| /external/XNNPACK/src/f32-velu/gen/ |
| D | velu-scalar-rr2-p6-x2.c | 59 float vt0 = vn0 * vminus_ln2_hi + vz0; in xnn_f32_velu_ukernel__scalar_rr2_p6_x2() local
62 vt0 = vn0 * vminus_ln2_lo + vt0; in xnn_f32_velu_ukernel__scalar_rr2_p6_x2()
67 vt0 = 0.0f; in xnn_f32_velu_ukernel__scalar_rr2_p6_x2()
74 float vp0 = vc6 * vt0 + vc5; in xnn_f32_velu_ukernel__scalar_rr2_p6_x2()
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()
89 vt0 *= vs0; in xnn_f32_velu_ukernel__scalar_rr2_p6_x2()
94 vp0 = vp0 * vt0 + vt0; in xnn_f32_velu_ukernel__scalar_rr2_p6_x2()
|
| D | velu-wasm-rr2-p6-x2.c | 59 float vt0 = vn0 * vminus_ln2_hi + vz0; in xnn_f32_velu_ukernel__wasm_rr2_p6_x2() local
62 vt0 = vn0 * vminus_ln2_lo + vt0; in xnn_f32_velu_ukernel__wasm_rr2_p6_x2()
66 float vp0 = vc6 * vt0 + vc5; in xnn_f32_velu_ukernel__wasm_rr2_p6_x2()
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()
81 vt0 *= vs0; in xnn_f32_velu_ukernel__wasm_rr2_p6_x2()
86 vp0 = vp0 * vt0 + vt0; in xnn_f32_velu_ukernel__wasm_rr2_p6_x2()
|
| D | velu-scalar-rr2-p6-x3.c | 64 float vt0 = vn0 * vminus_ln2_hi + vz0; in xnn_f32_velu_ukernel__scalar_rr2_p6_x3() local
68 vt0 = vn0 * vminus_ln2_lo + vt0; in xnn_f32_velu_ukernel__scalar_rr2_p6_x3()
74 vt0 = 0.0f; in xnn_f32_velu_ukernel__scalar_rr2_p6_x3()
85 float vp0 = vc6 * vt0 + vc5; in xnn_f32_velu_ukernel__scalar_rr2_p6_x3()
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()
105 vt0 *= vs0; in xnn_f32_velu_ukernel__scalar_rr2_p6_x3()
112 vp0 = vp0 * vt0 + vt0; in xnn_f32_velu_ukernel__scalar_rr2_p6_x3()
|
| D | velu-wasm-rr2-p6-x3.c | 64 float vt0 = vn0 * vminus_ln2_hi + vz0; in xnn_f32_velu_ukernel__wasm_rr2_p6_x3() local
68 vt0 = vn0 * vminus_ln2_lo + vt0; in xnn_f32_velu_ukernel__wasm_rr2_p6_x3()
73 float vp0 = vc6 * vt0 + vc5; in xnn_f32_velu_ukernel__wasm_rr2_p6_x3()
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()
93 vt0 *= vs0; in xnn_f32_velu_ukernel__wasm_rr2_p6_x3()
100 vp0 = vp0 * vt0 + vt0; in xnn_f32_velu_ukernel__wasm_rr2_p6_x3()
|
| D | velu-scalar-rr2-p6-x4.c | 69 float vt0 = vn0 * vminus_ln2_hi + vz0; in xnn_f32_velu_ukernel__scalar_rr2_p6_x4() local
74 vt0 = vn0 * vminus_ln2_lo + vt0; in xnn_f32_velu_ukernel__scalar_rr2_p6_x4()
81 vt0 = 0.0f; in xnn_f32_velu_ukernel__scalar_rr2_p6_x4()
96 float vp0 = vc6 * vt0 + vc5; in xnn_f32_velu_ukernel__scalar_rr2_p6_x4()
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()
121 vt0 *= vs0; in xnn_f32_velu_ukernel__scalar_rr2_p6_x4()
130 vp0 = vp0 * vt0 + vt0; in xnn_f32_velu_ukernel__scalar_rr2_p6_x4()
|
| D | velu-scalar-rr2-lut16-p3-x2.c | 61 float vt0 = vn0 * vminus_ln2_hi + vz0; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x2() local
66 vt0 = vn0 * vminus_ln2_lo + vt0; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x2()
69 vt0 = 0.0f; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x2()
77 float vp0 = vc3 * vt0 + vc2; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x2()
80 vp0 *= vt0; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x2()
83 vt0 *= vs0; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x2()
88 vp0 = vp0 * vt0 + vt0; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x2()
|
| D | velu-wasm-rr2-p6-x4.c | 69 float vt0 = vn0 * vminus_ln2_hi + vz0; in xnn_f32_velu_ukernel__wasm_rr2_p6_x4() local
