| /external/XNNPACK/src/f32-ppmm/gen/ |
| D | 4x2-minmax-scalar.c | 55 float vacc3x1 = vacc0x1; in xnn_f32_ppmm_minmax_ukernel_4x2__scalar() local
77 vacc3x1 += va3 * vb1; in xnn_f32_ppmm_minmax_ukernel_4x2__scalar()
90 vacc3x1 = math_min_f32(vacc3x1, vmax); in xnn_f32_ppmm_minmax_ukernel_4x2__scalar()
100 vacc3x1 = math_max_f32(vacc3x1, vmin); in xnn_f32_ppmm_minmax_ukernel_4x2__scalar()
104 c3[1] = vacc3x1; in xnn_f32_ppmm_minmax_ukernel_4x2__scalar()
|
| D | 4x4-minmax-scalar.c | 61 float vacc3x1 = vacc0x1; in xnn_f32_ppmm_minmax_ukernel_4x4__scalar() local
87 vacc3x1 += va3 * vb1; in xnn_f32_ppmm_minmax_ukernel_4x4__scalar()
108 vacc3x1 = math_min_f32(vacc3x1, vmax); in xnn_f32_ppmm_minmax_ukernel_4x4__scalar()
126 vacc3x1 = math_max_f32(vacc3x1, vmin); in xnn_f32_ppmm_minmax_ukernel_4x4__scalar()
138 c3[1] = vacc3x1; in xnn_f32_ppmm_minmax_ukernel_4x4__scalar()
165 c3[1] = vacc3x1; in xnn_f32_ppmm_minmax_ukernel_4x4__scalar()
|
| /external/XNNPACK/src/f32-spmm/gen/ |
| D | 8x2-minmax-scalar.c | 52 float vacc3x1 = vacc0x1; in xnn_f32_spmm_minmax_ukernel_8x2__scalar() local
82 vacc3x1 += vi3 * vw1; in xnn_f32_spmm_minmax_ukernel_8x2__scalar()
100 float vout3x1 = math_min_f32(vacc3x1, vmax); in xnn_f32_spmm_minmax_ukernel_8x2__scalar()
231 float vacc3x1 = vacc0x1; in xnn_f32_spmm_minmax_ukernel_8x2__scalar() local
249 vacc3x1 += vi3 * vw1; in xnn_f32_spmm_minmax_ukernel_8x2__scalar()
259 float vout3x1 = math_min_f32(vacc3x1, vmax); in xnn_f32_spmm_minmax_ukernel_8x2__scalar()
|
| D | 8x4-minmax-scalar.c | 52 float vacc3x1 = vacc0x1; in xnn_f32_spmm_minmax_ukernel_8x4__scalar() local
100 vacc3x1 += vi3 * vw1; in xnn_f32_spmm_minmax_ukernel_8x4__scalar()
134 float vout3x1 = math_min_f32(vacc3x1, vmax); in xnn_f32_spmm_minmax_ukernel_8x4__scalar()
315 float vacc3x1 = vacc0x1; in xnn_f32_spmm_minmax_ukernel_8x4__scalar() local
343 vacc3x1 += vi3 * vw1; in xnn_f32_spmm_minmax_ukernel_8x4__scalar()
361 float vout3x1 = math_min_f32(vacc3x1, vmax); in xnn_f32_spmm_minmax_ukernel_8x4__scalar()
|
| /external/XNNPACK/src/qs8-igemm/gen/ |
| D | 4x8c8-minmax-neon-mull-padal.c | 84 int32x4_t vacc3x1 = vacc0x1; in xnn_qs8_igemm_minmax_ukernel_4x8c8__neon_mull_padal() local
138 vacc3x1 = vpadalq_s16(vacc3x1, vprod3x1); in xnn_qs8_igemm_minmax_ukernel_4x8c8__neon_mull_padal()
213 const int32x4_t vsum3x01 = vpaddq_s32(vacc3x0, vacc3x1); in xnn_qs8_igemm_minmax_ukernel_4x8c8__neon_mull_padal()
269 const int32x2_t vpsum3x1 = vadd_s32(vget_low_s32(vacc3x1), vget_high_s32(vacc3x1)); in xnn_qs8_igemm_minmax_ukernel_4x8c8__neon_mull_padal()
|
| D | 4x8c8-minmax-neon-mlal-padal.c | 84 int32x4_t vacc3x1 = vacc0x1; in xnn_qs8_igemm_minmax_ukernel_4x8c8__neon_mlal_padal() local
158 vacc3x1 = vpadalq_s16(vacc3x1, vprod3x1); in xnn_qs8_igemm_minmax_ukernel_4x8c8__neon_mlal_padal()
265 vacc3x1 = vpadalq_s16(vacc3x1, vprod3x1); in xnn_qs8_igemm_minmax_ukernel_4x8c8__neon_mlal_padal()
