| /external/XNNPACK/src/f32-prelu/gen/ |
| D | wasmsimd-minmax-4x16.c | 55 const v128_t vzero = wasm_f32x4_splat(0.0f); in xnn_f32_prelu_ukernel__wasmsimd_minmax_4x16() local
87 v128_t vacc0x0123 = wasm_i32x4_max(vi0x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_4x16()
88 vi0x0123 = wasm_i32x4_min(vi0x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_4x16()
89 v128_t vacc0x4567 = wasm_i32x4_max(vi0x4567, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_4x16()
90 vi0x4567 = wasm_i32x4_min(vi0x4567, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_4x16()
91 v128_t vacc0x89AB = wasm_i32x4_max(vi0x89AB, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_4x16()
92 vi0x89AB = wasm_i32x4_min(vi0x89AB, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_4x16()
93 v128_t vacc0xCDEF = wasm_i32x4_max(vi0xCDEF, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_4x16()
94 vi0xCDEF = wasm_i32x4_min(vi0xCDEF, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_4x16()
95 v128_t vacc1x0123 = wasm_i32x4_max(vi1x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_4x16()
[all …]
|
| D | wasm-2x4.c | 41 const float vzero = 0.0f; in xnn_f32_prelu_ukernel__wasm_2x4() local
62 float vacc0x0 = __builtin_wasm_max_f32(vi0x0, vzero); in xnn_f32_prelu_ukernel__wasm_2x4()
63 vi0x0 = __builtin_wasm_min_f32(vi0x0, vzero); in xnn_f32_prelu_ukernel__wasm_2x4()
64 float vacc0x1 = __builtin_wasm_max_f32(vi0x1, vzero); in xnn_f32_prelu_ukernel__wasm_2x4()
65 vi0x1 = __builtin_wasm_min_f32(vi0x1, vzero); in xnn_f32_prelu_ukernel__wasm_2x4()
66 float vacc0x2 = __builtin_wasm_max_f32(vi0x2, vzero); in xnn_f32_prelu_ukernel__wasm_2x4()
67 vi0x2 = __builtin_wasm_min_f32(vi0x2, vzero); in xnn_f32_prelu_ukernel__wasm_2x4()
68 float vacc0x3 = __builtin_wasm_max_f32(vi0x3, vzero); in xnn_f32_prelu_ukernel__wasm_2x4()
69 vi0x3 = __builtin_wasm_min_f32(vi0x3, vzero); in xnn_f32_prelu_ukernel__wasm_2x4()
70 float vacc1x0 = __builtin_wasm_max_f32(vi1x0, vzero); in xnn_f32_prelu_ukernel__wasm_2x4()
[all …]
|
| D | wasmsimd-minmax-4x8.c | 55 const v128_t vzero = wasm_f32x4_splat(0.0f); in xnn_f32_prelu_ukernel__wasmsimd_minmax_4x8() local
77 v128_t vacc0x0123 = wasm_i32x4_max(vi0x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_4x8()
78 vi0x0123 = wasm_i32x4_min(vi0x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_4x8()
79 v128_t vacc0x4567 = wasm_i32x4_max(vi0x4567, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_4x8()
80 vi0x4567 = wasm_i32x4_min(vi0x4567, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_4x8()
81 v128_t vacc1x0123 = wasm_i32x4_max(vi1x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_4x8()
82 vi1x0123 = wasm_i32x4_min(vi1x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_4x8()
83 v128_t vacc1x4567 = wasm_i32x4_max(vi1x4567, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_4x8()
84 vi1x4567 = wasm_i32x4_min(vi1x4567, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_4x8()
85 v128_t vacc2x0123 = wasm_i32x4_max(vi2x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_4x8()