74 vt0 = vn0 * vminus_ln2_lo + vt0; in xnn_f32_velu_ukernel__wasm_rr2_p6_x4()
80 float vp0 = vc6 * vt0 + vc5; in xnn_f32_velu_ukernel__wasm_rr2_p6_x4()
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()
105 vt0 *= vs0; in xnn_f32_velu_ukernel__wasm_rr2_p6_x4()
114 vp0 = vp0 * vt0 + vt0; in xnn_f32_velu_ukernel__wasm_rr2_p6_x4()
|
| D | velu-wasm-rr2-lut16-p3-x2.c | 61 float vt0 = vn0 * vminus_ln2_hi + vz0; in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x2() local
66 vt0 = vn0 * vminus_ln2_lo + vt0; in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x2()
69 float vp0 = vc3 * vt0 + vc2; in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x2()
72 vp0 *= vt0; in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x2()
75 vt0 *= vs0; in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x2()
80 vp0 = vp0 * vt0 + vt0; in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x2()
|
| D | velu-scalar-rr2-p6-x5.c | 74 float vt0 = vn0 * vminus_ln2_hi + vz0; in xnn_f32_velu_ukernel__scalar_rr2_p6_x5() local
80 vt0 = vn0 * vminus_ln2_lo + vt0; in xnn_f32_velu_ukernel__scalar_rr2_p6_x5()
88 vt0 = 0.0f; in xnn_f32_velu_ukernel__scalar_rr2_p6_x5()
107 float vp0 = vc6 * vt0 + vc5; in xnn_f32_velu_ukernel__scalar_rr2_p6_x5()
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()
137 vt0 *= vs0; in xnn_f32_velu_ukernel__scalar_rr2_p6_x5()
148 vp0 = vp0 * vt0 + vt0; in xnn_f32_velu_ukernel__scalar_rr2_p6_x5()
|
| D | velu-scalar-rr2-lut16-p3-x3.c | 67 float vt0 = vn0 * vminus_ln2_hi + vz0; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x3() local
74 vt0 = vn0 * vminus_ln2_lo + vt0; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x3()
77 vt0 = 0.0f; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x3()
90 float vp0 = vc3 * vt0 + vc2; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x3()
94 vp0 *= vt0; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x3()
98 vt0 *= vs0; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x3()
105 vp0 = vp0 * vt0 + vt0; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x3()
|
| D | velu-wasm-rr2-p6-x5.c | 74 float vt0 = vn0 * vminus_ln2_hi + vz0; in xnn_f32_velu_ukernel__wasm_rr2_p6_x5() local
80 vt0 = vn0 * vminus_ln2_lo + vt0; in xnn_f32_velu_ukernel__wasm_rr2_p6_x5()
87 float vp0 = vc6 * vt0 + vc5; in xnn_f32_velu_ukernel__wasm_rr2_p6_x5()
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()
117 vt0 *= vs0; in xnn_f32_velu_ukernel__wasm_rr2_p6_x5()
128 vp0 = vp0 * vt0 + vt0; in xnn_f32_velu_ukernel__wasm_rr2_p6_x5()
|
| D | velu-scalar-rr2-p6-x6.c | 79 float vt0 = vn0 * vminus_ln2_hi + vz0; in xnn_f32_velu_ukernel__scalar_rr2_p6_x6() local
86 vt0 = vn0 * vminus_ln2_lo + vt0; in xnn_f32_velu_ukernel__scalar_rr2_p6_x6()
95 vt0 = 0.0f; in xnn_f32_velu_ukernel__scalar_rr2_p6_x6()
118 float vp0 = vc6 * vt0 + vc5; in xnn_f32_velu_ukernel__scalar_rr2_p6_x6()
125 vp0 = vp0 * vt0 + vc4; in xnn_f32_velu_ukernel__scalar_rr2_p6_x6()
132 vp0 = vp0 * vt0 + vc3; in xnn_f32_velu_ukernel__scalar_rr2_p6_x6()
139 vp0 = vp0 * vt0 + vc2; in xnn_f32_velu_ukernel__scalar_rr2_p6_x6()
146 vp0 *= vt0; in xnn_f32_velu_ukernel__scalar_rr2_p6_x6()
153 vt0 *= vs0; in xnn_f32_velu_ukernel__scalar_rr2_p6_x6()
166 vp0 = vp0 * vt0 + vt0; in xnn_f32_velu_ukernel__scalar_rr2_p6_x6()
