| /external/XNNPACK/src/qs8-dwconv/gen/ |
| D | up24x9-minmax-wasmsimd-mul16.c | 99 const v128_t vprod0x01234567 = wasm_i16x8_mul(vi0x01234567, vk0x01234567); in xnn_qs8_dwconv_minmax_ukernel_up24x9__wasmsimd_mul16()
100 const v128_t vprod0x89ABCDEF = wasm_i16x8_mul(vi0x89ABCDEF, vk0x89ABCDEF); in xnn_qs8_dwconv_minmax_ukernel_up24x9__wasmsimd_mul16()
101 const v128_t vprod0xGHIJKLMN = wasm_i16x8_mul(vi0xGHIJKLMN, vk0xGHIJKLMN); in xnn_qs8_dwconv_minmax_ukernel_up24x9__wasmsimd_mul16()
118 const v128_t vprod1x01234567 = wasm_i16x8_mul(vi1x01234567, vk1x01234567); in xnn_qs8_dwconv_minmax_ukernel_up24x9__wasmsimd_mul16()
119 const v128_t vprod1x89ABCDEF = wasm_i16x8_mul(vi1x89ABCDEF, vk1x89ABCDEF); in xnn_qs8_dwconv_minmax_ukernel_up24x9__wasmsimd_mul16()
120 const v128_t vprod1xGHIJKLMN = wasm_i16x8_mul(vi1xGHIJKLMN, vk1xGHIJKLMN); in xnn_qs8_dwconv_minmax_ukernel_up24x9__wasmsimd_mul16()
137 const v128_t vprod2x01234567 = wasm_i16x8_mul(vi2x01234567, vk2x01234567); in xnn_qs8_dwconv_minmax_ukernel_up24x9__wasmsimd_mul16()
138 const v128_t vprod2x89ABCDEF = wasm_i16x8_mul(vi2x89ABCDEF, vk2x89ABCDEF); in xnn_qs8_dwconv_minmax_ukernel_up24x9__wasmsimd_mul16()
139 const v128_t vprod2xGHIJKLMN = wasm_i16x8_mul(vi2xGHIJKLMN, vk2xGHIJKLMN); in xnn_qs8_dwconv_minmax_ukernel_up24x9__wasmsimd_mul16()
156 const v128_t vprod3x01234567 = wasm_i16x8_mul(vi3x01234567, vk3x01234567); in xnn_qs8_dwconv_minmax_ukernel_up24x9__wasmsimd_mul16()
[all …]
|
| D | up16x9-minmax-wasmsimd-mul16.c | 95 const v128_t vprod0x01234567 = wasm_i16x8_mul(vi0x01234567, vk0x01234567); in xnn_qs8_dwconv_minmax_ukernel_up16x9__wasmsimd_mul16()
96 const v128_t vprod0x89ABCDEF = wasm_i16x8_mul(vi0x89ABCDEF, vk0x89ABCDEF); in xnn_qs8_dwconv_minmax_ukernel_up16x9__wasmsimd_mul16()
109 const v128_t vprod1x01234567 = wasm_i16x8_mul(vi1x01234567, vk1x01234567); in xnn_qs8_dwconv_minmax_ukernel_up16x9__wasmsimd_mul16()
110 const v128_t vprod1x89ABCDEF = wasm_i16x8_mul(vi1x89ABCDEF, vk1x89ABCDEF); in xnn_qs8_dwconv_minmax_ukernel_up16x9__wasmsimd_mul16()
123 const v128_t vprod2x01234567 = wasm_i16x8_mul(vi2x01234567, vk2x01234567); in xnn_qs8_dwconv_minmax_ukernel_up16x9__wasmsimd_mul16()
124 const v128_t vprod2x89ABCDEF = wasm_i16x8_mul(vi2x89ABCDEF, vk2x89ABCDEF); in xnn_qs8_dwconv_minmax_ukernel_up16x9__wasmsimd_mul16()
137 const v128_t vprod3x01234567 = wasm_i16x8_mul(vi3x01234567, vk3x01234567); in xnn_qs8_dwconv_minmax_ukernel_up16x9__wasmsimd_mul16()
138 const v128_t vprod3x89ABCDEF = wasm_i16x8_mul(vi3x89ABCDEF, vk3x89ABCDEF); in xnn_qs8_dwconv_minmax_ukernel_up16x9__wasmsimd_mul16()
151 const v128_t vprod4x01234567 = wasm_i16x8_mul(vi4x01234567, vk4x01234567); in xnn_qs8_dwconv_minmax_ukernel_up16x9__wasmsimd_mul16()
152 const v128_t vprod4x89ABCDEF = wasm_i16x8_mul(vi4x89ABCDEF, vk4x89ABCDEF); in xnn_qs8_dwconv_minmax_ukernel_up16x9__wasmsimd_mul16()
[all …]
|
| D | up8x9-minmax-wasmsimd-mul16.c | 91 const v128_t vprod0x01234567 = wasm_i16x8_mul(vi0x01234567, vk0x01234567); in xnn_qs8_dwconv_minmax_ukernel_up8x9__wasmsimd_mul16()
100 const v128_t vprod1x01234567 = wasm_i16x8_mul(vi1x01234567, vk1x01234567); in xnn_qs8_dwconv_minmax_ukernel_up8x9__wasmsimd_mul16()
