| /external/XNNPACK/src/f32-vmulcaddc/gen/ |
| D | c8-minmax-wasmsimd-x86-2x.c | 90 v128_t vacc0 = wasm_v128_load(i0); in xnn_f32_vmulcaddc_minmax_ukernel_c8__wasmsimd_x86_2x() local
97 vacc0 = wasm_f32x4_add(vbias, wasm_f32x4_mul(vscale, vacc0)); in xnn_f32_vmulcaddc_minmax_ukernel_c8__wasmsimd_x86_2x()
100 vacc0 = wasm_v128_bitselect(vmin, vacc0, wasm_f32x4_lt(vacc0, vmin)); in xnn_f32_vmulcaddc_minmax_ukernel_c8__wasmsimd_x86_2x()
103 vacc0 = wasm_v128_bitselect(vacc0, vmax, wasm_f32x4_le(vacc0, vmax)); in xnn_f32_vmulcaddc_minmax_ukernel_c8__wasmsimd_x86_2x()
106 wasm_v128_store(o0, vacc0); in xnn_f32_vmulcaddc_minmax_ukernel_c8__wasmsimd_x86_2x()
116 v128_t vacc0 = wasm_v128_load(i0); in xnn_f32_vmulcaddc_minmax_ukernel_c8__wasmsimd_x86_2x() local
123 vacc0 = wasm_f32x4_add(vbias, wasm_f32x4_mul(vscale, vacc0)); in xnn_f32_vmulcaddc_minmax_ukernel_c8__wasmsimd_x86_2x()
126 vacc0 = wasm_v128_bitselect(vmin, vacc0, wasm_f32x4_lt(vacc0, vmin)); in xnn_f32_vmulcaddc_minmax_ukernel_c8__wasmsimd_x86_2x()
129 vacc0 = wasm_v128_bitselect(vacc0, vmax, wasm_f32x4_le(vacc0, vmax)); in xnn_f32_vmulcaddc_minmax_ukernel_c8__wasmsimd_x86_2x()
133 *((double*) o0) = wasm_f64x2_extract_lane(vacc0, 0); in xnn_f32_vmulcaddc_minmax_ukernel_c8__wasmsimd_x86_2x()
[all …]
|
| D | c8-minmax-wasmsimd-arm-2x.c | 90 v128_t vacc0 = wasm_v128_load(i0); in xnn_f32_vmulcaddc_minmax_ukernel_c8__wasmsimd_arm_2x() local
97 vacc0 = wasm_f32x4_add(vbias, wasm_f32x4_mul(vscale, vacc0)); in xnn_f32_vmulcaddc_minmax_ukernel_c8__wasmsimd_arm_2x()
100 vacc0 = wasm_f32x4_max(vacc0, vmin); in xnn_f32_vmulcaddc_minmax_ukernel_c8__wasmsimd_arm_2x()
103 vacc0 = wasm_f32x4_min(vacc0, vmax); in xnn_f32_vmulcaddc_minmax_ukernel_c8__wasmsimd_arm_2x()
106 wasm_v128_store(o0, vacc0); in xnn_f32_vmulcaddc_minmax_ukernel_c8__wasmsimd_arm_2x()
116 v128_t vacc0 = wasm_v128_load(i0); in xnn_f32_vmulcaddc_minmax_ukernel_c8__wasmsimd_arm_2x() local
123 vacc0 = wasm_f32x4_add(vbias, wasm_f32x4_mul(vscale, vacc0)); in xnn_f32_vmulcaddc_minmax_ukernel_c8__wasmsimd_arm_2x()
126 vacc0 = wasm_f32x4_max(vacc0, vmin); in xnn_f32_vmulcaddc_minmax_ukernel_c8__wasmsimd_arm_2x()
129 vacc0 = wasm_f32x4_min(vacc0, vmax); in xnn_f32_vmulcaddc_minmax_ukernel_c8__wasmsimd_arm_2x()
133 *((double*) o0) = wasm_f64x2_extract_lane(vacc0, 0); in xnn_f32_vmulcaddc_minmax_ukernel_c8__wasmsimd_arm_2x()
[all …]
|
| D | c4-minmax-wasmsimd-x86-2x.c | 78 v128_t vacc0 = wasm_v128_load(i0); in xnn_f32_vmulcaddc_minmax_ukernel_c4__wasmsimd_x86_2x() local
85 vacc0 = wasm_f32x4_add(vbias, wasm_f32x4_mul(vscale, vacc0)); in xnn_f32_vmulcaddc_minmax_ukernel_c4__wasmsimd_x86_2x()
88 vacc0 = wasm_v128_bitselect(vmin, vacc0, wasm_f32x4_lt(vacc0, vmin)); in xnn_f32_vmulcaddc_minmax_ukernel_c4__wasmsimd_x86_2x()
91 vacc0 = wasm_v128_bitselect(vacc0, vmax, wasm_f32x4_le(vacc0, vmax)); in xnn_f32_vmulcaddc_minmax_ukernel_c4__wasmsimd_x86_2x()
95 *((double*) o0) = wasm_f64x2_extract_lane(vacc0, 0); in xnn_f32_vmulcaddc_minmax_ukernel_c4__wasmsimd_x86_2x()
