| /external/XNNPACK/src/u8-clamp/ |
| D | scalar-x4.c | 26 uint8_t vt3 = x[3]; in xnn_u8_clamp_ukernel__scalar_x4() local
32 vt3 = XNN_UNPREDICTABLE(vt3 < voutput_min) ? voutput_min : vt3; in xnn_u8_clamp_ukernel__scalar_x4()
37 vt3 = XNN_UNPREDICTABLE(vt3 > voutput_max) ? voutput_max : vt3; in xnn_u8_clamp_ukernel__scalar_x4()
42 y[3] = vt3; in xnn_u8_clamp_ukernel__scalar_x4()
|
| /external/XNNPACK/src/f32-velu/gen/ |
| D | velu-scalar-rr2-p6-x4.c | 72 float vt3 = vn3 * vminus_ln2_hi + vz3; in xnn_f32_velu_ukernel__scalar_rr2_p6_x4() local
77 vt3 = vn3 * vminus_ln2_lo + vt3; in xnn_f32_velu_ukernel__scalar_rr2_p6_x4()
93 vt3 = 0.0f; in xnn_f32_velu_ukernel__scalar_rr2_p6_x4()
99 float vp3 = vc6 * vt3 + vc5; in xnn_f32_velu_ukernel__scalar_rr2_p6_x4()
104 vp3 = vp3 * vt3 + vc4; in xnn_f32_velu_ukernel__scalar_rr2_p6_x4()
109 vp3 = vp3 * vt3 + vc3; in xnn_f32_velu_ukernel__scalar_rr2_p6_x4()
114 vp3 = vp3 * vt3 + vc2; in xnn_f32_velu_ukernel__scalar_rr2_p6_x4()
119 vp3 *= vt3; in xnn_f32_velu_ukernel__scalar_rr2_p6_x4()
127 vt3 *= vs3; in xnn_f32_velu_ukernel__scalar_rr2_p6_x4()
133 vp3 = vp3 * vt3 + vt3; in xnn_f32_velu_ukernel__scalar_rr2_p6_x4()
|
| D | velu-wasm-rr2-p6-x4.c | 72 float vt3 = vn3 * vminus_ln2_hi + vz3; in xnn_f32_velu_ukernel__wasm_rr2_p6_x4() local
77 vt3 = vn3 * vminus_ln2_lo + vt3; in xnn_f32_velu_ukernel__wasm_rr2_p6_x4()
83 float vp3 = vc6 * vt3 + vc5; in xnn_f32_velu_ukernel__wasm_rr2_p6_x4()
88 vp3 = vp3 * vt3 + vc4; in xnn_f32_velu_ukernel__wasm_rr2_p6_x4()
93 vp3 = vp3 * vt3 + vc3; in xnn_f32_velu_ukernel__wasm_rr2_p6_x4()
98 vp3 = vp3 * vt3 + vc2; in xnn_f32_velu_ukernel__wasm_rr2_p6_x4()
103 vp3 *= vt3; in xnn_f32_velu_ukernel__wasm_rr2_p6_x4()
111 vt3 *= vs3; in xnn_f32_velu_ukernel__wasm_rr2_p6_x4()
117 vp3 = vp3 * vt3 + vt3; in xnn_f32_velu_ukernel__wasm_rr2_p6_x4()
|
| D | velu-scalar-rr2-p6-x5.c | 77 float vt3 = vn3 * vminus_ln2_hi + vz3; in xnn_f32_velu_ukernel__scalar_rr2_p6_x5() local
83 vt3 = vn3 * vminus_ln2_lo + vt3; in xnn_f32_velu_ukernel__scalar_rr2_p6_x5()
100 vt3 = 0.0f; in xnn_f32_velu_ukernel__scalar_rr2_p6_x5()
110 float vp3 = vc6 * vt3 + vc5; in xnn_f32_velu_ukernel__scalar_rr2_p6_x5()
116 vp3 = vp3 * vt3 + vc4; in xnn_f32_velu_ukernel__scalar_rr2_p6_x5()
122 vp3 = vp3 * vt3 + vc3; in xnn_f32_velu_ukernel__scalar_rr2_p6_x5()
128 vp3 = vp3 * vt3 + vc2; in xnn_f32_velu_ukernel__scalar_rr2_p6_x5()
134 vp3 *= vt3; in xnn_f32_velu_ukernel__scalar_rr2_p6_x5()
143 vt3 *= vs3; in xnn_f32_velu_ukernel__scalar_rr2_p6_x5()
151 vp3 = vp3 * vt3 + vt3; in xnn_f32_velu_ukernel__scalar_rr2_p6_x5()
|
| D | velu-wasm-rr2-p6-x5.c | 77 float vt3 = vn3 * vminus_ln2_hi + vz3; in xnn_f32_velu_ukernel__wasm_rr2_p6_x5() local
83 vt3 = vn3 * vminus_ln2_lo + vt3; in xnn_f32_velu_ukernel__wasm_rr2_p6_x5()
90 float vp3 = vc6 * vt3 + vc5; in xnn_f32_velu_ukernel__wasm_rr2_p6_x5()