|
| /external/XNNPACK/src/u8-clamp/ |
| D | scalar-x4.c | 23 uint8_t vt0 = x[0]; in xnn_u8_clamp_ukernel__scalar_x4() local
29 vt0 = XNN_UNPREDICTABLE(vt0 < voutput_min) ? voutput_min : vt0; in xnn_u8_clamp_ukernel__scalar_x4()
34 vt0 = XNN_UNPREDICTABLE(vt0 > voutput_max) ? voutput_max : vt0; in xnn_u8_clamp_ukernel__scalar_x4()
39 y[0] = vt0; in xnn_u8_clamp_ukernel__scalar_x4()
|
| /external/XNNPACK/src/f32-raddstoreexpminusmax/gen/ |
| D | scalar-p5-x2.c | 71 float vt0 = vn0 * vminus_ln2_hi + vx0; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x2() local
74 vt0 = vn0 * vminus_ln2_lo + vt0; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x2()
78 float vp0 = vc5 * vt0 + vc4; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x2()
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()
94 vt0 *= vs0; 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 | 72 float vt0 = vn0 * vminus_ln2_hi + vx0; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x2_acc2() local
75 vt0 = vn0 * vminus_ln2_lo + vt0; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x2_acc2()
79 float vp0 = vc5 * vt0 + vc4; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x2_acc2()
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()
95 vt0 *= vs0; 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 | 82 float vt0 = vn0 * vminus_ln2_hi + vx0; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4_acc2() local
87 vt0 = vn0 * vminus_ln2_lo + vt0; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4_acc2()
93 float vp0 = vc5 * vt0 + vc4; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4_acc2()
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()
117 vt0 *= vs0; 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 | 81 float vt0 = vn0 * vminus_ln2_hi + vx0; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4() local
86 vt0 = vn0 * vminus_ln2_lo + vt0; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4()
92 float vp0 = vc5 * vt0 + vc4; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4()
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()
116 vt0 *= vs0; 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 | 84 float vt0 = vn0 * vminus_ln2_hi + vx0; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4_acc4() local
89 vt0 = vn0 * vminus_ln2_lo + vt0; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4_acc4()
95 float vp0 = vc5 * vt0 + vc4; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_p5_x4_acc4()
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()
119 vt0 *= vs0; 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 | 56 float vt0 = vn0 * vln2_hi + vz0; in xnn_f32_sigmoid_ukernel__scalar_p5_div_x2() local
59 vt0 = vn0 * vln2_lo + vt0; in xnn_f32_sigmoid_ukernel__scalar_p5_div_x2()
62 float vp0 = vt0 * vc5 + vc4; in xnn_f32_sigmoid_ukernel__scalar_p5_div_x2()
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()
74 vt0 *= vs0; 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 | scalar-p5-div-x4.c | 66 float vt0 = vn0 * vln2_hi + vz0; in xnn_f32_sigmoid_ukernel__scalar_p5_div_x4() local
71 vt0 = vn0 * vln2_lo + vt0; in xnn_f32_sigmoid_ukernel__scalar_p5_div_x4()
76 float vp0 = vt0 * vc5 + vc4; in xnn_f32_sigmoid_ukernel__scalar_p5_div_x4()
81 vp0 = vt0 * vp0 + vc3; in xnn_f32_sigmoid_ukernel__scalar_p5_div_x4()