98 vacc0 = wasm_v32x4_shuffle(vacc0, vacc0, 2, 3, 2, 3); in xnn_f32_vmulcaddc_minmax_ukernel_c4__wasmsimd_x86_2x()
105 *o0++ = wasm_f32x4_extract_lane(vacc0, 0); in xnn_f32_vmulcaddc_minmax_ukernel_c4__wasmsimd_x86_2x()
|
| D | c4-minmax-wasmsimd-arm-2x.c | 78 v128_t vacc0 = wasm_v128_load(i0); in xnn_f32_vmulcaddc_minmax_ukernel_c4__wasmsimd_arm_2x() local
85 vacc0 = wasm_f32x4_add(vbias, wasm_f32x4_mul(vscale, vacc0)); in xnn_f32_vmulcaddc_minmax_ukernel_c4__wasmsimd_arm_2x()
88 vacc0 = wasm_f32x4_max(vacc0, vmin); in xnn_f32_vmulcaddc_minmax_ukernel_c4__wasmsimd_arm_2x()
91 vacc0 = wasm_f32x4_min(vacc0, vmax); in xnn_f32_vmulcaddc_minmax_ukernel_c4__wasmsimd_arm_2x()
95 *((double*) o0) = wasm_f64x2_extract_lane(vacc0, 0); in xnn_f32_vmulcaddc_minmax_ukernel_c4__wasmsimd_arm_2x()
98 vacc0 = wasm_v32x4_shuffle(vacc0, vacc0, 2, 3, 2, 3); in xnn_f32_vmulcaddc_minmax_ukernel_c4__wasmsimd_arm_2x()
105 *o0++ = wasm_f32x4_extract_lane(vacc0, 0); in xnn_f32_vmulcaddc_minmax_ukernel_c4__wasmsimd_arm_2x()
|
| D | c1-minmax-scalar-2x.c | 50 float vacc0 = *i0++; in xnn_f32_vmulcaddc_minmax_ukernel_c1__scalar_2x() local
55 vacc0 = vacc0 * vscale + vbias; in xnn_f32_vmulcaddc_minmax_ukernel_c1__scalar_2x()
58 vacc0 = math_max_f32(vacc0, vmin); in xnn_f32_vmulcaddc_minmax_ukernel_c1__scalar_2x()
61 vacc0 = math_min_f32(vacc0, vmax); in xnn_f32_vmulcaddc_minmax_ukernel_c1__scalar_2x()
64 *o0++ = vacc0; in xnn_f32_vmulcaddc_minmax_ukernel_c1__scalar_2x()
|
| D | c1-minmax-wasm-2x.c | 50 float vacc0 = *i0++; in xnn_f32_vmulcaddc_minmax_ukernel_c1__wasm_2x() local
55 vacc0 = vacc0 * vscale + vbias; in xnn_f32_vmulcaddc_minmax_ukernel_c1__wasm_2x()
58 vacc0 = __builtin_wasm_max_f32(vacc0, vmin); in xnn_f32_vmulcaddc_minmax_ukernel_c1__wasm_2x()
61 vacc0 = __builtin_wasm_min_f32(vacc0, vmax); in xnn_f32_vmulcaddc_minmax_ukernel_c1__wasm_2x()
64 *o0++ = vacc0; in xnn_f32_vmulcaddc_minmax_ukernel_c1__wasm_2x()
|
| D | c2-minmax-wasm-2x.c | 89 float vacc0 = *i0++; in xnn_f32_vmulcaddc_minmax_ukernel_c2__wasm_2x() local
94 vacc0 = vacc0 * vscale + vbias; in xnn_f32_vmulcaddc_minmax_ukernel_c2__wasm_2x()
97 vacc0 = __builtin_wasm_max_f32(vacc0, vmin); in xnn_f32_vmulcaddc_minmax_ukernel_c2__wasm_2x()
100 vacc0 = __builtin_wasm_min_f32(vacc0, vmax); in xnn_f32_vmulcaddc_minmax_ukernel_c2__wasm_2x()
103 *o0++ = vacc0; in xnn_f32_vmulcaddc_minmax_ukernel_c2__wasm_2x()
|
| /external/XNNPACK/src/f32-raddexpminusmax/gen/ |
| D | avx512f-p5-scalef-x192.c | 39 __m512 vacc0 = _mm512_setzero_ps(); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x192() local
195 vacc0 = _mm512_add_ps(vacc0, vf0); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x192()
196 vacc0 = _mm512_add_ps(vacc0, vf1); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x192()
197 vacc0 = _mm512_add_ps(vacc0, vf2); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x192()
198 vacc0 = _mm512_add_ps(vacc0, vf3); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x192()
199 vacc0 = _mm512_add_ps(vacc0, vf4); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x192()
