| /external/XNNPACK/src/f32-velu/gen/ |
| D | velu-wasm-rr2-p6-x2.c | 66 float vp0 = vc6 * vt0 + vc5; in xnn_f32_velu_ukernel__wasm_rr2_p6_x2() local
69 vp0 = vp0 * vt0 + vc4; in xnn_f32_velu_ukernel__wasm_rr2_p6_x2()
72 vp0 = vp0 * vt0 + vc3; in xnn_f32_velu_ukernel__wasm_rr2_p6_x2()
75 vp0 = vp0 * vt0 + vc2; in xnn_f32_velu_ukernel__wasm_rr2_p6_x2()
78 vp0 *= vt0; in xnn_f32_velu_ukernel__wasm_rr2_p6_x2()
86 vp0 = vp0 * vt0 + vt0; in xnn_f32_velu_ukernel__wasm_rr2_p6_x2()
89 const float ve0 = (vp0 + vs0) * valpha; in xnn_f32_velu_ukernel__wasm_rr2_p6_x2()
|
| D | velu-scalar-rr2-p6-x2.c | 74 float vp0 = vc6 * vt0 + vc5; in xnn_f32_velu_ukernel__scalar_rr2_p6_x2() local
77 vp0 = vp0 * vt0 + vc4; in xnn_f32_velu_ukernel__scalar_rr2_p6_x2()
80 vp0 = vp0 * vt0 + vc3; in xnn_f32_velu_ukernel__scalar_rr2_p6_x2()
83 vp0 = vp0 * vt0 + vc2; in xnn_f32_velu_ukernel__scalar_rr2_p6_x2()
86 vp0 *= vt0; in xnn_f32_velu_ukernel__scalar_rr2_p6_x2()
94 vp0 = vp0 * vt0 + vt0; in xnn_f32_velu_ukernel__scalar_rr2_p6_x2()
97 const float ve0 = (vp0 + vs0) * valpha; in xnn_f32_velu_ukernel__scalar_rr2_p6_x2()
|
| D | velu-scalar-rr2-p6-x3.c | 85 float vp0 = vc6 * vt0 + vc5; in xnn_f32_velu_ukernel__scalar_rr2_p6_x3() local
89 vp0 = vp0 * vt0 + vc4; in xnn_f32_velu_ukernel__scalar_rr2_p6_x3()
93 vp0 = vp0 * vt0 + vc3; in xnn_f32_velu_ukernel__scalar_rr2_p6_x3()
97 vp0 = vp0 * vt0 + vc2; in xnn_f32_velu_ukernel__scalar_rr2_p6_x3()
101 vp0 *= vt0; in xnn_f32_velu_ukernel__scalar_rr2_p6_x3()
112 vp0 = vp0 * vt0 + vt0; in xnn_f32_velu_ukernel__scalar_rr2_p6_x3()
116 const float ve0 = (vp0 + vs0) * valpha; in xnn_f32_velu_ukernel__scalar_rr2_p6_x3()
|
| D | velu-wasm-rr2-p6-x3.c | 73 float vp0 = vc6 * vt0 + vc5; in xnn_f32_velu_ukernel__wasm_rr2_p6_x3() local
77 vp0 = vp0 * vt0 + vc4; in xnn_f32_velu_ukernel__wasm_rr2_p6_x3()
81 vp0 = vp0 * vt0 + vc3; in xnn_f32_velu_ukernel__wasm_rr2_p6_x3()
85 vp0 = vp0 * vt0 + vc2; in xnn_f32_velu_ukernel__wasm_rr2_p6_x3()
89 vp0 *= vt0; in xnn_f32_velu_ukernel__wasm_rr2_p6_x3()
100 vp0 = vp0 * vt0 + vt0; in xnn_f32_velu_ukernel__wasm_rr2_p6_x3()
104 const float ve0 = (vp0 + vs0) * valpha; in xnn_f32_velu_ukernel__wasm_rr2_p6_x3()
|
| D | velu-wasm-rr2-p6-x4.c | 80 float vp0 = vc6 * vt0 + vc5; in xnn_f32_velu_ukernel__wasm_rr2_p6_x4() local
85 vp0 = vp0 * vt0 + vc4; in xnn_f32_velu_ukernel__wasm_rr2_p6_x4()