96 vp3 = vp3 * vt3 + vc4; in xnn_f32_velu_ukernel__wasm_rr2_p6_x5()
102 vp3 = vp3 * vt3 + vc3; in xnn_f32_velu_ukernel__wasm_rr2_p6_x5()
108 vp3 = vp3 * vt3 + vc2; in xnn_f32_velu_ukernel__wasm_rr2_p6_x5()
114 vp3 *= vt3; in xnn_f32_velu_ukernel__wasm_rr2_p6_x5()
123 vt3 *= vs3; in xnn_f32_velu_ukernel__wasm_rr2_p6_x5()
131 vp3 = vp3 * vt3 + vt3; in xnn_f32_velu_ukernel__wasm_rr2_p6_x5()
|
| D | velu-scalar-rr2-p6-x6.c | 82 float vt3 = vn3 * vminus_ln2_hi + vz3; in xnn_f32_velu_ukernel__scalar_rr2_p6_x6() local
89 vt3 = vn3 * vminus_ln2_lo + vt3; in xnn_f32_velu_ukernel__scalar_rr2_p6_x6()
107 vt3 = 0.0f; in xnn_f32_velu_ukernel__scalar_rr2_p6_x6()
121 float vp3 = vc6 * vt3 + vc5; in xnn_f32_velu_ukernel__scalar_rr2_p6_x6()
128 vp3 = vp3 * vt3 + vc4; in xnn_f32_velu_ukernel__scalar_rr2_p6_x6()
135 vp3 = vp3 * vt3 + vc3; in xnn_f32_velu_ukernel__scalar_rr2_p6_x6()
142 vp3 = vp3 * vt3 + vc2; in xnn_f32_velu_ukernel__scalar_rr2_p6_x6()
149 vp3 *= vt3; in xnn_f32_velu_ukernel__scalar_rr2_p6_x6()
159 vt3 *= vs3; in xnn_f32_velu_ukernel__scalar_rr2_p6_x6()
169 vp3 = vp3 * vt3 + vt3; in xnn_f32_velu_ukernel__scalar_rr2_p6_x6()
|
| D | velu-wasm-rr2-p6-x6.c | 82 float vt3 = vn3 * vminus_ln2_hi + vz3; in xnn_f32_velu_ukernel__wasm_rr2_p6_x6() local
89 vt3 = vn3 * vminus_ln2_lo + vt3; in xnn_f32_velu_ukernel__wasm_rr2_p6_x6()
97 float vp3 = vc6 * vt3 + vc5; in xnn_f32_velu_ukernel__wasm_rr2_p6_x6()
104 vp3 = vp3 * vt3 + vc4; in xnn_f32_velu_ukernel__wasm_rr2_p6_x6()
111 vp3 = vp3 * vt3 + vc3; in xnn_f32_velu_ukernel__wasm_rr2_p6_x6()
118 vp3 = vp3 * vt3 + vc2; in xnn_f32_velu_ukernel__wasm_rr2_p6_x6()
125 vp3 *= vt3; in xnn_f32_velu_ukernel__wasm_rr2_p6_x6()
135 vt3 *= vs3; in xnn_f32_velu_ukernel__wasm_rr2_p6_x6()
145 vp3 = vp3 * vt3 + vt3; in xnn_f32_velu_ukernel__wasm_rr2_p6_x6()
|
| D | velu-scalar-rr2-lut16-p3-x4.c | 79 float vt3 = vn3 * vminus_ln2_hi + vz3; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x4() local
97 vt3 = vn3 * vminus_ln2_lo + vt3; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x4()
100 vt3 = 0.0f; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x4()
106 float vp3 = vc3 * vt3 + vc2; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x4()
111 vp3 *= vt3; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x4()
119 vt3 *= vs3; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x4()
125 vp3 = vp3 * vt3 + vt3; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x4()
|
| D | velu-scalar-rr2-lut16-p3-x5.c | 85 float vt3 = vn3 * vminus_ln2_hi + vz3; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x5() local
105 vt3 = vn3 * vminus_ln2_lo + vt3; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x5()
108 vt3 = 0.0f; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x5()
119 float vp3 = vc3 * vt3 + vc2; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x5()
125 vp3 *= vt3; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x5()
