| /external/XNNPACK/src/f32-gemm/gen/ |
| D | 1x4-wasm.c | 41 float vacc00 = w[0]; in xnn_f32_gemm_ukernel_1x4__wasm() local
57 vacc00 += va0 * vb0; in xnn_f32_gemm_ukernel_1x4__wasm()
66 vacc00 = __builtin_wasm_max_f32(vacc00, vmin); in xnn_f32_gemm_ukernel_1x4__wasm()
72 vacc00 = __builtin_wasm_min_f32(vacc00, vmax); in xnn_f32_gemm_ukernel_1x4__wasm()
78 c0[0] = vacc00; in xnn_f32_gemm_ukernel_1x4__wasm()
89 c0[0] = vacc00; in xnn_f32_gemm_ukernel_1x4__wasm()
91 vacc00 = vacc02; in xnn_f32_gemm_ukernel_1x4__wasm()
95 c0[0] = vacc00; in xnn_f32_gemm_ukernel_1x4__wasm()
|
| D | 1x4-scalar.c | 41 float vacc00 = w[0]; in xnn_f32_gemm_ukernel_1x4__scalar() local
57 vacc00 += va0 * vb0; in xnn_f32_gemm_ukernel_1x4__scalar()
66 vacc00 = math_max_f32(vacc00, vmin); in xnn_f32_gemm_ukernel_1x4__scalar()
72 vacc00 = math_min_f32(vacc00, vmax); in xnn_f32_gemm_ukernel_1x4__scalar()
78 c0[0] = vacc00; in xnn_f32_gemm_ukernel_1x4__scalar()
89 c0[0] = vacc00; in xnn_f32_gemm_ukernel_1x4__scalar()
91 vacc00 = vacc02; in xnn_f32_gemm_ukernel_1x4__scalar()
95 c0[0] = vacc00; in xnn_f32_gemm_ukernel_1x4__scalar()
|
| D | 2x4-wasm.c | 47 float vacc00 = w[0]; in xnn_f32_gemm_ukernel_2x4__wasm() local
52 float vacc10 = vacc00; in xnn_f32_gemm_ukernel_2x4__wasm()
68 vacc00 += va0 * vb0; in xnn_f32_gemm_ukernel_2x4__wasm()
81 vacc00 = __builtin_wasm_max_f32(vacc00, vmin); in xnn_f32_gemm_ukernel_2x4__wasm()
91 vacc00 = __builtin_wasm_min_f32(vacc00, vmax); in xnn_f32_gemm_ukernel_2x4__wasm()
106 c0[0] = vacc00; in xnn_f32_gemm_ukernel_2x4__wasm()
122 c0[0] = vacc00; in xnn_f32_gemm_ukernel_2x4__wasm()
124 vacc00 = vacc02; in xnn_f32_gemm_ukernel_2x4__wasm()
129 c0[0] = vacc00; in xnn_f32_gemm_ukernel_2x4__wasm()
|
| D | 2x4-scalar.c | 47 float vacc00 = w[0]; in xnn_f32_gemm_ukernel_2x4__scalar() local
52 float vacc10 = vacc00; in xnn_f32_gemm_ukernel_2x4__scalar()
68 vacc00 += va0 * vb0; in xnn_f32_gemm_ukernel_2x4__scalar()
81 vacc00 = math_max_f32(vacc00, vmin); in xnn_f32_gemm_ukernel_2x4__scalar()
91 vacc00 = math_min_f32(vacc00, vmax); in xnn_f32_gemm_ukernel_2x4__scalar()
106 c0[0] = vacc00; in xnn_f32_gemm_ukernel_2x4__scalar()
122 c0[0] = vacc00; in xnn_f32_gemm_ukernel_2x4__scalar()
124 vacc00 = vacc02; in xnn_f32_gemm_ukernel_2x4__scalar()
129 c0[0] = vacc00; in xnn_f32_gemm_ukernel_2x4__scalar()
|
| D | 4x2-scalar.c | 59 float vacc00 = w[0]; in xnn_f32_gemm_ukernel_4x2__scalar() local
62 float vacc10 = vacc00; in xnn_f32_gemm_ukernel_4x2__scalar()
64 float vacc20 = vacc00; in xnn_f32_gemm_ukernel_4x2__scalar()
66 float vacc30 = vacc00; in xnn_f32_gemm_ukernel_4x2__scalar()
80 vacc00 += va0 * vb0; in xnn_f32_gemm_ukernel_4x2__scalar()
93 vacc00 = math_max_f32(vacc00, vmin); in xnn_f32_gemm_ukernel_4x2__scalar()