340 const int32x4_t vsum3x01 = vpaddq_s32(vacc3x0, vacc3x1); in xnn_qs8_igemm_minmax_ukernel_4x8c8__neon_mlal_padal()
396 const int32x2_t vpsum3x1 = vadd_s32(vget_low_s32(vacc3x1), vget_high_s32(vacc3x1)); in xnn_qs8_igemm_minmax_ukernel_4x8c8__neon_mlal_padal()
|
| D | 4x8c16-minmax-neon-mlal-padal.c | 84 int32x4_t vacc3x1 = vacc0x1; in xnn_qs8_igemm_minmax_ukernel_4x8c16__neon_mlal_padal() local
152 vacc3x1 = vpadalq_s16(vacc3x1, vprod3x1); in xnn_qs8_igemm_minmax_ukernel_4x8c16__neon_mlal_padal()
245 const int32x4_t vsum3x01 = vpaddq_s32(vacc3x0, vacc3x1); in xnn_qs8_igemm_minmax_ukernel_4x8c16__neon_mlal_padal()
301 const int32x2_t vpsum3x1 = vadd_s32(vget_low_s32(vacc3x1), vget_high_s32(vacc3x1)); in xnn_qs8_igemm_minmax_ukernel_4x8c16__neon_mlal_padal()
|
| D | 4x16c8-minmax-neon-mull-padal.c | 108 int32x4_t vacc3x1 = vacc0x1; in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mull_padal() local
170 vacc3x1 = vpadalq_s16(vacc3x1, vprod3x1); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mull_padal()
329 const int32x4_t vsum3x01 = vpaddq_s32(vacc3x0, vacc3x1); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mull_padal()
439 const int32x2_t vpsum3x1 = vadd_s32(vget_low_s32(vacc3x1), vget_high_s32(vacc3x1)); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mull_padal()
|
| D | 4x16c8-minmax-neon-mlal-padal.c | 108 int32x4_t vacc3x1 = vacc0x1; in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal() local
198 vacc3x1 = vpadalq_s16(vacc3x1, vprod3x1); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal()
409 vacc3x1 = vpadalq_s16(vacc3x1, vprod3x1); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal()
568 const int32x4_t vsum3x01 = vpaddq_s32(vacc3x0, vacc3x1); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal()
678 const int32x2_t vpsum3x1 = vadd_s32(vget_low_s32(vacc3x1), vget_high_s32(vacc3x1)); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal()
|
| D | 4x16c16-minmax-neon-mlal-padal.c | 108 int32x4_t vacc3x1 = vacc0x1; in xnn_qs8_igemm_minmax_ukernel_4x16c16__neon_mlal_padal() local
192 vacc3x1 = vpadalq_s16(vacc3x1, vprod3x1); in xnn_qs8_igemm_minmax_ukernel_4x16c16__neon_mlal_padal()
393 const int32x4_t vsum3x01 = vpaddq_s32(vacc3x0, vacc3x1); in xnn_qs8_igemm_minmax_ukernel_4x16c16__neon_mlal_padal()
503 const int32x2_t vpsum3x1 = vadd_s32(vget_low_s32(vacc3x1), vget_high_s32(vacc3x1)); in xnn_qs8_igemm_minmax_ukernel_4x16c16__neon_mlal_padal()
|
| /external/XNNPACK/src/qs8-gemm/gen/ |
| D | 4x8c8-minmax-neon-mull-padal.c | 87 int32x4_t vacc3x1 = vacc0x1; in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mull_padal() local
121 vacc3x1 = vpadalq_s16(vacc3x1, vprod3x1); in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mull_padal()
193 const int32x4_t vsum3x01 = vpaddq_s32(vacc3x0, vacc3x1); in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mull_padal()