134 vt3 *= vs3; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x5()
142 vp3 = vp3 * vt3 + vt3; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x5()
|
| D | velu-wasm-rr2-lut16-p3-x4.c | 79 float vt3 = vn3 * vminus_ln2_hi + vz3; in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x4() local
85 vt3 = vn3 * vminus_ln2_lo + vt3; in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x4()
90 float vp3 = vc3 * vt3 + vc2; in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x4()
95 vp3 *= vt3; in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x4()
103 vt3 *= vs3; in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x4()
109 vp3 = vp3 * vt3 + vt3; in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x4()
|
| D | velu-scalar-rr2-lut16-p3-x6.c | 91 float vt3 = vn3 * vminus_ln2_hi + vz3; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x6() local
113 vt3 = vn3 * vminus_ln2_lo + vt3; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x6()
116 vt3 = 0.0f; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x6()
132 float vp3 = vc3 * vt3 + vc2; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x6()
139 vp3 *= vt3; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x6()
149 vt3 *= vs3; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x6()
159 vp3 = vp3 * vt3 + vt3; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x6()
|
| D | velu-wasm-rr2-lut16-p3-x5.c | 85 float vt3 = vn3 * vminus_ln2_hi + vz3; in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x5() local
93 vt3 = vn3 * vminus_ln2_lo + vt3; in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x5()
99 float vp3 = vc3 * vt3 + vc2; in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x5()
105 vp3 *= vt3; in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x5()
114 vt3 *= vs3; in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x5()
122 vp3 = vp3 * vt3 + vt3; in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x5()
|
| D | velu-avx512f-rr1-p6-x64.c | 71 __m512 vt3 = _mm512_fmadd_ps(vn3, vminus_ln2, vz3); in xnn_f32_velu_ukernel__avx512f_rr1_p6_x64() local
76 __m512 vp3 = _mm512_fmadd_ps(vc6, vt3, vc5); in xnn_f32_velu_ukernel__avx512f_rr1_p6_x64()
81 vp3 = _mm512_fmadd_ps(vp3, vt3, vc4); in xnn_f32_velu_ukernel__avx512f_rr1_p6_x64()
86 vp3 = _mm512_fmadd_ps(vp3, vt3, vc3); in xnn_f32_velu_ukernel__avx512f_rr1_p6_x64()
91 vp3 = _mm512_fmadd_ps(vp3, vt3, vc2); in xnn_f32_velu_ukernel__avx512f_rr1_p6_x64()
99 vp3 = _mm512_mul_ps(vp3, vt3); in xnn_f32_velu_ukernel__avx512f_rr1_p6_x64()
100 vt3 = _mm512_mul_ps(vt3, vs3); in xnn_f32_velu_ukernel__avx512f_rr1_p6_x64()
110 vp3 = _mm512_fmadd_ps(vp3, vt3, vt3); in xnn_f32_velu_ukernel__avx512f_rr1_p6_x64()
|
| D | velu-avx2-rr1-p6-x32.c | 71 __m256 vt3 = _mm256_fmadd_ps(vn3, vminus_ln2, vz3); in xnn_f32_velu_ukernel__avx2_rr1_p6_x32() local
76 __m256 vp3 = _mm256_fmadd_ps(vc6, vt3, vc5); in xnn_f32_velu_ukernel__avx2_rr1_p6_x32()