[all …]
|
| D | wasmsimd-minmax-2x16.c | 43 const v128_t vzero = wasm_f32x4_splat(0.0f); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x16() local
65 v128_t vacc0x0123 = wasm_i32x4_max(vi0x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x16()
66 vi0x0123 = wasm_i32x4_min(vi0x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x16()
67 v128_t vacc0x4567 = wasm_i32x4_max(vi0x4567, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x16()
68 vi0x4567 = wasm_i32x4_min(vi0x4567, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x16()
69 v128_t vacc0x89AB = wasm_i32x4_max(vi0x89AB, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x16()
70 vi0x89AB = wasm_i32x4_min(vi0x89AB, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x16()
71 v128_t vacc0xCDEF = wasm_i32x4_max(vi0xCDEF, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x16()
72 vi0xCDEF = wasm_i32x4_min(vi0xCDEF, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x16()
73 v128_t vacc1x0123 = wasm_i32x4_max(vi1x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x16()
[all …]
|
| D | wasmsimd-minmax-2x8.c | 43 const v128_t vzero = wasm_f32x4_splat(0.0f); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x8() local
59 v128_t vacc0x0123 = wasm_i32x4_max(vi0x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x8()
60 vi0x0123 = wasm_i32x4_min(vi0x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x8()
61 v128_t vacc0x4567 = wasm_i32x4_max(vi0x4567, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x8()
62 vi0x4567 = wasm_i32x4_min(vi0x4567, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x8()
63 v128_t vacc1x0123 = wasm_i32x4_max(vi1x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x8()
64 vi1x0123 = wasm_i32x4_min(vi1x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x8()
65 v128_t vacc1x4567 = wasm_i32x4_max(vi1x4567, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x8()
66 vi1x4567 = wasm_i32x4_min(vi1x4567, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x8()
89 v128_t vacc0x0123 = wasm_i32x4_max(vi0x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x8()
[all …]
|
| D | wasmsimd-minmax-1x16.c | 37 const v128_t vzero = wasm_f32x4_splat(0.0f); in xnn_f32_prelu_ukernel__wasmsimd_minmax_1x16() local
54 v128_t vacc0x0123 = wasm_i32x4_max(vi0x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_1x16()
55 vi0x0123 = wasm_i32x4_min(vi0x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_1x16()
56 v128_t vacc0x4567 = wasm_i32x4_max(vi0x4567, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_1x16()
57 vi0x4567 = wasm_i32x4_min(vi0x4567, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_1x16()
58 v128_t vacc0x89AB = wasm_i32x4_max(vi0x89AB, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_1x16()
59 vi0x89AB = wasm_i32x4_min(vi0x89AB, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_1x16()
60 v128_t vacc0xCDEF = wasm_i32x4_max(vi0xCDEF, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_1x16()
61 vi0xCDEF = wasm_i32x4_min(vi0xCDEF, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_1x16()
