x6)); in xnn_qs8_gemm_minmax_ukernel_1x8c8__neon_mull_padal()
|
| D | 1x8c16-minmax-neon-mlal-padal.c | 50 …int32x4_t vacc0x6 = vld1q_lane_s32(w, vmovq_n_s32(0), 0); w = (const void*) ((uintptr_t) w + sizeo… in xnn_qs8_gemm_minmax_ukernel_1x8c16__neon_mlal_padal() local
87 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_gemm_minmax_ukernel_1x8c16__neon_mlal_padal()
99 const int32x4_t vsum0x67 = vpaddq_s32(vacc0x6, vacc0x7); in xnn_qs8_gemm_minmax_ukernel_1x8c16__neon_mlal_padal()
112 const int32x2_t vpsum0x6 = vadd_s32(vget_low_s32(vacc0x6), vget_high_s32(vacc0x6)); in xnn_qs8_gemm_minmax_ukernel_1x8c16__neon_mlal_padal()
|
| D | 2x8c8-minmax-neon-mull-padal.c | 56 …int32x4_t vacc0x6 = vld1q_lane_s32(w, vmovq_n_s32(0), 0); w = (const void*) ((uintptr_t) w + sizeo… in xnn_qs8_gemm_minmax_ukernel_2x8c8__neon_mull_padal() local
64 int32x4_t vacc1x6 = vacc0x6; in xnn_qs8_gemm_minmax_ukernel_2x8c8__neon_mull_padal()
107 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_gemm_minmax_ukernel_2x8c8__neon_mull_padal()
122 const int32x4_t vsum0x67 = vpaddq_s32(vacc0x6, vacc0x7); in xnn_qs8_gemm_minmax_ukernel_2x8c8__neon_mull_padal()
141 const int32x2_t vpsum0x6 = vadd_s32(vget_low_s32(vacc0x6), vget_high_s32(vacc0x6)); in xnn_qs8_gemm_minmax_ukernel_2x8c8__neon_mull_padal()
|
| D | 2x8c8-minmax-neon-mlal-padal.c | 56 …int32x4_t vacc0x6 = vld1q_lane_s32(w, vmovq_n_s32(0), 0); w = (const void*) ((uintptr_t) w + sizeo… in xnn_qs8_gemm_minmax_ukernel_2x8c8__neon_mlal_padal() local
64 int32x4_t vacc1x6 = vacc0x6; in xnn_qs8_gemm_minmax_ukernel_2x8c8__neon_mlal_padal()
131 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_gemm_minmax_ukernel_2x8c8__neon_mlal_padal()
182 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_gemm_minmax_ukernel_2x8c8__neon_mlal_padal()
197 const int32x4_t vsum0x67 = vpaddq_s32(vacc0x6, vacc0x7); in xnn_qs8_gemm_minmax_ukernel_2x8c8__neon_mlal_padal()
216 const int32x2_t vpsum0x6 = vadd_s32(vget_low_s32(vacc0x6), vget_high_s32(vacc0x6)); in xnn_qs8_gemm_minmax_ukernel_2x8c8__neon_mlal_padal()
|
| D | 2x8c16-minmax-neon-mlal-padal.c | 56 …int32x4_t vacc0x6 = vld1q_lane_s32(w, vmovq_n_s32(0), 0); w = (const void*) ((uintptr_t) w + sizeo… in xnn_qs8_gemm_minmax_ukernel_2x8c16__neon_mlal_padal() local
64 int32x4_t vacc1x6 = vacc0x6; in xnn_qs8_gemm_minmax_ukernel_2x8c16__neon_mlal_padal()
122 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_gemm_minmax_ukernel_2x8c16__neon_mlal_padal()
138 const int32x4_t vsum0x67 = vpaddq_s32(vacc0x6, vacc0x7); in xnn_qs8_gemm_minmax_ukernel_2x8c16__neon_mlal_padal()
157 const int32x2_t vpsum0x6 = vadd_s32(vget_low_s32(vacc0x6), vget_high_s32(vacc0x6)); in xnn_qs8_gemm_minmax_ukernel_2x8c16__neon_mlal_padal()
|