103 vacc00 = math_min_f32(vacc00, vmax); in xnn_f32_gemm_ukernel_4x2__scalar()
122 c0[0] = vacc00; in xnn_f32_gemm_ukernel_4x2__scalar()
137 c0[0] = vacc00; in xnn_f32_gemm_ukernel_4x2__scalar()
|
| D | 4x2-wasm.c | 59 float vacc00 = w[0]; in xnn_f32_gemm_ukernel_4x2__wasm() local
62 float vacc10 = vacc00; in xnn_f32_gemm_ukernel_4x2__wasm()
64 float vacc20 = vacc00; in xnn_f32_gemm_ukernel_4x2__wasm()
66 float vacc30 = vacc00; in xnn_f32_gemm_ukernel_4x2__wasm()
80 vacc00 += va0 * vb0; in xnn_f32_gemm_ukernel_4x2__wasm()
93 vacc00 = __builtin_wasm_max_f32(vacc00, vmin); in xnn_f32_gemm_ukernel_4x2__wasm()
103 vacc00 = __builtin_wasm_min_f32(vacc00, vmax); in xnn_f32_gemm_ukernel_4x2__wasm()
122 c0[0] = vacc00; in xnn_f32_gemm_ukernel_4x2__wasm()
137 c0[0] = vacc00; in xnn_f32_gemm_ukernel_4x2__wasm()
|
| D | 4x4-scalar.c | 59 float vacc00 = w[0]; in xnn_f32_gemm_ukernel_4x4__scalar() local
64 float vacc10 = vacc00; in xnn_f32_gemm_ukernel_4x4__scalar()
68 float vacc20 = vacc00; in xnn_f32_gemm_ukernel_4x4__scalar()
72 float vacc30 = vacc00; in xnn_f32_gemm_ukernel_4x4__scalar()
90 vacc00 += va0 * vb0; in xnn_f32_gemm_ukernel_4x4__scalar()
111 vacc00 = math_max_f32(vacc00, vmin); in xnn_f32_gemm_ukernel_4x4__scalar()
129 vacc00 = math_min_f32(vacc00, vmax); in xnn_f32_gemm_ukernel_4x4__scalar()
162 c0[0] = vacc00; in xnn_f32_gemm_ukernel_4x4__scalar()
188 c0[0] = vacc00; in xnn_f32_gemm_ukernel_4x4__scalar()
190 vacc00 = vacc02; in xnn_f32_gemm_ukernel_4x4__scalar()
[all …]
|
| D | 4x4-wasm.c | 59 float vacc00 = w[0]; in xnn_f32_gemm_ukernel_4x4__wasm() local
64 float vacc10 = vacc00; in xnn_f32_gemm_ukernel_4x4__wasm()
68 float vacc20 = vacc00; in xnn_f32_gemm_ukernel_4x4__wasm()
72 float vacc30 = vacc00; in xnn_f32_gemm_ukernel_4x4__wasm()
90 vacc00 += va0 * vb0; in xnn_f32_gemm_ukernel_4x4__wasm()
111 vacc00 = __builtin_wasm_max_f32(vacc00, vmin); in xnn_f32_gemm_ukernel_4x4__wasm()
129 vacc00 = __builtin_wasm_min_f32(vacc00, vmax); in xnn_f32_gemm_ukernel_4x4__wasm()
162 c0[0] = vacc00; in xnn_f32_gemm_ukernel_4x4__wasm()
188 c0[0] = vacc00; in xnn_f32_gemm_ukernel_4x4__wasm()
190 vacc00 = vacc02; in xnn_f32_gemm_ukernel_4x4__wasm()
[all …]
|
| /external/XNNPACK/src/f32-gemm/gen-inc/ |
| D | 1x4-wasm.c | 43 float vacc00 = acc[0]; in xnn_f32_gemminc_ukernel_1x4__wasm() local
59 vacc00 += va0 * vb0; in xnn_f32_gemminc_ukernel_1x4__wasm()
68 vacc00 = __builtin_wasm_max_f32(vacc00, vmin); in xnn_f32_gemminc_ukernel_1x4__wasm()
74 vacc00 = __builtin_wasm_min_f32(vacc00, vmax); in xnn_f32_gemminc_ukernel_1x4__wasm()
80 c0[0] = vacc00; in xnn_f32_gemminc_ukernel_1x4__wasm()
91 c0[0] = vacc00; in xnn_f32_gemminc_ukernel_1x4__wasm()
93 vacc00 = vacc02; in xnn_f32_gemminc_ukernel_1x4__wasm()
97 c0[0] = vacc00; in xnn_f32_gemminc_ukernel_1x4__wasm()