81 v128_t vacc0x0123 = wasm_i32x4_max(vi0x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_1x16()
[all …]
|
| D | wasmsimd-minmax-4x4.c | 55 const v128_t vzero = wasm_f32x4_splat(0.0f); in xnn_f32_prelu_ukernel__wasmsimd_minmax_4x4() local
72 v128_t vacc0x0123 = wasm_i32x4_max(vi0x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_4x4()
73 vi0x0123 = wasm_i32x4_min(vi0x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_4x4()
74 v128_t vacc1x0123 = wasm_i32x4_max(vi1x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_4x4()
75 vi1x0123 = wasm_i32x4_min(vi1x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_4x4()
76 v128_t vacc2x0123 = wasm_i32x4_max(vi2x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_4x4()
77 vi2x0123 = wasm_i32x4_min(vi2x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_4x4()
78 v128_t vacc3x0123 = wasm_i32x4_max(vi3x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_4x4()
79 vi3x0123 = wasm_i32x4_min(vi3x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_4x4()
108 v128_t vacc0x0123 = wasm_i32x4_max(vi0x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_4x4()
[all …]
|
| D | wasmsimd-bitselect-4x16.c | 55 const v128_t vzero = wasm_i32x4_splat(0); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_4x16() local
88 const v128_t vmask0x0123 = wasm_i32x4_lt(vi0x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_4x16()
90 const v128_t vmask0x4567 = wasm_i32x4_lt(vi0x4567, vzero); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_4x16()
92 const v128_t vmask0x89AB = wasm_i32x4_lt(vi0x89AB, vzero); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_4x16()
94 const v128_t vmask0xCDEF = wasm_i32x4_lt(vi0xCDEF, vzero); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_4x16()
96 const v128_t vmask1x0123 = wasm_i32x4_lt(vi1x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_4x16()
98 const v128_t vmask1x4567 = wasm_i32x4_lt(vi1x4567, vzero); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_4x16()
100 const v128_t vmask1x89AB = wasm_i32x4_lt(vi1x89AB, vzero); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_4x16()
102 const v128_t vmask1xCDEF = wasm_i32x4_lt(vi1xCDEF, vzero); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_4x16()
104 const v128_t vmask2x0123 = wasm_i32x4_lt(vi2x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_4x16()
[all …]
|
| D | wasmsimd-bitselect-4x8.c | 55 const v128_t vzero = wasm_i32x4_splat(0); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_4x8() local
78 const v128_t vmask0x0123 = wasm_i32x4_lt(vi0x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_4x8()
80 const v128_t vmask0x4567 = wasm_i32x4_lt(vi0x4567, vzero); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_4x8()
82 const v128_t vmask1x0123 = wasm_i32x4_lt(vi1x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_4x8()
84 const v128_t vmask1x4567 = wasm_i32x4_lt(vi1x4567, vzero); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_4x8()
86 const v128_t vmask2x0123 = wasm_i32x4_lt(vi2x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_4x8()