| D | 3x8c8-minmax-neon-mull-padal.c | 62 …int32x4_t vacc0x6 = vld1q_lane_s32(w, vmovq_n_s32(0), 0); w = (const void*) ((uintptr_t) w + sizeo… in xnn_qs8_gemm_minmax_ukernel_3x8c8__neon_mull_padal() local
70 int32x4_t vacc1x6 = vacc0x6; in xnn_qs8_gemm_minmax_ukernel_3x8c8__neon_mull_padal()
78 int32x4_t vacc2x6 = vacc0x6; in xnn_qs8_gemm_minmax_ukernel_3x8c8__neon_mull_padal()
135 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_gemm_minmax_ukernel_3x8c8__neon_mull_padal()
153 const int32x4_t vsum0x67 = vpaddq_s32(vacc0x6, vacc0x7); in xnn_qs8_gemm_minmax_ukernel_3x8c8__neon_mull_padal()
178 const int32x2_t vpsum0x6 = vadd_s32(vget_low_s32(vacc0x6), vget_high_s32(vacc0x6)); in xnn_qs8_gemm_minmax_ukernel_3x8c8__neon_mull_padal()
|
| D | 3x8c8-minmax-neon-mlal-padal.c | 62 …int32x4_t vacc0x6 = vld1q_lane_s32(w, vmovq_n_s32(0), 0); w = (const void*) ((uintptr_t) w + sizeo… in xnn_qs8_gemm_minmax_ukernel_3x8c8__neon_mlal_padal() local
70 int32x4_t vacc1x6 = vacc0x6; in xnn_qs8_gemm_minmax_ukernel_3x8c8__neon_mlal_padal()
78 int32x4_t vacc2x6 = vacc0x6; in xnn_qs8_gemm_minmax_ukernel_3x8c8__neon_mlal_padal()
167 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_gemm_minmax_ukernel_3x8c8__neon_mlal_padal()
236 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_gemm_minmax_ukernel_3x8c8__neon_mlal_padal()
254 const int32x4_t vsum0x67 = vpaddq_s32(vacc0x6, vacc0x7); in xnn_qs8_gemm_minmax_ukernel_3x8c8__neon_mlal_padal()
279 const int32x2_t vpsum0x6 = vadd_s32(vget_low_s32(vacc0x6), vget_high_s32(vacc0x6)); in xnn_qs8_gemm_minmax_ukernel_3x8c8__neon_mlal_padal()
|
| D | 1x16c8-minmax-neon-mull-padal.c | 50 …int32x4_t vacc0x6 = vld1q_lane_s32(w, vmovq_n_s32(0), 0); w = (const void*) ((uintptr_t) w + sizeo… in xnn_qs8_gemm_minmax_ukernel_1x16c8__neon_mull_padal() local
87 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_gemm_minmax_ukernel_1x16c8__neon_mull_padal()
123 const int32x4_t vsum0x67 = vpaddq_s32(vacc0x6, vacc0x7); in xnn_qs8_gemm_minmax_ukernel_1x16c8__neon_mull_padal()
142 const int32x2_t vpsum0x6 = vadd_s32(vget_low_s32(vacc0x6), vget_high_s32(vacc0x6)); in xnn_qs8_gemm_minmax_ukernel_1x16c8__neon_mull_padal()
|
| D | 4x8c8-minmax-neon-mull-padal.c | 68 …int32x4_t vacc0x6 = vld1q_lane_s32(w, vmovq_n_s32(0), 0); w = (const void*) ((uintptr_t) w + sizeo… in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mull_padal() local
76 int32x4_t vacc1x6 = vacc0x6; in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mull_padal()
84 int32x4_t vacc2x6 = vacc0x6; in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mull_padal()
92 int32x4_t vacc3x6 = vacc0x6; in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mull_padal()
163 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mull_padal()