200 vacc0 = _mm512_add_ps(vacc0, vf5); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x192()
201 vacc0 = _mm512_add_ps(vacc0, vf6); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x192()
202 vacc0 = _mm512_add_ps(vacc0, vf7); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x192()
203 vacc0 = _mm512_add_ps(vacc0, vf8); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x192()
[all …]
|
| D | avx512f-p5-scalef-x160.c | 39 __m512 vacc0 = _mm512_setzero_ps(); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x160() local
173 vacc0 = _mm512_add_ps(vacc0, vf0); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x160()
174 vacc0 = _mm512_add_ps(vacc0, vf1); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x160()
175 vacc0 = _mm512_add_ps(vacc0, vf2); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x160()
176 vacc0 = _mm512_add_ps(vacc0, vf3); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x160()
177 vacc0 = _mm512_add_ps(vacc0, vf4); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x160()
178 vacc0 = _mm512_add_ps(vacc0, vf5); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x160()
179 vacc0 = _mm512_add_ps(vacc0, vf6); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x160()
180 vacc0 = _mm512_add_ps(vacc0, vf7); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x160()
181 vacc0 = _mm512_add_ps(vacc0, vf8); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x160()
[all …]
|
| D | avx512f-p5-scalef-x144.c | 39 __m512 vacc0 = _mm512_setzero_ps(); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x144() local
162 vacc0 = _mm512_add_ps(vacc0, vf0); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x144()
163 vacc0 = _mm512_add_ps(vacc0, vf1); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x144()
164 vacc0 = _mm512_add_ps(vacc0, vf2); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x144()
165 vacc0 = _mm512_add_ps(vacc0, vf3); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x144()
166 vacc0 = _mm512_add_ps(vacc0, vf4); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x144()
167 vacc0 = _mm512_add_ps(vacc0, vf5); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x144()
168 vacc0 = _mm512_add_ps(vacc0, vf6); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x144()
169 vacc0 = _mm512_add_ps(vacc0, vf7); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x144()
170 vacc0 = _mm512_add_ps(vacc0, vf8); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x144()
[all …]
|
| D | avx512f-p5-scalef-x128.c | 39 __m512 vacc0 = _mm512_setzero_ps(); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x128() local
151 vacc0 = _mm512_add_ps(vacc0, vf0); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x128()
152 vacc0 = _mm512_add_ps(vacc0, vf1); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x128()
153 vacc0 = _mm512_add_ps(vacc0, vf2); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x128()
154 vacc0 = _mm512_add_ps(vacc0, vf3); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x128()
155 vacc0 = _mm512_add_ps(vacc0, vf4); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x128()