88 const v128_t vmask2x4567 = wasm_i32x4_lt(vi2x4567, vzero); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_4x8()
90 const v128_t vmask3x0123 = wasm_i32x4_lt(vi3x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_4x8()
92 const v128_t vmask3x4567 = wasm_i32x4_lt(vi3x4567, vzero); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_4x8()
130 const v128_t vmask0x0123 = wasm_i32x4_lt(vi0x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_4x8()
[all …]
|
| D | wasmsimd-minmax-1x8.c | 37 const v128_t vzero = wasm_f32x4_splat(0.0f); in xnn_f32_prelu_ukernel__wasmsimd_minmax_1x8() local
50 v128_t vacc0x0123 = wasm_i32x4_max(vi0x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_1x8()
51 vi0x0123 = wasm_i32x4_min(vi0x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_1x8()
52 v128_t vacc0x4567 = wasm_i32x4_max(vi0x4567, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_1x8()
53 vi0x4567 = wasm_i32x4_min(vi0x4567, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_1x8()
69 v128_t vacc0x0123 = wasm_i32x4_max(vi0x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_1x8()
70 vi0x0123 = wasm_i32x4_min(vi0x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_1x8()
84 v128_t vacc0x0123 = wasm_i32x4_max(vi0x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_1x8()
85 vi0x0123 = wasm_i32x4_min(vi0x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_1x8()
|
| D | wasmsimd-bitselect-2x16.c | 43 const v128_t vzero = wasm_i32x4_splat(0); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_2x16() local
66 const v128_t vmask0x0123 = wasm_i32x4_lt(vi0x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_2x16()
68 const v128_t vmask0x4567 = wasm_i32x4_lt(vi0x4567, vzero); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_2x16()
70 const v128_t vmask0x89AB = wasm_i32x4_lt(vi0x89AB, vzero); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_2x16()
72 const v128_t vmask0xCDEF = wasm_i32x4_lt(vi0xCDEF, vzero); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_2x16()
74 const v128_t vmask1x0123 = wasm_i32x4_lt(vi1x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_2x16()
76 const v128_t vmask1x4567 = wasm_i32x4_lt(vi1x4567, vzero); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_2x16()
78 const v128_t vmask1x89AB = wasm_i32x4_lt(vi1x89AB, vzero); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_2x16()
80 const v128_t vmask1xCDEF = wasm_i32x4_lt(vi1xCDEF, vzero); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_2x16()
112 const v128_t vmask0x0123 = wasm_i32x4_lt(vi0x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_bitselect_2x16()
[all …]
|
| D | wasmsimd-minmax-2x4.c | 43 const v128_t vzero = wasm_f32x4_splat(0.0f); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x4() local
56 v128_t vacc0x0123 = wasm_i32x4_max(vi0x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x4()
57 vi0x0123 = wasm_i32x4_min(vi0x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x4()
58 v128_t vacc1x0123 = wasm_i32x4_max(vi1x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x4()
