| /external/XNNPACK/src/f32-ppmm/gen/ |
| D | 4x2-minmax-scalar.c | 54 float vacc3x0 = vacc0x0; in xnn_f32_ppmm_minmax_ukernel_4x2__scalar() local
73 vacc3x0 += va3 * vb0; in xnn_f32_ppmm_minmax_ukernel_4x2__scalar()
86 vacc3x0 = math_min_f32(vacc3x0, vmax); in xnn_f32_ppmm_minmax_ukernel_4x2__scalar()
96 vacc3x0 = math_max_f32(vacc3x0, vmin); in xnn_f32_ppmm_minmax_ukernel_4x2__scalar()
103 c3[0] = vacc3x0; in xnn_f32_ppmm_minmax_ukernel_4x2__scalar()
122 *c3 = vacc3x0; in xnn_f32_ppmm_minmax_ukernel_4x2__scalar()
|
| D | 4x4-minmax-scalar.c | 60 float vacc3x0 = vacc0x0; in xnn_f32_ppmm_minmax_ukernel_4x4__scalar() local
83 vacc3x0 += va3 * vb0; in xnn_f32_ppmm_minmax_ukernel_4x4__scalar()
104 vacc3x0 = math_min_f32(vacc3x0, vmax); in xnn_f32_ppmm_minmax_ukernel_4x4__scalar()
122 vacc3x0 = math_max_f32(vacc3x0, vmin); in xnn_f32_ppmm_minmax_ukernel_4x4__scalar()
137 c3[0] = vacc3x0; in xnn_f32_ppmm_minmax_ukernel_4x4__scalar()
164 c3[0] = vacc3x0; in xnn_f32_ppmm_minmax_ukernel_4x4__scalar()
173 vacc3x0 = vacc3x2; in xnn_f32_ppmm_minmax_ukernel_4x4__scalar()
184 *c3 = vacc3x0; in xnn_f32_ppmm_minmax_ukernel_4x4__scalar()
|
| /external/XNNPACK/src/f32-spmm/gen/ |
| D | 8x1-minmax-scalar.c | 44 float vacc3x0 = vacc0x0; in xnn_f32_spmm_minmax_ukernel_8x1__scalar() local
65 vacc3x0 += vi3 * vw0; in xnn_f32_spmm_minmax_ukernel_8x1__scalar()
75 float vout3x0 = math_min_f32(vacc3x0, vmax); in xnn_f32_spmm_minmax_ukernel_8x1__scalar()
185 float vacc3x0 = vacc0x0; in xnn_f32_spmm_minmax_ukernel_8x1__scalar() local
198 vacc3x0 += vi3 * vw0; in xnn_f32_spmm_minmax_ukernel_8x1__scalar()
204 float vout3x0 = math_min_f32(vacc3x0, vmax); in xnn_f32_spmm_minmax_ukernel_8x1__scalar()
|
| D | 8x2-minmax-scalar.c | 44 float vacc3x0 = vacc0x0; in xnn_f32_spmm_minmax_ukernel_8x2__scalar() local
74 vacc3x0 += vi3 * vw0; in xnn_f32_spmm_minmax_ukernel_8x2__scalar()
92 float vout3x0 = math_min_f32(vacc3x0, vmax); in xnn_f32_spmm_minmax_ukernel_8x2__scalar()
227 float vacc3x0 = vacc0x0; in xnn_f32_spmm_minmax_ukernel_8x2__scalar() local
245 vacc3x0 += vi3 * vw0; in xnn_f32_spmm_minmax_ukernel_8x2__scalar()
255 float vout3x0 = math_min_f32(vacc3x0, vmax); in xnn_f32_spmm_minmax_ukernel_8x2__scalar()
|
| D | 4x1-minmax-scalar.c | 44 float vacc3x0 = vacc0x0; in xnn_f32_spmm_minmax_ukernel_4x1__scalar() local
57 vacc3x0 += vi3 * vw0; in xnn_f32_spmm_minmax_ukernel_4x1__scalar()
63 float vout3x0 = math_min_f32(vacc3x0, vmax); in xnn_f32_spmm_minmax_ukernel_4x1__scalar()
|
| D | 8x4-minmax-scalar.c | 44 float vacc3x0 = vacc0x0; in xnn_f32_spmm_minmax_ukernel_8x4__scalar() local
92 vacc3x0 += vi3 * vw0; in xnn_f32_spmm_minmax_ukernel_8x4__scalar()
126 float vout3x0 = math_min_f32(vacc3x0, vmax); in xnn_f32_spmm_minmax_ukernel_8x4__scalar()
311 float vacc3x0 = vacc0x0; in xnn_f32_spmm_minmax_ukernel_8x4__scalar() local
339 vacc3x0 += vi3 * vw0; in xnn_f32_spmm_minmax_ukernel_8x4__scalar()
357 float vout3x0 = math_min_f32(vacc3x0, vmax); in xnn_f32_spmm_minmax_ukernel_8x4__scalar()
|
| /external/XNNPACK/src/qs8-igemm/gen/ |
| D | 4x8c8-minmax-neon-mull-padal.c | 83 int32x4_t vacc3x0 = vacc0x0; in xnn_qs8_igemm_minmax_ukernel_4x8c8__neon_mull_padal() local
129 vacc3x0 = vpadalq_s16(vacc3x0, vprod3x0); 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()
268 const int32x2_t vpsum3x0 = vadd_s32(vget_low_s32(vacc3x0), vget_high_s32(vacc3x0)); in xnn_qs8_igemm_minmax_ukernel_4x8c8__neon_mull_padal()
|
| D | 4x8c8-minmax-neon-mlal-padal.c | 83 int32x4_t vacc3x0 = vacc0x0; in xnn_qs8_igemm_minmax_ukernel_4x8c8__neon_mlal_padal() local
145 vacc3x0 = vpadalq_s16(vacc3x0, vprod3x0); in xnn_qs8_igemm_minmax_ukernel_4x8c8__neon_mlal_padal()
256 vacc3x0 = vpadalq_s16(vacc3x0, vprod3x0); 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()
395 const int32x2_t vpsum3x0 = vadd_s32(vget_low_s32(vacc3x0), vget_high_s32(vacc3x0)); in xnn_qs8_igemm_minmax_ukernel_4x8c8__neon_mlal_padal()
|
| D | 4x8c16-minmax-neon-mlal-padal.c | 83 int32x4_t vacc3x0 = vacc0x0; in xnn_qs8_igemm_minmax_ukernel_4x8c16__neon_mlal_padal() local