86 vp0 = vt0 * vp0 + vc2; in xnn_f32_sigmoid_ukernel__scalar_p5_div_x4()
91 vp0 = vt0 * vp0 + vc1; in xnn_f32_sigmoid_ukernel__scalar_p5_div_x4()
96 vt0 *= vs0; in xnn_f32_sigmoid_ukernel__scalar_p5_div_x4()
101 const float ve0 = vt0 * vp0 + vs0; in xnn_f32_sigmoid_ukernel__scalar_p5_div_x4()
|
| /external/XNNPACK/src/f32-vscaleexpminusmax/gen/ |
| D | avx2-p5-x8.c | 65 __m256 vt0 = _mm256_fmadd_ps(vn0, vminus_ln2_hi, vx0); in xnn_f32_vscaleexpminusmax_ukernel__avx2_p5_x8() local
67 vt0 = _mm256_fmadd_ps(vn0, vminus_ln2_lo, vt0); in xnn_f32_vscaleexpminusmax_ukernel__avx2_p5_x8()
70 __m256 vp0 = _mm256_fmadd_ps(vc5, vt0, vc4); in xnn_f32_vscaleexpminusmax_ukernel__avx2_p5_x8()
72 vp0 = _mm256_fmadd_ps(vp0, vt0, vc3); in xnn_f32_vscaleexpminusmax_ukernel__avx2_p5_x8()
74 vp0 = _mm256_fmadd_ps(vp0, vt0, vc2); in xnn_f32_vscaleexpminusmax_ukernel__avx2_p5_x8()
76 vp0 = _mm256_fmadd_ps(vp0, vt0, vc1); in xnn_f32_vscaleexpminusmax_ukernel__avx2_p5_x8()
82 vt0 = _mm256_mul_ps(vt0, vs0); in xnn_f32_vscaleexpminusmax_ukernel__avx2_p5_x8()
84 __m256 vf0 = _mm256_fmadd_ps(vt0, vp0, vs0); in xnn_f32_vscaleexpminusmax_ukernel__avx2_p5_x8()
|
| D | avx512f-p5-scalef-x16.c | 54 __m512 vt0 = _mm512_fmadd_ps(vn0, vminus_ln2_hi, vx0); in xnn_f32_vscaleexpminusmax_ukernel__avx512f_p5_scalef_x16() local
56 vt0 = _mm512_fmadd_ps(vn0, vminus_ln2_lo, vt0); in xnn_f32_vscaleexpminusmax_ukernel__avx512f_p5_scalef_x16()
59 __m512 vp0 = _mm512_fmadd_ps(vc5, vt0, vc4); in xnn_f32_vscaleexpminusmax_ukernel__avx512f_p5_scalef_x16()
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()
|
| D | avx2-p5-x16.c | 70 __m256 vt0 = _mm256_fmadd_ps(vn0, vminus_ln2_hi, vx0); in xnn_f32_vscaleexpminusmax_ukernel__avx2_p5_x16() local
73 vt0 = _mm256_fmadd_ps(vn0, vminus_ln2_lo, vt0); in xnn_f32_vscaleexpminusmax_ukernel__avx2_p5_x16()
77 __m256 vp0 = _mm256_fmadd_ps(vc5, vt0, vc4); in xnn_f32_vscaleexpminusmax_ukernel__avx2_p5_x16()
80 vp0 = _mm256_fmadd_ps(vp0, vt0, vc3); in xnn_f32_vscaleexpminusmax_ukernel__avx2_p5_x16()
83 vp0 = _mm256_fmadd_ps(vp0, vt0, vc2); in xnn_f32_vscaleexpminusmax_ukernel__avx2_p5_x16()
86 vp0 = _mm256_fmadd_ps(vp0, vt0, vc1); in xnn_f32_vscaleexpminusmax_ukernel__avx2_p5_x16()
93 vt0 = _mm256_mul_ps(vt0, vs0); in xnn_f32_vscaleexpminusmax_ukernel__avx2_p5_x16()
96 __m256 vf0 = _mm256_fmadd_ps(vt0, vp0, vs0); in xnn_f32_vscaleexpminusmax_ukernel__avx2_p5_x16()
|
| /external/XNNPACK/src/f32-vscaleextexp/gen/ |
| D | avx512f-p5-scalef-x16.c | 52 __m512 vt0 = _mm512_fmadd_ps(vn0, vminus_ln2_hi, vx0); in xnn_f32_vscaleextexp_ukernel__avx512f_p5_scalef_x16() local
54 vt0 = _mm512_fmadd_ps(vn0, vminus_ln2_lo, vt0); in xnn_f32_vscaleextexp_ukernel__avx512f_p5_scalef_x16()
57 __m512 vp0 = _mm512_fmadd_ps(vc5, vt0, vc4); in xnn_f32_vscaleextexp_ukernel__avx512f_p5_scalef_x16()
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()
|
| D | avx2-p5-x8.c | 58 __m256 vt0 = _mm256_fmadd_ps(vn0, vminus_ln2_hi, vx0); in xnn_f32_vscaleextexp_ukernel__avx2_p5_x8() local
60 vt0 = _mm256_fmadd_ps(vn0, vminus_ln2_lo, vt0); in xnn_f32_vscaleextexp_ukernel__avx2_p5_x8()
63 __m256 vp0 = _mm256_fmadd_ps(vc5, vt0, vc4); in xnn_f32_vscaleextexp_ukernel__avx2_p5_x8()
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()
|