|
| D | 1x4-scalar.c | 43 float vacc00 = acc[0]; in xnn_f32_gemminc_ukernel_1x4__scalar() local
59 vacc00 += va0 * vb0; in xnn_f32_gemminc_ukernel_1x4__scalar()
68 vacc00 = math_max_f32(vacc00, vmin); in xnn_f32_gemminc_ukernel_1x4__scalar()
74 vacc00 = math_min_f32(vacc00, vmax); in xnn_f32_gemminc_ukernel_1x4__scalar()
80 c0[0] = vacc00; in xnn_f32_gemminc_ukernel_1x4__scalar()
91 c0[0] = vacc00; in xnn_f32_gemminc_ukernel_1x4__scalar()
93 vacc00 = vacc02; in xnn_f32_gemminc_ukernel_1x4__scalar()
97 c0[0] = vacc00; in xnn_f32_gemminc_ukernel_1x4__scalar()
|
| D | 2x4-scalar.c | 49 float vacc00 = acc[0]; in xnn_f32_gemminc_ukernel_2x4__scalar() local
70 vacc00 += va0 * vb0; in xnn_f32_gemminc_ukernel_2x4__scalar()
83 vacc00 = math_max_f32(vacc00, vmin); in xnn_f32_gemminc_ukernel_2x4__scalar()
93 vacc00 = math_min_f32(vacc00, vmax); in xnn_f32_gemminc_ukernel_2x4__scalar()
108 c0[0] = vacc00; in xnn_f32_gemminc_ukernel_2x4__scalar()
124 c0[0] = vacc00; in xnn_f32_gemminc_ukernel_2x4__scalar()
126 vacc00 = vacc02; in xnn_f32_gemminc_ukernel_2x4__scalar()
131 c0[0] = vacc00; in xnn_f32_gemminc_ukernel_2x4__scalar()
|
| D | 2x4-wasm.c | 49 float vacc00 = acc[0]; in xnn_f32_gemminc_ukernel_2x4__wasm() local
70 vacc00 += va0 * vb0; in xnn_f32_gemminc_ukernel_2x4__wasm()
83 vacc00 = __builtin_wasm_max_f32(vacc00, vmin); in xnn_f32_gemminc_ukernel_2x4__wasm()
93 vacc00 = __builtin_wasm_min_f32(vacc00, vmax); in xnn_f32_gemminc_ukernel_2x4__wasm()
108 c0[0] = vacc00; in xnn_f32_gemminc_ukernel_2x4__wasm()
124 c0[0] = vacc00; in xnn_f32_gemminc_ukernel_2x4__wasm()
126 vacc00 = vacc02; in xnn_f32_gemminc_ukernel_2x4__wasm()
131 c0[0] = vacc00; in xnn_f32_gemminc_ukernel_2x4__wasm()
|
| D | 4x4-wasm.c | 61 float vacc00 = acc[0]; in xnn_f32_gemminc_ukernel_4x4__wasm() local
92 vacc00 += va0 * vb0; in xnn_f32_gemminc_ukernel_4x4__wasm()
113 vacc00 = __builtin_wasm_max_f32(vacc00, vmin); in xnn_f32_gemminc_ukernel_4x4__wasm()
131 vacc00 = __builtin_wasm_min_f32(vacc00, vmax); in xnn_f32_gemminc_ukernel_4x4__wasm()
164 c0[0] = vacc00; in xnn_f32_gemminc_ukernel_4x4__wasm()
190 c0[0] = vacc00; in xnn_f32_gemminc_ukernel_4x4__wasm()
192 vacc00 = vacc02; in xnn_f32_gemminc_ukernel_4x4__wasm()
199 c0[0] = vacc00; in xnn_f32_gemminc_ukernel_4x4__wasm()
|
| D | 4x4-scalar.c | 61 float vacc00 = acc[0]; in xnn_f32_gemminc_ukernel_4x4__scalar() local
92 vacc00 += va0 * vb0; in xnn_f32_gemminc_ukernel_4x4__scalar()
113 vacc00 = math_max_f32(vacc00, vmin); in xnn_f32_gemminc_ukernel_4x4__scalar()
131 vacc00 = math_min_f32(vacc00, vmax); in xnn_f32_gemminc_ukernel_4x4__scalar()
164 c0[0] = vacc00; in xnn_f32_gemminc_ukernel_4x4__scalar()
190 c0[0] = vacc00; in xnn_f32_gemminc_ukernel_4x4__scalar()
192 vacc00 = vacc02; in xnn_f32_gemminc_ukernel_4x4__scalar()