184 const int32x4_t vsum0x67 = vpaddq_s32(vacc0x6, vacc0x7); in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mull_padal()
215 const int32x2_t vpsum0x6 = vadd_s32(vget_low_s32(vacc0x6), vget_high_s32(vacc0x6)); in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mull_padal()
|
| D | 3x8c16-minmax-neon-mlal-padal.c | 62 …int32x4_t vacc0x6 = vld1q_lane_s32(w, vmovq_n_s32(0), 0); w = (const void*) ((uintptr_t) w + sizeo… in xnn_qs8_gemm_minmax_ukernel_3x8c16__neon_mlal_padal() local
70 int32x4_t vacc1x6 = vacc0x6; in xnn_qs8_gemm_minmax_ukernel_3x8c16__neon_mlal_padal()
78 int32x4_t vacc2x6 = vacc0x6; in xnn_qs8_gemm_minmax_ukernel_3x8c16__neon_mlal_padal()
157 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_gemm_minmax_ukernel_3x8c16__neon_mlal_padal()
177 const int32x4_t vsum0x67 = vpaddq_s32(vacc0x6, vacc0x7); in xnn_qs8_gemm_minmax_ukernel_3x8c16__neon_mlal_padal()
202 const int32x2_t vpsum0x6 = vadd_s32(vget_low_s32(vacc0x6), vget_high_s32(vacc0x6)); in xnn_qs8_gemm_minmax_ukernel_3x8c16__neon_mlal_padal()
|
| D | 1x16c8-minmax-neon-mlal-padal.c | 50 …int32x4_t vacc0x6 = vld1q_lane_s32(w, vmovq_n_s32(0), 0); w = (const void*) ((uintptr_t) w + sizeo… in xnn_qs8_gemm_minmax_ukernel_1x16c8__neon_mlal_padal() local
111 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_gemm_minmax_ukernel_1x16c8__neon_mlal_padal()
176 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_gemm_minmax_ukernel_1x16c8__neon_mlal_padal()
212 const int32x4_t vsum0x67 = vpaddq_s32(vacc0x6, vacc0x7); in xnn_qs8_gemm_minmax_ukernel_1x16c8__neon_mlal_padal()
231 const int32x2_t vpsum0x6 = vadd_s32(vget_low_s32(vacc0x6), vget_high_s32(vacc0x6)); in xnn_qs8_gemm_minmax_ukernel_1x16c8__neon_mlal_padal()
|
| D | 4x8c8-minmax-neon-mlal-padal.c | 68 …int32x4_t vacc0x6 = vld1q_lane_s32(w, vmovq_n_s32(0), 0); w = (const void*) ((uintptr_t) w + sizeo… in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mlal_padal() local
76 int32x4_t vacc1x6 = vacc0x6; in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mlal_padal()
84 int32x4_t vacc2x6 = vacc0x6; in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mlal_padal()
92 int32x4_t vacc3x6 = vacc0x6; in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mlal_padal()
203 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mlal_padal()
290 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mlal_padal()
311 const int32x4_t vsum0x67 = vpaddq_s32(vacc0x6, vacc0x7); in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mlal_padal()
342 const int32x2_t vpsum0x6 = vadd_s32(vget_low_s32(vacc0x6), vget_high_s32(vacc0x6)); in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mlal_padal()
|
| /external/XNNPACK/src/qs8-igemm/gen/ |
| D | 1x8c8-minmax-neon-mull-padal.c | 53 …int32x4_t vacc0x6 = vld1q_lane_s32(w, vmovq_n_s32(0), 0); w = (const void*) ((uintptr_t) w + sizeo… in xnn_qs8_igemm_minmax_ukernel_1x8c8__neon_mull_padal() local