81 vp3 = _mm256_fmadd_ps(vp3, vt3, vc4); in xnn_f32_velu_ukernel__avx2_rr1_p6_x32()
86 vp3 = _mm256_fmadd_ps(vp3, vt3, vc3); in xnn_f32_velu_ukernel__avx2_rr1_p6_x32()
91 vp3 = _mm256_fmadd_ps(vp3, vt3, vc2); in xnn_f32_velu_ukernel__avx2_rr1_p6_x32()
99 vp3 = _mm256_mul_ps(vp3, vt3); in xnn_f32_velu_ukernel__avx2_rr1_p6_x32()
100 vt3 = _mm256_mul_ps(vt3, vs3); in xnn_f32_velu_ukernel__avx2_rr1_p6_x32()
109 vp3 = _mm256_fmadd_ps(vp3, vt3, vt3); in xnn_f32_velu_ukernel__avx2_rr1_p6_x32()
|
| D | velu-avx-rr2-p6-x32.c | 80 __m256 vt3 = _mm256_add_ps(_mm256_mul_ps(vn3, vminus_ln2_hi), vz3); in xnn_f32_velu_ukernel__avx_rr2_p6_x32() local
86 vt3 = _mm256_add_ps(_mm256_mul_ps(vn3, vminus_ln2_lo), vt3); in xnn_f32_velu_ukernel__avx_rr2_p6_x32()
91 __m256 vp3 = _mm256_add_ps(_mm256_mul_ps(vc6, vt3), vc5); in xnn_f32_velu_ukernel__avx_rr2_p6_x32()
96 vp3 = _mm256_add_ps(_mm256_mul_ps(vp3, vt3), vc4); in xnn_f32_velu_ukernel__avx_rr2_p6_x32()
101 vp3 = _mm256_add_ps(_mm256_mul_ps(vp3, vt3), vc3); in xnn_f32_velu_ukernel__avx_rr2_p6_x32()
106 vp3 = _mm256_add_ps(_mm256_mul_ps(vp3, vt3), vc2); in xnn_f32_velu_ukernel__avx_rr2_p6_x32()
111 vp3 = _mm256_mul_ps(vp3, vt3); in xnn_f32_velu_ukernel__avx_rr2_p6_x32()
119 vt3 = _mm256_mul_ps(vt3, vs3); in xnn_f32_velu_ukernel__avx_rr2_p6_x32()
125 vp3 = _mm256_add_ps(_mm256_mul_ps(vp3, vt3), vt3); in xnn_f32_velu_ukernel__avx_rr2_p6_x32()
|
| /external/XNNPACK/src/f32-raddstoreexpminusmax/gen/ |
| D | scalar-p5-x4.c | 84 float vt3 = vn3 * vminus_ln2_hi + vx3; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4() local
89 vt3 = vn3 * vminus_ln2_lo + vt3; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4()
95 float vp3 = vc5 * vt3 + vc4; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4()
100 vp3 = vp3 * vt3 + vc3; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4()
105 vp3 = vp3 * vt3 + vc2; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4()
110 vp3 = vp3 * vt3 + vc1; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4()
119 vt3 *= vs3; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4()
124 float vf3 = vt3 * vp3 + vs3; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4()
|
| D | scalar-p5-x4-acc2.c | 85 float vt3 = vn3 * vminus_ln2_hi + vx3; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4_acc2() local
90 vt3 = vn3 * vminus_ln2_lo + vt3; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4_acc2()
96 float vp3 = vc5 * vt3 + vc4; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4_acc2()
101 vp3 = vp3 * vt3 + vc3; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4_acc2()
106 vp3 = vp3 * vt3 + vc2; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4_acc2()
111 vp3 = vp3 * vt3 + vc1; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4_acc2()
120 vt3 *= vs3; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4_acc2()