90 vp0 = vp0 * vt0 + vc3; in xnn_f32_velu_ukernel__wasm_rr2_p6_x4()
95 vp0 = vp0 * vt0 + vc2; in xnn_f32_velu_ukernel__wasm_rr2_p6_x4()
100 vp0 *= vt0; in xnn_f32_velu_ukernel__wasm_rr2_p6_x4()
114 vp0 = vp0 * vt0 + vt0; in xnn_f32_velu_ukernel__wasm_rr2_p6_x4()
119 const float ve0 = (vp0 + vs0) * valpha; in xnn_f32_velu_ukernel__wasm_rr2_p6_x4()
|
| D | velu-scalar-rr2-p6-x4.c | 96 float vp0 = vc6 * vt0 + vc5; in xnn_f32_velu_ukernel__scalar_rr2_p6_x4() local
101 vp0 = vp0 * vt0 + vc4; in xnn_f32_velu_ukernel__scalar_rr2_p6_x4()
106 vp0 = vp0 * vt0 + vc3; in xnn_f32_velu_ukernel__scalar_rr2_p6_x4()
111 vp0 = vp0 * vt0 + vc2; in xnn_f32_velu_ukernel__scalar_rr2_p6_x4()
116 vp0 *= vt0; in xnn_f32_velu_ukernel__scalar_rr2_p6_x4()
130 vp0 = vp0 * vt0 + vt0; in xnn_f32_velu_ukernel__scalar_rr2_p6_x4()
135 const float ve0 = (vp0 + vs0) * valpha; in xnn_f32_velu_ukernel__scalar_rr2_p6_x4()
|
| D | velu-avx2-rr1-p6-x16.c | 61 __m256 vp0 = _mm256_fmadd_ps(vc6, vt0, vc5); in xnn_f32_velu_ukernel__avx2_rr1_p6_x16() local
64 vp0 = _mm256_fmadd_ps(vp0, vt0, vc4); in xnn_f32_velu_ukernel__avx2_rr1_p6_x16()
67 vp0 = _mm256_fmadd_ps(vp0, vt0, vc3); in xnn_f32_velu_ukernel__avx2_rr1_p6_x16()
70 vp0 = _mm256_fmadd_ps(vp0, vt0, vc2); in xnn_f32_velu_ukernel__avx2_rr1_p6_x16()
73 vp0 = _mm256_mul_ps(vp0, vt0); in xnn_f32_velu_ukernel__avx2_rr1_p6_x16()
79 vp0 = _mm256_fmadd_ps(vp0, vt0, vt0); in xnn_f32_velu_ukernel__avx2_rr1_p6_x16()
83 const __m256 ve0 = _mm256_fmadd_ps(vp0, valpha, vs0); in xnn_f32_velu_ukernel__avx2_rr1_p6_x16()
|
| D | velu-avx512f-rr1-p6-x32.c | 61 __m512 vp0 = _mm512_fmadd_ps(vc6, vt0, vc5); in xnn_f32_velu_ukernel__avx512f_rr1_p6_x32() local
64 vp0 = _mm512_fmadd_ps(vp0, vt0, vc4); in xnn_f32_velu_ukernel__avx512f_rr1_p6_x32()
67 vp0 = _mm512_fmadd_ps(vp0, vt0, vc3); in xnn_f32_velu_ukernel__avx512f_rr1_p6_x32()
70 vp0 = _mm512_fmadd_ps(vp0, vt0, vc2); in xnn_f32_velu_ukernel__avx512f_rr1_p6_x32()
73 vp0 = _mm512_mul_ps(vp0, vt0); in xnn_f32_velu_ukernel__avx512f_rr1_p6_x32()
81 vp0 = _mm512_fmadd_ps(vp0, vt0, vt0); in xnn_f32_velu_ukernel__avx512f_rr1_p6_x32()
85 __m512 vy0 = _mm512_fmadd_ps(vp0, valpha, vs0); in xnn_f32_velu_ukernel__avx512f_rr1_p6_x32()
|
| D | velu-wasm-rr2-p6-x5.c | 87 float vp0 = vc6 * vt0 + vc5; in xnn_f32_velu_ukernel__wasm_rr2_p6_x5() local
93 vp0 = vp0 * vt0 + vc4; in xnn_f32_velu_ukernel__wasm_rr2_p6_x5()
99 vp0 = vp0 * vt0 + vc3; in xnn_f32_velu_ukernel__wasm_rr2_p6_x5()