90 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_igemm_minmax_ukernel_1x8c8__neon_mull_padal()
105 const int32x4_t vsum0x67 = vpaddq_s32(vacc0x6, vacc0x7); in xnn_qs8_igemm_minmax_ukernel_1x8c8__neon_mull_padal()
118 const int32x2_t vpsum0x6 = vadd_s32(vget_low_s32(vacc0x6), vget_high_s32(vacc0x6)); in xnn_qs8_igemm_minmax_ukernel_1x8c8__neon_mull_padal()
|
| D | 1x8c8-minmax-neon-mlal-padal.c | 53 …int32x4_t vacc0x6 = vld1q_lane_s32(w, vmovq_n_s32(0), 0); w = (const void*) ((uintptr_t) w + sizeo… in xnn_qs8_igemm_minmax_ukernel_1x8c8__neon_mlal_padal() local
106 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_igemm_minmax_ukernel_1x8c8__neon_mlal_padal()
139 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_igemm_minmax_ukernel_1x8c8__neon_mlal_padal()
154 const int32x4_t vsum0x67 = vpaddq_s32(vacc0x6, vacc0x7); in xnn_qs8_igemm_minmax_ukernel_1x8c8__neon_mlal_padal()
167 const int32x2_t vpsum0x6 = vadd_s32(vget_low_s32(vacc0x6), vget_high_s32(vacc0x6)); in xnn_qs8_igemm_minmax_ukernel_1x8c8__neon_mlal_padal()
|
| D | 1x8c16-minmax-neon-mlal-padal.c | 53 …int32x4_t vacc0x6 = vld1q_lane_s32(w, vmovq_n_s32(0), 0); w = (const void*) ((uintptr_t) w + sizeo… in xnn_qs8_igemm_minmax_ukernel_1x8c16__neon_mlal_padal() local
98 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_igemm_minmax_ukernel_1x8c16__neon_mlal_padal()
113 const int32x4_t vsum0x67 = vpaddq_s32(vacc0x6, vacc0x7); in xnn_qs8_igemm_minmax_ukernel_1x8c16__neon_mlal_padal()
126 const int32x2_t vpsum0x6 = vadd_s32(vget_low_s32(vacc0x6), vget_high_s32(vacc0x6)); in xnn_qs8_igemm_minmax_ukernel_1x8c16__neon_mlal_padal()
|
| D | 2x8c8-minmax-neon-mull-padal.c | 57 …int32x4_t vacc0x6 = vld1q_lane_s32(w, vmovq_n_s32(0), 0); w = (const void*) ((uintptr_t) w + sizeo… in xnn_qs8_igemm_minmax_ukernel_2x8c8__neon_mull_padal() local
65 int32x4_t vacc1x6 = vacc0x6; in xnn_qs8_igemm_minmax_ukernel_2x8c8__neon_mull_padal()
120 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_igemm_minmax_ukernel_2x8c8__neon_mull_padal()
138 const int32x4_t vsum0x67 = vpaddq_s32(vacc0x6, vacc0x7); in xnn_qs8_igemm_minmax_ukernel_2x8c8__neon_mull_padal()
157 const int32x2_t vpsum0x6 = vadd_s32(vget_low_s32(vacc0x6), vget_high_s32(vacc0x6)); in xnn_qs8_igemm_minmax_ukernel_2x8c8__neon_mull_padal()
|
| D | 2x8c8-minmax-neon-mlal-padal.c | 57 …int32x4_t vacc0x6 = vld1q_lane_s32(w, vmovq_n_s32(0), 0); w = (const void*) ((uintptr_t) w + sizeo… in xnn_qs8_igemm_minmax_ukernel_2x8c8__neon_mlal_padal() local
65 int32x4_t vacc1x6 = vacc0x6; in xnn_qs8_igemm_minmax_ukernel_2x8c8__neon_mlal_padal()
144 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_igemm_minmax_ukernel_2x8c8__neon_mlal_padal()