59 vi1x0123 = wasm_i32x4_min(vi1x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x4()
78 v128_t vacc0x0123 = wasm_i32x4_max(vi0x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x4()
79 vi0x0123 = wasm_i32x4_min(vi0x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x4()
80 v128_t vacc1x0123 = wasm_i32x4_max(vi1x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x4()
81 vi1x0123 = wasm_i32x4_min(vi1x0123, vzero); in xnn_f32_prelu_ukernel__wasmsimd_minmax_2x4()
|
| /external/XNNPACK/src/f32-vlrelu/gen/ |
| D | vlrelu-wasm-x4.c | 26 const float vzero = 0.0f; in xnn_f32_vlrelu_ukernel__wasm_x4() local
35 const float vnegx0 = __builtin_wasm_min_f32(vx0, vzero); in xnn_f32_vlrelu_ukernel__wasm_x4()
36 const float vnegx1 = __builtin_wasm_min_f32(vx1, vzero); in xnn_f32_vlrelu_ukernel__wasm_x4()
37 const float vnegx2 = __builtin_wasm_min_f32(vx2, vzero); in xnn_f32_vlrelu_ukernel__wasm_x4()
38 const float vnegx3 = __builtin_wasm_min_f32(vx3, vzero); in xnn_f32_vlrelu_ukernel__wasm_x4()
41 const float vposx0 = __builtin_wasm_max_f32(vx0, vzero); in xnn_f32_vlrelu_ukernel__wasm_x4()
43 const float vposx1 = __builtin_wasm_max_f32(vx1, vzero); in xnn_f32_vlrelu_ukernel__wasm_x4()
45 const float vposx2 = __builtin_wasm_max_f32(vx2, vzero); in xnn_f32_vlrelu_ukernel__wasm_x4()
47 const float vposx3 = __builtin_wasm_max_f32(vx3, vzero); in xnn_f32_vlrelu_ukernel__wasm_x4()
63 const float vnegx = __builtin_wasm_min_f32(vx, vzero); in xnn_f32_vlrelu_ukernel__wasm_x4()
[all …]
|
| D | vlrelu-wasmsimd-minmax-x8.c | 28 const v128_t vzero = wasm_f32x4_splat(0.0f); in xnn_f32_vlrelu_ukernel__wasmsimd_minmax_x8() local
34 v128_t vacc0123 = wasm_i32x4_max(vx0123, vzero); in xnn_f32_vlrelu_ukernel__wasmsimd_minmax_x8()
35 vx0123 = wasm_i32x4_min(vx0123, vzero); in xnn_f32_vlrelu_ukernel__wasmsimd_minmax_x8()
36 v128_t vacc4567 = wasm_i32x4_max(vx4567, vzero); in xnn_f32_vlrelu_ukernel__wasmsimd_minmax_x8()
37 vx4567 = wasm_i32x4_min(vx4567, vzero); in xnn_f32_vlrelu_ukernel__wasmsimd_minmax_x8()
49 v128_t vacc = wasm_i32x4_max(vx, vzero); in xnn_f32_vlrelu_ukernel__wasmsimd_minmax_x8()
50 vx = wasm_i32x4_min(vx, vzero); in xnn_f32_vlrelu_ukernel__wasmsimd_minmax_x8()
57 v128_t vacc = wasm_i32x4_max(vx, vzero); in xnn_f32_vlrelu_ukernel__wasmsimd_minmax_x8()
58 vx = wasm_i32x4_min(vx, vzero); in xnn_f32_vlrelu_ukernel__wasmsimd_minmax_x8()
|
| D | vlrelu-wasm-x2.c | 26 const float vzero = 0.0f; in xnn_f32_vlrelu_ukernel__wasm_x2() local
33 const float vnegx0 = __builtin_wasm_min_f32(vx0, vzero); in xnn_f32_vlrelu_ukernel__wasm_x2()
34 const float vnegx1 = __builtin_wasm_min_f32(vx1, vzero); in xnn_f32_vlrelu_ukernel__wasm_x2()
37 const float vposx0 = __builtin_wasm_max_f32(vx0, vzero); in xnn_f32_vlrelu_ukernel__wasm_x2()
39 const float vposx1 = __builtin_wasm_max_f32(vx1, vzero); in xnn_f32_vlrelu_ukernel__wasm_x2()
50 const float vnegx = __builtin_wasm_min_f32(vx, vzero); in xnn_f32_vlrelu_ukernel__wasm_x2()
52 const float vposx = __builtin_wasm_max_f32(vx, vzero); in xnn_f32_vlrelu_ukernel__wasm_x2()
|