156 vacc0 = _mm512_add_ps(vacc0, vf5); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x128()
157 vacc0 = _mm512_add_ps(vacc0, vf6); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x128()
158 vacc0 = _mm512_add_ps(vacc0, vf7); in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x128()
161 __m512 vacc = vacc0; in xnn_f32_raddexpminusmax_ukernel__avx512f_p5_scalef_x128()
|
| /external/XNNPACK/src/f32-raddstoreexpminusmax/gen/ |
| D | avx512f-p5-scalef-x192.c | 40 __m512 vacc0 = _mm512_setzero_ps(); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x192() local
211 vacc0 = _mm512_add_ps(vacc0, vf0); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x192()
212 vacc0 = _mm512_add_ps(vacc0, vf1); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x192()
213 vacc0 = _mm512_add_ps(vacc0, vf2); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x192()
214 vacc0 = _mm512_add_ps(vacc0, vf3); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x192()
215 vacc0 = _mm512_add_ps(vacc0, vf4); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x192()
216 vacc0 = _mm512_add_ps(vacc0, vf5); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x192()
217 vacc0 = _mm512_add_ps(vacc0, vf6); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x192()
218 vacc0 = _mm512_add_ps(vacc0, vf7); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x192()
219 vacc0 = _mm512_add_ps(vacc0, vf8); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x192()
[all …]
|
| D | avx512f-p5-scalef-x160.c | 40 __m512 vacc0 = _mm512_setzero_ps(); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x160() local
187 vacc0 = _mm512_add_ps(vacc0, vf0); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x160()
188 vacc0 = _mm512_add_ps(vacc0, vf1); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x160()
189 vacc0 = _mm512_add_ps(vacc0, vf2); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x160()
190 vacc0 = _mm512_add_ps(vacc0, vf3); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x160()
191 vacc0 = _mm512_add_ps(vacc0, vf4); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x160()
192 vacc0 = _mm512_add_ps(vacc0, vf5); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x160()
193 vacc0 = _mm512_add_ps(vacc0, vf6); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x160()
194 vacc0 = _mm512_add_ps(vacc0, vf7); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x160()
195 vacc0 = _mm512_add_ps(vacc0, vf8); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x160()
[all …]
|
| D | avx512f-p5-scalef-x144.c | 40 __m512 vacc0 = _mm512_setzero_ps(); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x144() local
175 vacc0 = _mm512_add_ps(vacc0, vf0); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x144()
176 vacc0 = _mm512_add_ps(vacc0, vf1); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x144()
177 vacc0 = _mm512_add_ps(vacc0, vf2); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x144()
178 vacc0 = _mm512_add_ps(vacc0, vf3); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x144()
179 vacc0 = _mm512_add_ps(vacc0, vf4); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x144()