195 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_igemm_minmax_ukernel_2x8c8__neon_mlal_padal()
213 const int32x4_t vsum0x67 = vpaddq_s32(vacc0x6, vacc0x7); in xnn_qs8_igemm_minmax_ukernel_2x8c8__neon_mlal_padal()
232 const int32x2_t vpsum0x6 = vadd_s32(vget_low_s32(vacc0x6), vget_high_s32(vacc0x6)); in xnn_qs8_igemm_minmax_ukernel_2x8c8__neon_mlal_padal()
|
| D | 2x8c16-minmax-neon-mlal-padal.c | 57 …int32x4_t vacc0x6 = vld1q_lane_s32(w, vmovq_n_s32(0), 0); w = (const void*) ((uintptr_t) w + sizeo… in xnn_qs8_igemm_minmax_ukernel_2x8c16__neon_mlal_padal() local
65 int32x4_t vacc1x6 = vacc0x6; in xnn_qs8_igemm_minmax_ukernel_2x8c16__neon_mlal_padal()
135 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_igemm_minmax_ukernel_2x8c16__neon_mlal_padal()
154 const int32x4_t vsum0x67 = vpaddq_s32(vacc0x6, vacc0x7); in xnn_qs8_igemm_minmax_ukernel_2x8c16__neon_mlal_padal()
173 const int32x2_t vpsum0x6 = vadd_s32(vget_low_s32(vacc0x6), vget_high_s32(vacc0x6)); in xnn_qs8_igemm_minmax_ukernel_2x8c16__neon_mlal_padal()
|
| D | 3x8c8-minmax-neon-mull-padal.c | 61 …int32x4_t vacc0x6 = vld1q_lane_s32(w, vmovq_n_s32(0), 0); w = (const void*) ((uintptr_t) w + sizeo… in xnn_qs8_igemm_minmax_ukernel_3x8c8__neon_mull_padal() local
69 int32x4_t vacc1x6 = vacc0x6; in xnn_qs8_igemm_minmax_ukernel_3x8c8__neon_mull_padal()
77 int32x4_t vacc2x6 = vacc0x6; in xnn_qs8_igemm_minmax_ukernel_3x8c8__neon_mull_padal()
150 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_igemm_minmax_ukernel_3x8c8__neon_mull_padal()
171 const int32x4_t vsum0x67 = vpaddq_s32(vacc0x6, vacc0x7); in xnn_qs8_igemm_minmax_ukernel_3x8c8__neon_mull_padal()
196 const int32x2_t vpsum0x6 = vadd_s32(vget_low_s32(vacc0x6), vget_high_s32(vacc0x6)); in xnn_qs8_igemm_minmax_ukernel_3x8c8__neon_mull_padal()
|
| D | 4x8c8-minmax-neon-mull-padal.c | 65 …int32x4_t vacc0x6 = vld1q_lane_s32(w, vmovq_n_s32(0), 0); w = (const void*) ((uintptr_t) w + sizeo… in xnn_qs8_igemm_minmax_ukernel_4x8c8__neon_mull_padal() local
73 int32x4_t vacc1x6 = vacc0x6; in xnn_qs8_igemm_minmax_ukernel_4x8c8__neon_mull_padal()
81 int32x4_t vacc2x6 = vacc0x6; in xnn_qs8_igemm_minmax_ukernel_4x8c8__neon_mull_padal()
89 int32x4_t vacc3x6 = vacc0x6; in xnn_qs8_igemm_minmax_ukernel_4x8c8__neon_mull_padal()
180 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_igemm_minmax_ukernel_4x8c8__neon_mull_padal()
204 const int32x4_t vsum0x67 = vpaddq_s32(vacc0x6, vacc0x7); in xnn_qs8_igemm_minmax_ukernel_4x8c8__neon_mull_padal()
235 const int32x2_t vpsum0x6 = vadd_s32(vget_low_s32(vacc0x6), vget_high_s32(vacc0x6)); in xnn_qs8_igemm_minmax_ukernel_4x8c8__neon_mull_padal()
|
| D | 1x16c8-minmax-neon-mull-padal.c | 53 …int32x4_t vacc0x6 = vld1q_lane_s32(w, vmovq_n_s32(0), 0); w = (const void*) ((uintptr_t) w + sizeo… in xnn_qs8_igemm_minmax_ukernel_1x16c8__neon_mull_padal() local