105 vp0 = vp0 * vt0 + vc2; in xnn_f32_velu_ukernel__wasm_rr2_p6_x5()
111 vp0 *= vt0; in xnn_f32_velu_ukernel__wasm_rr2_p6_x5()
128 vp0 = vp0 * vt0 + vt0; in xnn_f32_velu_ukernel__wasm_rr2_p6_x5()
134 const float ve0 = (vp0 + vs0) * valpha; in xnn_f32_velu_ukernel__wasm_rr2_p6_x5()
|
| D | velu-scalar-rr2-p6-x5.c | 107 float vp0 = vc6 * vt0 + vc5; in xnn_f32_velu_ukernel__scalar_rr2_p6_x5() local
113 vp0 = vp0 * vt0 + vc4; in xnn_f32_velu_ukernel__scalar_rr2_p6_x5()
119 vp0 = vp0 * vt0 + vc3; in xnn_f32_velu_ukernel__scalar_rr2_p6_x5()
125 vp0 = vp0 * vt0 + vc2; in xnn_f32_velu_ukernel__scalar_rr2_p6_x5()
131 vp0 *= vt0; in xnn_f32_velu_ukernel__scalar_rr2_p6_x5()
148 vp0 = vp0 * vt0 + vt0; in xnn_f32_velu_ukernel__scalar_rr2_p6_x5()
154 const float ve0 = (vp0 + vs0) * valpha; in xnn_f32_velu_ukernel__scalar_rr2_p6_x5()
|
| /external/XNNPACK/src/f32-vscaleextexp/gen/ |
| D | avx512f-p5-scalef-x16.c | 57 __m512 vp0 = _mm512_fmadd_ps(vc5, vt0, vc4); in xnn_f32_vscaleextexp_ukernel__avx512f_p5_scalef_x16() local
59 vp0 = _mm512_fmadd_ps(vp0, vt0, vc3); in xnn_f32_vscaleextexp_ukernel__avx512f_p5_scalef_x16()
61 vp0 = _mm512_fmadd_ps(vp0, vt0, vc2); in xnn_f32_vscaleextexp_ukernel__avx512f_p5_scalef_x16()
63 vp0 = _mm512_fmadd_ps(vp0, vt0, vc1); in xnn_f32_vscaleextexp_ukernel__avx512f_p5_scalef_x16()
65 vp0 = _mm512_fmadd_ps(vp0, vt0, vc0); in xnn_f32_vscaleextexp_ukernel__avx512f_p5_scalef_x16()
73 __m512 vf0 = _mm512_mul_ps(vp0, vscalev); in xnn_f32_vscaleextexp_ukernel__avx512f_p5_scalef_x16()
|
| D | avx2-p5-x8.c | 63 __m256 vp0 = _mm256_fmadd_ps(vc5, vt0, vc4); in xnn_f32_vscaleextexp_ukernel__avx2_p5_x8() local
65 vp0 = _mm256_fmadd_ps(vp0, vt0, vc3); in xnn_f32_vscaleextexp_ukernel__avx2_p5_x8()
67 vp0 = _mm256_fmadd_ps(vp0, vt0, vc2); in xnn_f32_vscaleextexp_ukernel__avx2_p5_x8()
69 vp0 = _mm256_fmadd_ps(vp0, vt0, vc1); in xnn_f32_vscaleextexp_ukernel__avx2_p5_x8()
71 vp0 = _mm256_fmadd_ps(vp0, vt0, vc0); in xnn_f32_vscaleextexp_ukernel__avx2_p5_x8()
79 __m256 vf0 = _mm256_mul_ps(vp0, vscalev); in xnn_f32_vscaleextexp_ukernel__avx2_p5_x8()
|
| D | avx512f-p5-scalef-x32.c | 61 __m512 vp0 = _mm512_fmadd_ps(vc5, vt0, vc4); in xnn_f32_vscaleextexp_ukernel__avx512f_p5_scalef_x32() local
64 vp0 = _mm512_fmadd_ps(vp0, vt0, vc3); in xnn_f32_vscaleextexp_ukernel__avx512f_p5_scalef_x32()