180 vacc0 = _mm512_add_ps(vacc0, vf5); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x144()
181 vacc0 = _mm512_add_ps(vacc0, vf6); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x144()
182 vacc0 = _mm512_add_ps(vacc0, vf7); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x144()
183 vacc0 = _mm512_add_ps(vacc0, vf8); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x144()
[all …]
|
| D | wasmsimd-p5-x16-acc4.c | 43 v128_t vacc0 = wasm_f32x4_splat(0.0f); in xnn_f32_raddstoreexpminusmax_ukernel__wasmsimd_p5_x16_acc4() local
44 v128_t vacc1 = vacc0; in xnn_f32_raddstoreexpminusmax_ukernel__wasmsimd_p5_x16_acc4()
45 v128_t vacc2 = vacc0; in xnn_f32_raddstoreexpminusmax_ukernel__wasmsimd_p5_x16_acc4()
46 v128_t vacc3 = vacc0; in xnn_f32_raddstoreexpminusmax_ukernel__wasmsimd_p5_x16_acc4()
142 vacc0 = wasm_f32x4_add(vacc0, vf0123); in xnn_f32_raddstoreexpminusmax_ukernel__wasmsimd_p5_x16_acc4()
143 vacc0 = wasm_f32x4_add(vacc0, vf4567); in xnn_f32_raddstoreexpminusmax_ukernel__wasmsimd_p5_x16_acc4()
144 vacc0 = wasm_f32x4_add(vacc0, vf89AB); in xnn_f32_raddstoreexpminusmax_ukernel__wasmsimd_p5_x16_acc4()
145 vacc0 = wasm_f32x4_add(vacc0, vfCDEF); in xnn_f32_raddstoreexpminusmax_ukernel__wasmsimd_p5_x16_acc4()
148 vacc0 = wasm_f32x4_add(vacc0, vacc1); in xnn_f32_raddstoreexpminusmax_ukernel__wasmsimd_p5_x16_acc4()
150 vacc0 = wasm_f32x4_add(vacc0, vacc2); in xnn_f32_raddstoreexpminusmax_ukernel__wasmsimd_p5_x16_acc4()
[all …]
|
| D | avx512f-p5-scalef-x128.c | 40 __m512 vacc0 = _mm512_setzero_ps(); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x128() local
163 vacc0 = _mm512_add_ps(vacc0, vf0); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x128()
164 vacc0 = _mm512_add_ps(vacc0, vf1); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x128()
165 vacc0 = _mm512_add_ps(vacc0, vf2); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x128()
166 vacc0 = _mm512_add_ps(vacc0, vf3); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x128()
167 vacc0 = _mm512_add_ps(vacc0, vf4); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x128()
168 vacc0 = _mm512_add_ps(vacc0, vf5); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x128()
169 vacc0 = _mm512_add_ps(vacc0, vf6); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x128()
170 vacc0 = _mm512_add_ps(vacc0, vf7); in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x128()
173 __m512 vacc = vacc0; in xnn_f32_raddstoreexpminusmax_ukernel__avx512f_p5_scalef_x128()
|
| /external/XNNPACK/src/f32-spmm/gen/ |
| D | 8x1-minmax-scalar.c | 110 float vacc0 = *w++; in xnn_f32_spmm_minmax_ukernel_8x1__scalar() local
111 float vacc1 = vacc0; in xnn_f32_spmm_minmax_ukernel_8x1__scalar()
112 float vacc2 = vacc0; in xnn_f32_spmm_minmax_ukernel_8x1__scalar()
113 float vacc3 = vacc0; in xnn_f32_spmm_minmax_ukernel_8x1__scalar()
114 float vacc4 = vacc0; in xnn_f32_spmm_minmax_ukernel_8x1__scalar()
115 float vacc5 = vacc0; in xnn_f32_spmm_minmax_ukernel_8x1__scalar()
116 float vacc6 = vacc0; in xnn_f32_spmm_minmax_ukernel_8x1__scalar()