| /external/XNNPACK/src/f32-relu/gen/ |
| D | wasm-x8.c | 28 const float vzero = 0.0f; in xnn_f32_relu_ukernel__wasm_x8() local
41 vacc0 = __builtin_wasm_max_f32(vacc0, vzero); in xnn_f32_relu_ukernel__wasm_x8()
42 vacc1 = __builtin_wasm_max_f32(vacc1, vzero); in xnn_f32_relu_ukernel__wasm_x8()
43 vacc2 = __builtin_wasm_max_f32(vacc2, vzero); in xnn_f32_relu_ukernel__wasm_x8()
44 vacc3 = __builtin_wasm_max_f32(vacc3, vzero); in xnn_f32_relu_ukernel__wasm_x8()
45 vacc4 = __builtin_wasm_max_f32(vacc4, vzero); in xnn_f32_relu_ukernel__wasm_x8()
46 vacc5 = __builtin_wasm_max_f32(vacc5, vzero); in xnn_f32_relu_ukernel__wasm_x8()
47 vacc6 = __builtin_wasm_max_f32(vacc6, vzero); in xnn_f32_relu_ukernel__wasm_x8()
48 vacc7 = __builtin_wasm_max_f32(vacc7, vzero); in xnn_f32_relu_ukernel__wasm_x8()
63 vacc = __builtin_wasm_max_f32(vacc, vzero); in xnn_f32_relu_ukernel__wasm_x8()
|
| D | wasmsimd-x16.c | 29 const v128_t vzero = wasm_f32x4_splat(0.0f); in xnn_f32_relu_ukernel__wasmsimd_x16() local
38 vacc0123 = wasm_i32x4_max(vacc0123, vzero); in xnn_f32_relu_ukernel__wasmsimd_x16()
39 vacc4567 = wasm_i32x4_max(vacc4567, vzero); in xnn_f32_relu_ukernel__wasmsimd_x16()
40 vacc89AB = wasm_i32x4_max(vacc89AB, vzero); in xnn_f32_relu_ukernel__wasmsimd_x16()
41 vaccCDEF = wasm_i32x4_max(vaccCDEF, vzero); in xnn_f32_relu_ukernel__wasmsimd_x16()
53 vacc = wasm_i32x4_max(vacc, vzero); in xnn_f32_relu_ukernel__wasmsimd_x16()
61 vacc = wasm_i32x4_max(vacc, vzero); in xnn_f32_relu_ukernel__wasmsimd_x16()
|
| D | wasm-x4.c | 28 const float vzero = 0.0f; in xnn_f32_relu_ukernel__wasm_x4() local
37 vacc0 = __builtin_wasm_max_f32(vacc0, vzero); in xnn_f32_relu_ukernel__wasm_x4()
38 vacc1 = __builtin_wasm_max_f32(vacc1, vzero); in xnn_f32_relu_ukernel__wasm_x4()
39 vacc2 = __builtin_wasm_max_f32(vacc2, vzero); in xnn_f32_relu_ukernel__wasm_x4()
40 vacc3 = __builtin_wasm_max_f32(vacc3, vzero); in xnn_f32_relu_ukernel__wasm_x4()
51 vacc = __builtin_wasm_max_f32(vacc, vzero); in xnn_f32_relu_ukernel__wasm_x4()
|
| /external/XNNPACK/src/qu8-gavgpool/ |
| D | 7p7x-minmax-sse2-c8.c | 40 const __m128i vzero = _mm_setzero_si128(); in xnn_qu8_gavgpool_minmax_ukernel_7p7x__sse2_c8() local
52 const __m128i vxi0 = _mm_unpacklo_epi8(vi0, vzero); in xnn_qu8_gavgpool_minmax_ukernel_7p7x__sse2_c8()
53 const __m128i vxi1 = _mm_unpacklo_epi8(vi1, vzero); in xnn_qu8_gavgpool_minmax_ukernel_7p7x__sse2_c8()
54 const __m128i vxi2 = _mm_unpacklo_epi8(vi2, vzero); in xnn_qu8_gavgpool_minmax_ukernel_7p7x__sse2_c8()
55 const __m128i vxi3 = _mm_unpacklo_epi8(vi3, vzero); in xnn_qu8_gavgpool_minmax_ukernel_7p7x__sse2_c8()
56 const __m128i vxi4 = _mm_unpacklo_epi8(vi4, vzero); in xnn_qu8_gavgpool_minmax_ukernel_7p7x__sse2_c8()
57 const __m128i vxi5 = _mm_unpacklo_epi8(vi5, vzero); in xnn_qu8_gavgpool_minmax_ukernel_7p7x__sse2_c8()
58 const __m128i vxi6 = _mm_unpacklo_epi8(vi6, vzero); in xnn_qu8_gavgpool_minmax_ukernel_7p7x__sse2_c8()
68 const __m128i vacc_lo = _mm_add_epi32(vbias, _mm_unpacklo_epi16(vsum, vzero)); in xnn_qu8_gavgpool_minmax_ukernel_7p7x__sse2_c8()