67 vp0 = _mm512_fmadd_ps(vp0, vt0, vc2); in xnn_f32_vscaleextexp_ukernel__avx512f_p5_scalef_x32()
70 vp0 = _mm512_fmadd_ps(vp0, vt0, vc1); in xnn_f32_vscaleextexp_ukernel__avx512f_p5_scalef_x32()
73 vp0 = _mm512_fmadd_ps(vp0, vt0, vc0); in xnn_f32_vscaleextexp_ukernel__avx512f_p5_scalef_x32()
82 __m512 vf0 = _mm512_mul_ps(vp0, vscalev); in xnn_f32_vscaleextexp_ukernel__avx512f_p5_scalef_x32()
|
| D | avx512f-p5-scalef-x48.c | 65 __m512 vp0 = _mm512_fmadd_ps(vc5, vt0, vc4); in xnn_f32_vscaleextexp_ukernel__avx512f_p5_scalef_x48() local
69 vp0 = _mm512_fmadd_ps(vp0, vt0, vc3); in xnn_f32_vscaleextexp_ukernel__avx512f_p5_scalef_x48()
73 vp0 = _mm512_fmadd_ps(vp0, vt0, vc2); in xnn_f32_vscaleextexp_ukernel__avx512f_p5_scalef_x48()
77 vp0 = _mm512_fmadd_ps(vp0, vt0, vc1); in xnn_f32_vscaleextexp_ukernel__avx512f_p5_scalef_x48()
81 vp0 = _mm512_fmadd_ps(vp0, vt0, vc0); in xnn_f32_vscaleextexp_ukernel__avx512f_p5_scalef_x48()
91 __m512 vf0 = _mm512_mul_ps(vp0, vscalev); in xnn_f32_vscaleextexp_ukernel__avx512f_p5_scalef_x48()
|
| D | avx2-p5-x16.c | 67 __m256 vp0 = _mm256_fmadd_ps(vc5, vt0, vc4); in xnn_f32_vscaleextexp_ukernel__avx2_p5_x16() local
70 vp0 = _mm256_fmadd_ps(vp0, vt0, vc3); in xnn_f32_vscaleextexp_ukernel__avx2_p5_x16()
73 vp0 = _mm256_fmadd_ps(vp0, vt0, vc2); in xnn_f32_vscaleextexp_ukernel__avx2_p5_x16()
76 vp0 = _mm256_fmadd_ps(vp0, vt0, vc1); in xnn_f32_vscaleextexp_ukernel__avx2_p5_x16()
79 vp0 = _mm256_fmadd_ps(vp0, vt0, vc0); in xnn_f32_vscaleextexp_ukernel__avx2_p5_x16()
88 __m256 vf0 = _mm256_mul_ps(vp0, vscalev); in xnn_f32_vscaleextexp_ukernel__avx2_p5_x16()
|
| /external/XNNPACK/src/f32-vscaleexpminusmax/gen/ |
| D | avx512f-p5-scalef-x16.c | 59 __m512 vp0 = _mm512_fmadd_ps(vc5, vt0, vc4); in xnn_f32_vscaleexpminusmax_ukernel__avx512f_p5_scalef_x16() local
61 vp0 = _mm512_fmadd_ps(vp0, vt0, vc3); in xnn_f32_vscaleexpminusmax_ukernel__avx512f_p5_scalef_x16()
63 vp0 = _mm512_fmadd_ps(vp0, vt0, vc2); in xnn_f32_vscaleexpminusmax_ukernel__avx512f_p5_scalef_x16()
65 vp0 = _mm512_fmadd_ps(vp0, vt0, vc1); in xnn_f32_vscaleexpminusmax_ukernel__avx512f_p5_scalef_x16()
67 vp0 = _mm512_fmadd_ps(vp0, vt0, vc0); in xnn_f32_vscaleexpminusmax_ukernel__avx512f_p5_scalef_x16()
72 __m512 vf0 = _mm512_scalef_ps(vp0, vn0); in xnn_f32_vscaleexpminusmax_ukernel__avx512f_p5_scalef_x16()
|
| D | avx512f-p5-scalef-x32.c | 64 __m512 vp0 = _mm512_fmadd_ps(vc5, vt0, vc4); in xnn_f32_vscaleexpminusmax_ukernel__avx512f_p5_scalef_x32() local