117 float vacc7 = vacc0; in xnn_f32_spmm_minmax_ukernel_8x1__scalar()
131 vacc0 += vi0 * vw; in xnn_f32_spmm_minmax_ukernel_8x1__scalar()
141 float vout0 = math_min_f32(vacc0, vmax); in xnn_f32_spmm_minmax_ukernel_8x1__scalar()
[all …]
|
| /external/XNNPACK/src/f32-hswish/gen/ |
| D | hswish-wasm-x2.c | 38 float vacc0 = vx0 + vthree; in xnn_f32_hswish_ukernel__wasm_x2() local
43 vacc0 = __builtin_wasm_max_f32(vacc0, vzero); in xnn_f32_hswish_ukernel__wasm_x2()
46 vacc0 = __builtin_wasm_min_f32(vacc0, vsix); in xnn_f32_hswish_ukernel__wasm_x2()
49 vacc0 *= vx0; in xnn_f32_hswish_ukernel__wasm_x2()
52 y[0] = vacc0; in xnn_f32_hswish_ukernel__wasm_x2()
|
| D | hswish-scalar-x2.c | 38 float vacc0 = vx0 + vthree; in xnn_f32_hswish_ukernel__scalar_x2() local
43 vacc0 = math_max_f32(vacc0, vzero); in xnn_f32_hswish_ukernel__scalar_x2()
46 vacc0 = math_min_f32(vacc0, vsix); in xnn_f32_hswish_ukernel__scalar_x2()
49 vacc0 *= vx0; in xnn_f32_hswish_ukernel__scalar_x2()
52 y[0] = vacc0; in xnn_f32_hswish_ukernel__scalar_x2()
|
| D | hswish-wasm-x4.c | 40 float vacc0 = vx0 + vthree; in xnn_f32_hswish_ukernel__wasm_x4() local
49 vacc0 = __builtin_wasm_max_f32(vacc0, vzero); in xnn_f32_hswish_ukernel__wasm_x4()
54 vacc0 = __builtin_wasm_min_f32(vacc0, vsix); in xnn_f32_hswish_ukernel__wasm_x4()
59 vacc0 *= vx0; in xnn_f32_hswish_ukernel__wasm_x4()
64 y[0] = vacc0; in xnn_f32_hswish_ukernel__wasm_x4()
|
| D | hswish-scalar-x4.c | 40 float vacc0 = vx0 + vthree; in xnn_f32_hswish_ukernel__scalar_x4() local
49 vacc0 = math_max_f32(vacc0, vzero); in xnn_f32_hswish_ukernel__scalar_x4()
54 vacc0 = math_min_f32(vacc0, vsix); in xnn_f32_hswish_ukernel__scalar_x4()
59 vacc0 *= vx0; in xnn_f32_hswish_ukernel__scalar_x4()
64 y[0] = vacc0; in xnn_f32_hswish_ukernel__scalar_x4()
|
| /external/XNNPACK/src/f32-clamp/gen/ |
| D | wasm-x2.c | 32 float vacc0 = x[0]; in xnn_f32_clamp_ukernel__wasm_x2() local
36 vacc0 = __builtin_wasm_max_f32(vacc0, vy_min); in xnn_f32_clamp_ukernel__wasm_x2()
39 vacc0 = __builtin_wasm_min_f32(vacc0, vy_max); in xnn_f32_clamp_ukernel__wasm_x2()
42 y[0] = vacc0; in xnn_f32_clamp_ukernel__wasm_x2()
|
| D | scalar-x2.c | 32 float vacc0 = x[0]; in xnn_f32_clamp_ukernel__scalar_x2() local
36 vacc0 = math_max_f32(vacc0, vy_min); in xnn_f32_clamp_ukernel__scalar_x2()
39 vacc0 = math_min_f32(vacc0, vy_max); in xnn_f32_clamp_ukernel__scalar_x2()
42 y[0] = vacc0; in xnn_f32_clamp_ukernel__scalar_x2()
|
| D | scalar-x4.c | 32 float vacc0 = x[0]; in xnn_f32_clamp_ukernel__scalar_x4() local
38 vacc0 = math_max_f32(vacc0, vy_min); in xnn_f32_clamp_ukernel__scalar_x4()
43 vacc0 = math_min_f32(vacc0, vy_max); in xnn_f32_clamp_ukernel__scalar_x4()
48 y[0] = vacc0; in xnn_f32_clamp_ukernel__scalar_x4()
|
| D | wasm-x4.c | 32 float vacc0 = x[0]; in xnn_f32_clamp_ukernel__wasm_x4() local
38 vacc0 = __builtin_wasm_max_f32(vacc0, vy_min); in xnn_f32_clamp_ukernel__wasm_x4()
43 vacc0 = __builtin_wasm_min_f32(vacc0, vy_max); in xnn_f32_clamp_ukernel__wasm_x4()
48 y[0] = vacc0; in xnn_f32_clamp_ukernel__wasm_x4()
|