140 vacc3x0 = vpadalq_s16(vacc3x0, vprod3x0); 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()
300 const int32x2_t vpsum3x0 = vadd_s32(vget_low_s32(vacc3x0), vget_high_s32(vacc3x0)); in xnn_qs8_igemm_minmax_ukernel_4x8c16__neon_mlal_padal()
|
| D | 4x16c8-minmax-neon-mull-padal.c | 107 int32x4_t vacc3x0 = vacc0x0; in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mull_padal() local
161 vacc3x0 = vpadalq_s16(vacc3x0, vprod3x0); 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()
438 const int32x2_t vpsum3x0 = vadd_s32(vget_low_s32(vacc3x0), vget_high_s32(vacc3x0)); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mull_padal()
|
| D | 4x16c8-minmax-neon-mlal-padal.c | 107 int32x4_t vacc3x0 = vacc0x0; in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal() local
185 vacc3x0 = vpadalq_s16(vacc3x0, vprod3x0); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal()
400 vacc3x0 = vpadalq_s16(vacc3x0, vprod3x0); 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()
677 const int32x2_t vpsum3x0 = vadd_s32(vget_low_s32(vacc3x0), vget_high_s32(vacc3x0)); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal()
|
| D | 4x16c16-minmax-neon-mlal-padal.c | 107 int32x4_t vacc3x0 = vacc0x0; in xnn_qs8_igemm_minmax_ukernel_4x16c16__neon_mlal_padal() local
180 vacc3x0 = vpadalq_s16(vacc3x0, vprod3x0); 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()
502 const int32x2_t vpsum3x0 = vadd_s32(vget_low_s32(vacc3x0), vget_high_s32(vacc3x0)); in xnn_qs8_igemm_minmax_ukernel_4x16c16__neon_mlal_padal()
|
| /external/XNNPACK/src/qs8-gemm/gen/ |
| D | 4x8c8-minmax-neon-mull-padal.c | 86 int32x4_t vacc3x0 = vacc0x0; in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mull_padal() local
112 vacc3x0 = vpadalq_s16(vacc3x0, vprod3x0); 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()
248 const int32x2_t vpsum3x0 = vadd_s32(vget_low_s32(vacc3x0), vget_high_s32(vacc3x0)); in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mull_padal()
|
| D | 4x8c8-minmax-neon-mlal-padal.c | 86 int32x4_t vacc3x0 = vacc0x0; in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mlal_padal() local
128 vacc3x0 = vpadalq_s16(vacc3x0, vprod3x0); in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mlal_padal()
239 vacc3x0 = vpadalq_s16(vacc3x0, vprod3x0); 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()
375 const int32x2_t vpsum3x0 = vadd_s32(vget_low_s32(vacc3x0), vget_high_s32(vacc3x0)); in xnn_qs8_gemm_minmax_ukernel_4x8c8__neon_mlal_padal()
|
| D | 4x8c16-minmax-neon-mlal-padal.c | 86 int32x4_t vacc3x0 = vacc0x0; in xnn_qs8_gemm_minmax_ukernel_4x8c16__neon_mlal_padal() local
123 vacc3x0 = vpadalq_s16(vacc3x0, vprod3x0); 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()
280 const int32x2_t vpsum3x0 = vadd_s32(vget_low_s32(vacc3x0), vget_high_s32(vacc3x0)); in xnn_qs8_gemm_minmax_ukernel_4x8c16__neon_mlal_padal()
|
| D | 4x16c8-minmax-neon-mull-padal.c | 110 int32x4_t vacc3x0 = vacc0x0; in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mull_padal() local
144 vacc3x0 = vpadalq_s16(vacc3x0, vprod3x0); 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()
418 const int32x2_t vpsum3x0 = vadd_s32(vget_low_s32(vacc3x0), vget_high_s32(vacc3x0)); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mull_padal()
|
| D | 4x16c8-minmax-neon-mlal-padal.c | 110 int32x4_t vacc3x0 = vacc0x0; in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal() local
168 vacc3x0 = vpadalq_s16(vacc3x0, vprod3x0); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal()
383 vacc3x0 = vpadalq_s16(vacc3x0, vprod3x0); 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()
657 const int32x2_t vpsum3x0 = vadd_s32(vget_low_s32(vacc3x0), vget_high_s32(vacc3x0)); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal()
|
| D | 4x16c16-minmax-neon-mlal-padal.c | 110 int32x4_t vacc3x0 = vacc0x0; in xnn_qs8_gemm_minmax_ukernel_4x16c16__neon_mlal_padal() local
163 vacc3x0 = vpadalq_s16(vacc3x0, vprod3x0); 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()
482 const int32x2_t vpsum3x0 = vadd_s32(vget_low_s32(vacc3x0), vget_high_s32(vacc3x0)); in xnn_qs8_gemm_minmax_ukernel_4x16c16__neon_mlal_padal()
|