98 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_igemm_minmax_ukernel_1x16c8__neon_mull_padal()
137 const int32x4_t vsum0x67 = vpaddq_s32(vacc0x6, vacc0x7); in xnn_qs8_igemm_minmax_ukernel_1x16c8__neon_mull_padal()
156 const int32x2_t vpsum0x6 = vadd_s32(vget_low_s32(vacc0x6), vget_high_s32(vacc0x6)); in xnn_qs8_igemm_minmax_ukernel_1x16c8__neon_mull_padal()
|
| D | 1x16c8-minmax-neon-mlal-padal.c | 53 …int32x4_t vacc0x6 = vld1q_lane_s32(w, vmovq_n_s32(0), 0); w = (const void*) ((uintptr_t) w + sizeo… in xnn_qs8_igemm_minmax_ukernel_1x16c8__neon_mlal_padal() local
122 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_igemm_minmax_ukernel_1x16c8__neon_mlal_padal()
187 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_igemm_minmax_ukernel_1x16c8__neon_mlal_padal()
226 const int32x4_t vsum0x67 = vpaddq_s32(vacc0x6, vacc0x7); in xnn_qs8_igemm_minmax_ukernel_1x16c8__neon_mlal_padal()
245 const int32x2_t vpsum0x6 = vadd_s32(vget_low_s32(vacc0x6), vget_high_s32(vacc0x6)); in xnn_qs8_igemm_minmax_ukernel_1x16c8__neon_mlal_padal()
|
| D | 3x8c16-minmax-neon-mlal-padal.c | 61 …int32x4_t vacc0x6 = vld1q_lane_s32(w, vmovq_n_s32(0), 0); w = (const void*) ((uintptr_t) w + sizeo… in xnn_qs8_igemm_minmax_ukernel_3x8c16__neon_mlal_padal() local
69 int32x4_t vacc1x6 = vacc0x6; in xnn_qs8_igemm_minmax_ukernel_3x8c16__neon_mlal_padal()
77 int32x4_t vacc2x6 = vacc0x6; in xnn_qs8_igemm_minmax_ukernel_3x8c16__neon_mlal_padal()
172 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_igemm_minmax_ukernel_3x8c16__neon_mlal_padal()
195 const int32x4_t vsum0x67 = vpaddq_s32(vacc0x6, vacc0x7); in xnn_qs8_igemm_minmax_ukernel_3x8c16__neon_mlal_padal()
220 const int32x2_t vpsum0x6 = vadd_s32(vget_low_s32(vacc0x6), vget_high_s32(vacc0x6)); in xnn_qs8_igemm_minmax_ukernel_3x8c16__neon_mlal_padal()
|
| D | 3x8c8-minmax-neon-mlal-padal.c | 61 …int32x4_t vacc0x6 = vld1q_lane_s32(w, vmovq_n_s32(0), 0); w = (const void*) ((uintptr_t) w + sizeo… in xnn_qs8_igemm_minmax_ukernel_3x8c8__neon_mlal_padal() local
69 int32x4_t vacc1x6 = vacc0x6; in xnn_qs8_igemm_minmax_ukernel_3x8c8__neon_mlal_padal()
77 int32x4_t vacc2x6 = vacc0x6; in xnn_qs8_igemm_minmax_ukernel_3x8c8__neon_mlal_padal()
182 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_igemm_minmax_ukernel_3x8c8__neon_mlal_padal()
251 vacc0x6 = vpadalq_s16(vacc0x6, vprod0x6); in xnn_qs8_igemm_minmax_ukernel_3x8c8__neon_mlal_padal()
272 const int32x4_t vsum0x67 = vpaddq_s32(vacc0x6, vacc0x7); in xnn_qs8_igemm_minmax_ukernel_3x8c8__neon_mlal_padal()
297 const int32x2_t vpsum0x6 = vadd_s32(vget_low_s32(vacc0x6), vget_high_s32(vacc0x6)); in xnn_qs8_igemm_minmax_ukernel_3x8c8__neon_mlal_padal()
|