69 const __m128i vacc_hi = _mm_add_epi32(vbias, _mm_unpackhi_epi16(vsum, vzero)); in xnn_qu8_gavgpool_minmax_ukernel_7p7x__sse2_c8()
[all …]
|
| D | 7x-minmax-sse2-c8.c | 56 const __m128i vzero = _mm_setzero_si128(); in xnn_qu8_gavgpool_minmax_ukernel_7x__sse2_c8() local
70 const __m128i vxi0 = _mm_unpacklo_epi8(vi0, vzero); in xnn_qu8_gavgpool_minmax_ukernel_7x__sse2_c8()
71 const __m128i vxi1 = _mm_unpacklo_epi8(vi1, vzero); in xnn_qu8_gavgpool_minmax_ukernel_7x__sse2_c8()
72 const __m128i vxi2 = _mm_unpacklo_epi8(vi2, vzero); in xnn_qu8_gavgpool_minmax_ukernel_7x__sse2_c8()
73 const __m128i vxi3 = _mm_unpacklo_epi8(vi3, vzero); in xnn_qu8_gavgpool_minmax_ukernel_7x__sse2_c8()
74 const __m128i vxi4 = _mm_unpacklo_epi8(vi4, vzero); in xnn_qu8_gavgpool_minmax_ukernel_7x__sse2_c8()
75 const __m128i vxi5 = _mm_unpacklo_epi8(vi5, vzero); in xnn_qu8_gavgpool_minmax_ukernel_7x__sse2_c8()
76 const __m128i vxi6 = _mm_unpacklo_epi8(vi6, vzero); in xnn_qu8_gavgpool_minmax_ukernel_7x__sse2_c8()
86 __m128i vacc_lo = _mm_add_epi32(vbias, _mm_unpacklo_epi16(vsum, vzero)); in xnn_qu8_gavgpool_minmax_ukernel_7x__sse2_c8()
87 __m128i vacc_hi = _mm_add_epi32(vbias, _mm_unpackhi_epi16(vsum, vzero)); in xnn_qu8_gavgpool_minmax_ukernel_7x__sse2_c8()
[all …]
|
| /external/libaom/libaom/aom_dsp/x86/ |
| D | sum_squares_avx2.c | 111 __m256i vzero = _mm256_setzero_si256(); in aom_var_2d_u8_avx2() local
112 __m256i v_acc_sum = vzero; in aom_var_2d_u8_avx2()
113 __m256i v_acc_sqs = vzero; in aom_var_2d_u8_avx2()
127 __m256i vsrc0 = _mm256_unpacklo_epi8(vsrc[k], vzero); in aom_var_2d_u8_avx2()
128 __m256i vsrc1 = _mm256_unpackhi_epi8(vsrc[k], vzero); in aom_var_2d_u8_avx2()
141 v_acc_sum = vzero; in aom_var_2d_u8_avx2()
142 v_acc_sqs = vzero; in aom_var_2d_u8_avx2()
148 __m256i vsrc0 = _mm256_unpacklo_epi8(vsrc, vzero); in aom_var_2d_u8_avx2()
149 __m256i vsrc1 = _mm256_unpackhi_epi8(vsrc, vzero); in aom_var_2d_u8_avx2()
164 v_acc_sum = vzero; in aom_var_2d_u8_avx2()
[all …]
|
| D | sum_squares_sse2.c | 229 __m128i vzero = _mm_setzero_si128(); in aom_var_2d_u8_sse2() local
230 __m128i v_acc_sum = vzero; in aom_var_2d_u8_sse2()
231 __m128i v_acc_sqs = vzero; in aom_var_2d_u8_sse2()
245 __m128i vsrc0 = _mm_unpacklo_epi8(vsrc[k], vzero); in aom_var_2d_u8_sse2()
246 __m128i vsrc1 = _mm_unpackhi_epi8(vsrc[k], vzero); in aom_var_2d_u8_sse2()
259 v_acc_sum = vzero; in aom_var_2d_u8_sse2()
260 v_acc_sqs = vzero; in aom_var_2d_u8_sse2()
266 __m128i vsrc0 = _mm_unpacklo_epi8(vsrc, vzero); in aom_var_2d_u8_sse2()
267 __m128i vsrc1 = _mm_unpackhi_epi8(vsrc, vzero); in aom_var_2d_u8_sse2()
282 v_acc_sum = vzero; in aom_var_2d_u8_sse2()
[all …]
|
| /external/XNNPACK/src/qu8-avgpool/ |
| D | 9p8x-minmax-sse2-c8.c | 34 const __m128i vzero = _mm_setzero_si128(); in xnn_qu8_avgpool_minmax_ukernel_9p8x__sse2_c8() local
99 const __m128i vxi0 = _mm_unpacklo_epi8(vi0, vzero); in xnn_qu8_avgpool_minmax_ukernel_9p8x__sse2_c8()
100 const __m128i vxi1 = _mm_unpacklo_epi8(vi1, vzero); in xnn_qu8_avgpool_minmax_ukernel_9p8x__sse2_c8()