125 float vf3 = vt3 * vp3 + vs3; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4_acc2()
|
| D | scalar-p5-x4-acc4.c | 87 float vt3 = vn3 * vminus_ln2_hi + vx3; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4_acc4() local
92 vt3 = vn3 * vminus_ln2_lo + vt3; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4_acc4()
98 float vp3 = vc5 * vt3 + vc4; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4_acc4()
103 vp3 = vp3 * vt3 + vc3; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4_acc4()
108 vp3 = vp3 * vt3 + vc2; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4_acc4()
113 vp3 = vp3 * vt3 + vc1; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4_acc4()
122 vt3 *= vs3; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4_acc4()
127 float vf3 = vt3 * vp3 + vs3; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4_acc4()
|
| D | scalar-lut64-p2-x4-acc2.c | 107 float vt3 = vn3 * vminus_ln2_o64_hi + vx3; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_lut64_p2_x4_acc2() local
112 vt3 = vn3 * vminus_ln2_o64_lo + vt3; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_lut64_p2_x4_acc2()
118 float vp3 = vt3 * vc2; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_lut64_p2_x4_acc2()
123 vp3 = vp3 * vt3 + vt3; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_lut64_p2_x4_acc2()
|
| D | scalar-lut64-p2-x4-acc4.c | 109 float vt3 = vn3 * vminus_ln2_o64_hi + vx3; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_lut64_p2_x4_acc4() local
114 vt3 = vn3 * vminus_ln2_o64_lo + vt3; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_lut64_p2_x4_acc4()
120 float vp3 = vt3 * vc2; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_lut64_p2_x4_acc4()
125 vp3 = vp3 * vt3 + vt3; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_lut64_p2_x4_acc4()
|
| /external/XNNPACK/src/f32-sigmoid/gen/ |
| D | scalar-p5-div-x4.c | 69 float vt3 = vn3 * vln2_hi + vz3; in xnn_f32_sigmoid_ukernel__scalar_p5_div_x4() local
74 vt3 = vn3 * vln2_lo + vt3; in xnn_f32_sigmoid_ukernel__scalar_p5_div_x4()
79 float vp3 = vt3 * vc5 + vc4; in xnn_f32_sigmoid_ukernel__scalar_p5_div_x4()
84 vp3 = vt3 * vp3 + vc3; in xnn_f32_sigmoid_ukernel__scalar_p5_div_x4()
89 vp3 = vt3 * vp3 + vc2; in xnn_f32_sigmoid_ukernel__scalar_p5_div_x4()
94 vp3 = vt3 * vp3 + vc1; in xnn_f32_sigmoid_ukernel__scalar_p5_div_x4()
99 vt3 *= vs3; in xnn_f32_sigmoid_ukernel__scalar_p5_div_x4()
104 const float ve3 = vt3 * vp3 + vs3; in xnn_f32_sigmoid_ukernel__scalar_p5_div_x4()
|
| /external/XNNPACK/src/f32-vscaleexpminusmax/gen/ |
| D | avx2-p5-x32.c | 83 __m256 vt3 = _mm256_fmadd_ps(vn3, vminus_ln2_hi, vx3); in xnn_f32_vscaleexpminusmax_ukernel__avx2_p5_x32() local
88 vt3 = _mm256_fmadd_ps(vn3, vminus_ln2_lo, vt3); in xnn_f32_vscaleexpminusmax_ukernel__avx2_p5_x32()
94 __m256 vp3 = _mm256_fmadd_ps(vc5, vt3, vc4); in xnn_f32_vscaleexpminusmax_ukernel__avx2_p5_x32()