199 c0[0] = vacc00; in xnn_f32_gemminc_ukernel_4x4__scalar()
|
| /external/XNNPACK/src/f32-igemm/gen/ |
| D | 1x4-wasm.c | 45 float vacc00 = w[0]; in xnn_f32_igemm_ukernel_1x4__wasm() local
70 vacc00 += va0 * vb0; in xnn_f32_igemm_ukernel_1x4__wasm()
81 vacc00 = __builtin_wasm_max_f32(vacc00, vmin); in xnn_f32_igemm_ukernel_1x4__wasm()
87 vacc00 = __builtin_wasm_min_f32(vacc00, vmax); in xnn_f32_igemm_ukernel_1x4__wasm()
93 c0[0] = vacc00; in xnn_f32_igemm_ukernel_1x4__wasm()
103 c0[0] = vacc00; in xnn_f32_igemm_ukernel_1x4__wasm()
105 vacc00 = vacc02; in xnn_f32_igemm_ukernel_1x4__wasm()
109 c0[0] = vacc00; in xnn_f32_igemm_ukernel_1x4__wasm()
|
| D | 1x4-scalar.c | 45 float vacc00 = w[0]; in xnn_f32_igemm_ukernel_1x4__scalar() local
70 vacc00 += va0 * vb0; in xnn_f32_igemm_ukernel_1x4__scalar()
81 vacc00 = math_max_f32(vacc00, vmin); in xnn_f32_igemm_ukernel_1x4__scalar()
87 vacc00 = math_min_f32(vacc00, vmax); in xnn_f32_igemm_ukernel_1x4__scalar()
93 c0[0] = vacc00; in xnn_f32_igemm_ukernel_1x4__scalar()
103 c0[0] = vacc00; in xnn_f32_igemm_ukernel_1x4__scalar()
105 vacc00 = vacc02; in xnn_f32_igemm_ukernel_1x4__scalar()
109 c0[0] = vacc00; in xnn_f32_igemm_ukernel_1x4__scalar()
|
| D | 2x4-wasm.c | 49 float vacc00 = w[0]; in xnn_f32_igemm_ukernel_2x4__wasm() local
53 float vacc10 = vacc00; in xnn_f32_igemm_ukernel_2x4__wasm()
84 vacc00 += va0 * vb0; in xnn_f32_igemm_ukernel_2x4__wasm()
99 vacc00 = __builtin_wasm_max_f32(vacc00, vmin); in xnn_f32_igemm_ukernel_2x4__wasm()
109 vacc00 = __builtin_wasm_min_f32(vacc00, vmax); in xnn_f32_igemm_ukernel_2x4__wasm()
124 c0[0] = vacc00; in xnn_f32_igemm_ukernel_2x4__wasm()
138 c0[0] = vacc00; in xnn_f32_igemm_ukernel_2x4__wasm()
140 vacc00 = vacc02; in xnn_f32_igemm_ukernel_2x4__wasm()
145 c0[0] = vacc00; in xnn_f32_igemm_ukernel_2x4__wasm()
|
| D | 2x4-scalar.c | 49 float vacc00 = w[0]; in xnn_f32_igemm_ukernel_2x4__scalar() local
53 float vacc10 = vacc00; in xnn_f32_igemm_ukernel_2x4__scalar()
84 vacc00 += va0 * vb0; in xnn_f32_igemm_ukernel_2x4__scalar()
99 vacc00 = math_max_f32(vacc00, vmin); in xnn_f32_igemm_ukernel_2x4__scalar()
109 vacc00 = math_min_f32(vacc00, vmax); in xnn_f32_igemm_ukernel_2x4__scalar()
124 c0[0] = vacc00; in xnn_f32_igemm_ukernel_2x4__scalar()
138 c0[0] = vacc00; in xnn_f32_igemm_ukernel_2x4__scalar()
140 vacc00 = vacc02; in xnn_f32_igemm_ukernel_2x4__scalar()
145 c0[0] = vacc00; in xnn_f32_igemm_ukernel_2x4__scalar()
|
| D | 4x2-scalar.c | 57 float vacc00 = w[0]; in xnn_f32_igemm_ukernel_4x2__scalar() local
59 float vacc10 = vacc00; in xnn_f32_igemm_ukernel_4x2__scalar()
61 float vacc20 = vacc00; in xnn_f32_igemm_ukernel_4x2__scalar()
63 float vacc30 = vacc00; in xnn_f32_igemm_ukernel_4x2__scalar()
102 vacc00 += va0 * vb0; in xnn_f32_igemm_ukernel_4x2__scalar()