67 vp0 = _mm512_fmadd_ps(vp0, vt0, vc3); in xnn_f32_vscaleexpminusmax_ukernel__avx512f_p5_scalef_x32()
70 vp0 = _mm512_fmadd_ps(vp0, vt0, vc2); in xnn_f32_vscaleexpminusmax_ukernel__avx512f_p5_scalef_x32()
73 vp0 = _mm512_fmadd_ps(vp0, vt0, vc1); in xnn_f32_vscaleexpminusmax_ukernel__avx512f_p5_scalef_x32()
76 vp0 = _mm512_fmadd_ps(vp0, vt0, vc0); in xnn_f32_vscaleexpminusmax_ukernel__avx512f_p5_scalef_x32()
82 __m512 vf0 = _mm512_scalef_ps(vp0, vn0); in xnn_f32_vscaleexpminusmax_ukernel__avx512f_p5_scalef_x32()
|
| D | avx512f-p5-scalef-x48.c | 69 __m512 vp0 = _mm512_fmadd_ps(vc5, vt0, vc4); in xnn_f32_vscaleexpminusmax_ukernel__avx512f_p5_scalef_x48() local
73 vp0 = _mm512_fmadd_ps(vp0, vt0, vc3); in xnn_f32_vscaleexpminusmax_ukernel__avx512f_p5_scalef_x48()
77 vp0 = _mm512_fmadd_ps(vp0, vt0, vc2); in xnn_f32_vscaleexpminusmax_ukernel__avx512f_p5_scalef_x48()
81 vp0 = _mm512_fmadd_ps(vp0, vt0, vc1); in xnn_f32_vscaleexpminusmax_ukernel__avx512f_p5_scalef_x48()
85 vp0 = _mm512_fmadd_ps(vp0, vt0, vc0); in xnn_f32_vscaleexpminusmax_ukernel__avx512f_p5_scalef_x48()
92 __m512 vf0 = _mm512_scalef_ps(vp0, vn0); in xnn_f32_vscaleexpminusmax_ukernel__avx512f_p5_scalef_x48()
|
| /external/XNNPACK/src/f32-raddstoreexpminusmax/gen/ |
| D | scalar-p5-x2.c | 78 float vp0 = vc5 * vt0 + vc4; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x2() local
81 vp0 = vp0 * vt0 + vc3; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x2()
84 vp0 = vp0 * vt0 + vc2; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x2()
87 vp0 = vp0 * vt0 + vc1; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x2()
97 float vf0 = vt0 * vp0 + vs0; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x2()
|
| D | scalar-p5-x2-acc2.c | 79 float vp0 = vc5 * vt0 + vc4; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x2_acc2() local
82 vp0 = vp0 * vt0 + vc3; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x2_acc2()
85 vp0 = vp0 * vt0 + vc2; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x2_acc2()
88 vp0 = vp0 * vt0 + vc1; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x2_acc2()
98 float vf0 = vt0 * vp0 + vs0; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x2_acc2()
|
| D | scalar-p5-x4-acc2.c | 93 float vp0 = vc5 * vt0 + vc4; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4_acc2() local