99 vp3 = _mm256_fmadd_ps(vp3, vt3, vc3); in xnn_f32_vscaleexpminusmax_ukernel__avx2_p5_x32()
104 vp3 = _mm256_fmadd_ps(vp3, vt3, vc2); in xnn_f32_vscaleexpminusmax_ukernel__avx2_p5_x32()
109 vp3 = _mm256_fmadd_ps(vp3, vt3, vc1); in xnn_f32_vscaleexpminusmax_ukernel__avx2_p5_x32()
118 vt3 = _mm256_mul_ps(vt3, vs3); in xnn_f32_vscaleexpminusmax_ukernel__avx2_p5_x32()
123 __m256 vf3 = _mm256_fmadd_ps(vt3, vp3, vs3); in xnn_f32_vscaleexpminusmax_ukernel__avx2_p5_x32()
|
| D | avx2-p5-x40.c | 88 __m256 vt3 = _mm256_fmadd_ps(vn3, vminus_ln2_hi, vx3); in xnn_f32_vscaleexpminusmax_ukernel__avx2_p5_x40() local
94 vt3 = _mm256_fmadd_ps(vn3, vminus_ln2_lo, vt3); in xnn_f32_vscaleexpminusmax_ukernel__avx2_p5_x40()
101 __m256 vp3 = _mm256_fmadd_ps(vc5, vt3, vc4); in xnn_f32_vscaleexpminusmax_ukernel__avx2_p5_x40()
107 vp3 = _mm256_fmadd_ps(vp3, vt3, vc3); in xnn_f32_vscaleexpminusmax_ukernel__avx2_p5_x40()
113 vp3 = _mm256_fmadd_ps(vp3, vt3, vc2); in xnn_f32_vscaleexpminusmax_ukernel__avx2_p5_x40()
119 vp3 = _mm256_fmadd_ps(vp3, vt3, vc1); in xnn_f32_vscaleexpminusmax_ukernel__avx2_p5_x40()
129 vt3 = _mm256_mul_ps(vt3, vs3); in xnn_f32_vscaleexpminusmax_ukernel__avx2_p5_x40()
135 __m256 vf3 = _mm256_fmadd_ps(vt3, vp3, vs3); in xnn_f32_vscaleexpminusmax_ukernel__avx2_p5_x40()
|
| D | avx512f-p5-scalef-x64.c | 66 __m512 vt3 = _mm512_fmadd_ps(vn3, vminus_ln2_hi, vx3); in xnn_f32_vscaleexpminusmax_ukernel__avx512f_p5_scalef_x64() local
71 vt3 = _mm512_fmadd_ps(vn3, vminus_ln2_lo, vt3); in xnn_f32_vscaleexpminusmax_ukernel__avx512f_p5_scalef_x64()
77 __m512 vp3 = _mm512_fmadd_ps(vc5, vt3, vc4); in xnn_f32_vscaleexpminusmax_ukernel__avx512f_p5_scalef_x64()
82 vp3 = _mm512_fmadd_ps(vp3, vt3, vc3); in xnn_f32_vscaleexpminusmax_ukernel__avx512f_p5_scalef_x64()
87 vp3 = _mm512_fmadd_ps(vp3, vt3, vc2); in xnn_f32_vscaleexpminusmax_ukernel__avx512f_p5_scalef_x64()
92 vp3 = _mm512_fmadd_ps(vp3, vt3, vc1); in xnn_f32_vscaleexpminusmax_ukernel__avx512f_p5_scalef_x64()
97 vp3 = _mm512_fmadd_ps(vp3, vt3, vc0); in xnn_f32_vscaleexpminusmax_ukernel__avx512f_p5_scalef_x64()
|
| /external/XNNPACK/src/f32-vscaleextexp/gen/ |
| D | avx512f-p5-scalef-x64.c | 61 __m512 vt3 = _mm512_fmadd_ps(vn3, vminus_ln2_hi, vx3); in xnn_f32_vscaleextexp_ukernel__avx512f_p5_scalef_x64() local
66 vt3 = _mm512_fmadd_ps(vn3, vminus_ln2_lo, vt3); in xnn_f32_vscaleextexp_ukernel__avx512f_p5_scalef_x64()
72 __m512 vp3 = _mm512_fmadd_ps(vc5, vt3, vc4); in xnn_f32_vscaleextexp_ukernel__avx512f_p5_scalef_x64()
77 vp3 = _mm512_fmadd_ps(vp3, vt3, vc3); in xnn_f32_vscaleextexp_ukernel__avx512f_p5_scalef_x64()
82 vp3 = _mm512_fmadd_ps(vp3, vt3, vc2); in xnn_f32_vscaleextexp_ukernel__avx512f_p5_scalef_x64()
87 vp3 = _mm512_fmadd_ps(vp3, vt3, vc1); in xnn_f32_vscaleextexp_ukernel__avx512f_p5_scalef_x64()
92 vp3 = _mm512_fmadd_ps(vp3, vt3, vc0); in xnn_f32_vscaleextexp_ukernel__avx512f_p5_scalef_x64()
|