117 vacc00 = math_max_f32(vacc00, vmin); in xnn_f32_igemm_ukernel_4x2__scalar()
127 vacc00 = math_min_f32(vacc00, vmax); in xnn_f32_igemm_ukernel_4x2__scalar()
146 c0[0] = vacc00; in xnn_f32_igemm_ukernel_4x2__scalar()
157 c0[0] = vacc00; in xnn_f32_igemm_ukernel_4x2__scalar()
|
| D | 4x2-wasm.c | 57 float vacc00 = w[0]; in xnn_f32_igemm_ukernel_4x2__wasm() local
59 float vacc10 = vacc00; in xnn_f32_igemm_ukernel_4x2__wasm()
61 float vacc20 = vacc00; in xnn_f32_igemm_ukernel_4x2__wasm()
63 float vacc30 = vacc00; in xnn_f32_igemm_ukernel_4x2__wasm()
102 vacc00 += va0 * vb0; in xnn_f32_igemm_ukernel_4x2__wasm()
117 vacc00 = __builtin_wasm_max_f32(vacc00, vmin); in xnn_f32_igemm_ukernel_4x2__wasm()
127 vacc00 = __builtin_wasm_min_f32(vacc00, vmax); in xnn_f32_igemm_ukernel_4x2__wasm()
146 c0[0] = vacc00; in xnn_f32_igemm_ukernel_4x2__wasm()
157 c0[0] = vacc00; in xnn_f32_igemm_ukernel_4x2__wasm()
|
| D | 4x4-scalar.c | 57 float vacc00 = w[0]; in xnn_f32_igemm_ukernel_4x4__scalar() local
61 float vacc10 = vacc00; in xnn_f32_igemm_ukernel_4x4__scalar()
65 float vacc20 = vacc00; in xnn_f32_igemm_ukernel_4x4__scalar()
69 float vacc30 = vacc00; in xnn_f32_igemm_ukernel_4x4__scalar()
112 vacc00 += va0 * vb0; in xnn_f32_igemm_ukernel_4x4__scalar()
135 vacc00 = math_max_f32(vacc00, vmin); in xnn_f32_igemm_ukernel_4x4__scalar()
153 vacc00 = math_min_f32(vacc00, vmax); in xnn_f32_igemm_ukernel_4x4__scalar()
186 c0[0] = vacc00; in xnn_f32_igemm_ukernel_4x4__scalar()
208 c0[0] = vacc00; in xnn_f32_igemm_ukernel_4x4__scalar()
210 vacc00 = vacc02; in xnn_f32_igemm_ukernel_4x4__scalar()
[all …]
|
| D | 4x4-wasm.c | 57 float vacc00 = w[0]; in xnn_f32_igemm_ukernel_4x4__wasm() local
61 float vacc10 = vacc00; in xnn_f32_igemm_ukernel_4x4__wasm()
65 float vacc20 = vacc00; in xnn_f32_igemm_ukernel_4x4__wasm()
69 float vacc30 = vacc00; in xnn_f32_igemm_ukernel_4x4__wasm()
112 vacc00 += va0 * vb0; in xnn_f32_igemm_ukernel_4x4__wasm()
135 vacc00 = __builtin_wasm_max_f32(vacc00, vmin); in xnn_f32_igemm_ukernel_4x4__wasm()
153 vacc00 = __builtin_wasm_min_f32(vacc00, vmax); in xnn_f32_igemm_ukernel_4x4__wasm()
186 c0[0] = vacc00; in xnn_f32_igemm_ukernel_4x4__wasm()
208 c0[0] = vacc00; in xnn_f32_igemm_ukernel_4x4__wasm()
210 vacc00 = vacc02; in xnn_f32_igemm_ukernel_4x4__wasm()
[all …]
|
| /external/XNNPACK/src/q8-gemm/ |
| D | 2x4c8-sse2.c | 65 __m128i vacc00 = _mm_cvtsi32_si128((int) ((const int32_t*) w)[0]); in xnn_q8_gemm_ukernel_2x4c8__sse2() local
69 __m128i vacc10 = vacc00; in xnn_q8_gemm_ukernel_2x4c8__sse2()
94 vacc00 = _mm_add_epi32(vacc00, _mm_madd_epi16(vxa0, vxb0)); in xnn_q8_gemm_ukernel_2x4c8__sse2()
104 __m128i vacc0x0123 = sse_reduce4_i32(vacc00, vacc01, vacc02, vacc03); in xnn_q8_gemm_ukernel_2x4c8__sse2()
|