98 vp0 = vp0 * vt0 + vc3; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4_acc2()
103 vp0 = vp0 * vt0 + vc2; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4_acc2()
108 vp0 = vp0 * vt0 + vc1; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4_acc2()
122 float vf0 = vt0 * vp0 + vs0; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4_acc2()
|
| D | scalar-p5-x4.c | 92 float vp0 = vc5 * vt0 + vc4; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4() local
97 vp0 = vp0 * vt0 + vc3; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4()
102 vp0 = vp0 * vt0 + vc2; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4()
107 vp0 = vp0 * vt0 + vc1; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4()
121 float vf0 = vt0 * vp0 + vs0; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4()
|
| D | scalar-p5-x4-acc4.c | 95 float vp0 = vc5 * vt0 + vc4; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4_acc4() local
100 vp0 = vp0 * vt0 + vc3; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4_acc4()
105 vp0 = vp0 * vt0 + vc2; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4_acc4()
110 vp0 = vp0 * vt0 + vc1; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4_acc4()
124 float vf0 = vt0 * vp0 + vs0; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4_acc4()
|
| /external/XNNPACK/src/f32-sigmoid/gen/ |
| D | scalar-p5-div-x2.c | 62 float vp0 = vt0 * vc5 + vc4; in xnn_f32_sigmoid_ukernel__scalar_p5_div_x2() local
65 vp0 = vt0 * vp0 + vc3; in xnn_f32_sigmoid_ukernel__scalar_p5_div_x2()
68 vp0 = vt0 * vp0 + vc2; in xnn_f32_sigmoid_ukernel__scalar_p5_div_x2()
71 vp0 = vt0 * vp0 + vc1; in xnn_f32_sigmoid_ukernel__scalar_p5_div_x2()
77 const float ve0 = vt0 * vp0 + vs0; in xnn_f32_sigmoid_ukernel__scalar_p5_div_x2()
|
| D | avx512f-rr1-p5-scalef-div-x32.c | 54 __m512 vp0 = _mm512_fmadd_ps(vc5, vt0, vc4); in xnn_f32_sigmoid_ukernel__avx512f_rr1_p5_scalef_div_x32() local
57 vp0 = _mm512_fmadd_ps(vp0, vt0, vc3); in xnn_f32_sigmoid_ukernel__avx512f_rr1_p5_scalef_div_x32()
60 vp0 = _mm512_fmadd_ps(vp0, vt0, vc2); in xnn_f32_sigmoid_ukernel__avx512f_rr1_p5_scalef_div_x32()
63 vp0 = _mm512_fmadd_ps(vp0, vt0, vc1); in xnn_f32_sigmoid_ukernel__avx512f_rr1_p5_scalef_div_x32()
66 vp0 = _mm512_fmadd_ps(vp0, vt0, vone); in xnn_f32_sigmoid_ukernel__avx512f_rr1_p5_scalef_div_x32()
69 const __m512 ve0 = _mm512_scalef_ps(vp0, vn0); in xnn_f32_sigmoid_ukernel__avx512f_rr1_p5_scalef_div_x32()
|