101 const __m128i vxi2 = _mm_unpacklo_epi8(vi2, vzero); in xnn_qu8_avgpool_minmax_ukernel_9p8x__sse2_c8()
102 const __m128i vxi3 = _mm_unpacklo_epi8(vi3, vzero); in xnn_qu8_avgpool_minmax_ukernel_9p8x__sse2_c8()
103 const __m128i vxi4 = _mm_unpacklo_epi8(vi4, vzero); in xnn_qu8_avgpool_minmax_ukernel_9p8x__sse2_c8()
104 const __m128i vxi5 = _mm_unpacklo_epi8(vi5, vzero); in xnn_qu8_avgpool_minmax_ukernel_9p8x__sse2_c8()
105 const __m128i vxi6 = _mm_unpacklo_epi8(vi6, vzero); in xnn_qu8_avgpool_minmax_ukernel_9p8x__sse2_c8()
106 const __m128i vxi7 = _mm_unpacklo_epi8(vi7, vzero); in xnn_qu8_avgpool_minmax_ukernel_9p8x__sse2_c8()
107 const __m128i vxi8 = _mm_unpacklo_epi8(vi8, vzero); in xnn_qu8_avgpool_minmax_ukernel_9p8x__sse2_c8()
[all …]
|
| D | 9x-minmax-sse2-c8.c | 34 const __m128i vzero = _mm_setzero_si128(); in xnn_qu8_avgpool_minmax_ukernel_9x__sse2_c8() local
123 const __m128i vxi0 = _mm_unpacklo_epi8(vi0, vzero); in xnn_qu8_avgpool_minmax_ukernel_9x__sse2_c8()
124 const __m128i vxi1 = _mm_unpacklo_epi8(vi1, vzero); in xnn_qu8_avgpool_minmax_ukernel_9x__sse2_c8()
125 const __m128i vxi2 = _mm_unpacklo_epi8(vi2, vzero); in xnn_qu8_avgpool_minmax_ukernel_9x__sse2_c8()
126 const __m128i vxi3 = _mm_unpacklo_epi8(vi3, vzero); in xnn_qu8_avgpool_minmax_ukernel_9x__sse2_c8()
127 const __m128i vxi4 = _mm_unpacklo_epi8(vi4, vzero); in xnn_qu8_avgpool_minmax_ukernel_9x__sse2_c8()
128 const __m128i vxi5 = _mm_unpacklo_epi8(vi5, vzero); in xnn_qu8_avgpool_minmax_ukernel_9x__sse2_c8()
129 const __m128i vxi6 = _mm_unpacklo_epi8(vi6, vzero); in xnn_qu8_avgpool_minmax_ukernel_9x__sse2_c8()
130 const __m128i vxi7 = _mm_unpacklo_epi8(vi7, vzero); in xnn_qu8_avgpool_minmax_ukernel_9x__sse2_c8()
131 const __m128i vxi8 = _mm_unpacklo_epi8(vi8, vzero); in xnn_qu8_avgpool_minmax_ukernel_9x__sse2_c8()
[all …]
|
| /external/XNNPACK/src/qu8-dwconv/ |
| D | up8x9-minmax-sse2.c | 27 const __m128i vzero = _mm_setzero_si128(); in xnn_qu8_dwconv_minmax_ukernel_up8x9__sse2() local
76 const __m128i vxi0 = _mm_unpacklo_epi8(vi0, vzero); in xnn_qu8_dwconv_minmax_ukernel_up8x9__sse2()
78 const __m128i vxk0 = _mm_sub_epi16(_mm_unpacklo_epi8(vk0, vzero), vkernel_zero_point); in xnn_qu8_dwconv_minmax_ukernel_up8x9__sse2()
85 const __m128i vxi1 = _mm_unpacklo_epi8(vi1, vzero); in xnn_qu8_dwconv_minmax_ukernel_up8x9__sse2()
87 const __m128i vxk1 = _mm_sub_epi16(_mm_unpacklo_epi8(vk1, vzero), vkernel_zero_point); in xnn_qu8_dwconv_minmax_ukernel_up8x9__sse2()
94 const __m128i vxi2 = _mm_unpacklo_epi8(vi2, vzero); in xnn_qu8_dwconv_minmax_ukernel_up8x9__sse2()
96 const __m128i vxk2 = _mm_sub_epi16(_mm_unpacklo_epi8(vk2, vzero), vkernel_zero_point); in xnn_qu8_dwconv_minmax_ukernel_up8x9__sse2()
103 const __m128i vxi3 = _mm_unpacklo_epi8(vi3, vzero); in xnn_qu8_dwconv_minmax_ukernel_up8x9__sse2()
105 const __m128i vxk3 = _mm_sub_epi16(_mm_unpacklo_epi8(vk3, vzero), vkernel_zero_point); in xnn_qu8_dwconv_minmax_ukernel_up8x9__sse2()
112 const __m128i vxi4 = _mm_unpacklo_epi8(vi4, vzero); in xnn_qu8_dwconv_minmax_ukernel_up8x9__sse2()
[all …]
|