249 const int32x2_t vpsum3x1 = vadd_s32(vget_low_s32(vacc3x1), vget_high_s32(vacc3x1)); in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mull_padal()
|
| D | 4x8c8-minmax-neon-mlal-padal.c | 87 int32x4_t vacc3x1 = vacc0x1; in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mlal_padal() local
141 vacc3x1 = vpadalq_s16(vacc3x1, vprod3x1); in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mlal_padal()
248 vacc3x1 = vpadalq_s16(vacc3x1, vprod3x1); in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mlal_padal()
320 const int32x4_t vsum3x01 = vpaddq_s32(vacc3x0, vacc3x1); in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mlal_padal()
376 const int32x2_t vpsum3x1 = vadd_s32(vget_low_s32(vacc3x1), vget_high_s32(vacc3x1)); in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mlal_padal()
|
| D | 4x8c16-minmax-neon-mlal-padal.c | 87 int32x4_t vacc3x1 = vacc0x1; in xnn_qs8_gemm_minmax_ukernel_4x8c16__neon_mlal_padal() local
135 vacc3x1 = vpadalq_s16(vacc3x1, vprod3x1); in xnn_qs8_gemm_minmax_ukernel_4x8c16__neon_mlal_padal()
225 const int32x4_t vsum3x01 = vpaddq_s32(vacc3x0, vacc3x1); in xnn_qs8_gemm_minmax_ukernel_4x8c16__neon_mlal_padal()
281 const int32x2_t vpsum3x1 = vadd_s32(vget_low_s32(vacc3x1), vget_high_s32(vacc3x1)); in xnn_qs8_gemm_minmax_ukernel_4x8c16__neon_mlal_padal()
|
| D | 4x16c8-minmax-neon-mull-padal.c | 111 int32x4_t vacc3x1 = vacc0x1; in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mull_padal() local
153 vacc3x1 = vpadalq_s16(vacc3x1, vprod3x1); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mull_padal()
309 const int32x4_t vsum3x01 = vpaddq_s32(vacc3x0, vacc3x1); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mull_padal()
419 const int32x2_t vpsum3x1 = vadd_s32(vget_low_s32(vacc3x1), vget_high_s32(vacc3x1)); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mull_padal()
|
| D | 4x16c8-minmax-neon-mlal-padal.c | 111 int32x4_t vacc3x1 = vacc0x1; in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal() local
181 vacc3x1 = vpadalq_s16(vacc3x1, vprod3x1); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal()
392 vacc3x1 = vpadalq_s16(vacc3x1, vprod3x1); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal()
548 const int32x4_t vsum3x01 = vpaddq_s32(vacc3x0, vacc3x1); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal()
658 const int32x2_t vpsum3x1 = vadd_s32(vget_low_s32(vacc3x1), vget_high_s32(vacc3x1)); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal()
|
| D | 4x16c16-minmax-neon-mlal-padal.c | 111 int32x4_t vacc3x1 = vacc0x1; in xnn_qs8_gemm_minmax_ukernel_4x16c16__neon_mlal_padal() local
175 vacc3x1 = vpadalq_s16(vacc3x1, vprod3x1); in xnn_qs8_gemm_minmax_ukernel_4x16c16__neon_mlal_padal()
373 const int32x4_t vsum3x01 = vpaddq_s32(vacc3x0, vacc3x1); in xnn_qs8_gemm_minmax_ukernel_4x16c16__neon_mlal_padal()
483 const int32x2_t vpsum3x1 = vadd_s32(vget_low_s32(vacc3x1), vget_high_s32(vacc3x1)); in xnn_qs8_gemm_minmax_ukernel_4x16c16__neon_mlal_padal()
|