109 const v128_t vprod2x01234567 = wasm_i16x8_mul(vi2x01234567, vk2x01234567); in xnn_qs8_dwconv_minmax_ukernel_up8x9__wasmsimd_mul16()
118 const v128_t vprod3x01234567 = wasm_i16x8_mul(vi3x01234567, vk3x01234567); in xnn_qs8_dwconv_minmax_ukernel_up8x9__wasmsimd_mul16()
127 const v128_t vprod4x01234567 = wasm_i16x8_mul(vi4x01234567, vk4x01234567); in xnn_qs8_dwconv_minmax_ukernel_up8x9__wasmsimd_mul16()
136 const v128_t vprod5x01234567 = wasm_i16x8_mul(vi5x01234567, vk5x01234567); in xnn_qs8_dwconv_minmax_ukernel_up8x9__wasmsimd_mul16()
145 const v128_t vprod6x01234567 = wasm_i16x8_mul(vi6x01234567, vk6x01234567); in xnn_qs8_dwconv_minmax_ukernel_up8x9__wasmsimd_mul16()
154 const v128_t vprod7x01234567 = wasm_i16x8_mul(vi7x01234567, vk7x01234567); in xnn_qs8_dwconv_minmax_ukernel_up8x9__wasmsimd_mul16()
163 const v128_t vprod8x01234567 = wasm_i16x8_mul(vi8x01234567, vk8x01234567); in xnn_qs8_dwconv_minmax_ukernel_up8x9__wasmsimd_mul16()
216 const v128_t vprod0x01234567 = wasm_i16x8_mul(vi0x01234567, vk0x01234567); in xnn_qs8_dwconv_minmax_ukernel_up8x9__wasmsimd_mul16()
[all …]
|
| /external/XNNPACK/src/qs8-gemm/gen/ |
| D | 3x4c8-xw-minmax-wasmsimd.c | 82 const v128_t vprod0x0 = wasm_i16x8_mul(vxa0, vxb0); in xnn_qs8_gemm_xw_minmax_ukernel_3x4c8__wasmsimd()
85 const v128_t vprod1x0 = wasm_i16x8_mul(vxa1, vxb0); in xnn_qs8_gemm_xw_minmax_ukernel_3x4c8__wasmsimd()
88 const v128_t vprod2x0 = wasm_i16x8_mul(vxa2, vxb0); in xnn_qs8_gemm_xw_minmax_ukernel_3x4c8__wasmsimd()
93 const v128_t vprod0x1 = wasm_i16x8_mul(vxa0, vxb1); in xnn_qs8_gemm_xw_minmax_ukernel_3x4c8__wasmsimd()
96 const v128_t vprod1x1 = wasm_i16x8_mul(vxa1, vxb1); in xnn_qs8_gemm_xw_minmax_ukernel_3x4c8__wasmsimd()
99 const v128_t vprod2x1 = wasm_i16x8_mul(vxa2, vxb1); in xnn_qs8_gemm_xw_minmax_ukernel_3x4c8__wasmsimd()
104 const v128_t vprod0x2 = wasm_i16x8_mul(vxa0, vxb2); in xnn_qs8_gemm_xw_minmax_ukernel_3x4c8__wasmsimd()
107 const v128_t vprod1x2 = wasm_i16x8_mul(vxa1, vxb2); in xnn_qs8_gemm_xw_minmax_ukernel_3x4c8__wasmsimd()
110 const v128_t vprod2x2 = wasm_i16x8_mul(vxa2, vxb2); in xnn_qs8_gemm_xw_minmax_ukernel_3x4c8__wasmsimd()
115 const v128_t vprod0x3 = wasm_i16x8_mul(vxa0, vxb3); in xnn_qs8_gemm_xw_minmax_ukernel_3x4c8__wasmsimd()
[all …]
|
| D | 3x4c8-minmax-wasmsimd-ld128.c | 84 const v128_t vprod0x0 = wasm_i16x8_mul(vxb0, vxa0); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld128()
86 const v128_t vprod1x0 = wasm_i16x8_mul(vxb0, vxa1); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld128()
88 const v128_t vprod2x0 = wasm_i16x8_mul(vxb0, vxa2); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld128()
91 const v128_t vprod0x1 = wasm_i16x8_mul(vxb1, vxa0); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld128()
94 const v128_t vprod1x1 = wasm_i16x8_mul(vxb1, vxa1); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld128()
97 const v128_t vprod2x1 = wasm_i16x8_mul(vxb1, vxa2); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld128()
108 const v128_t vprod0x2 = wasm_i16x8_mul(vxb2, vxa0); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld128()
110 const v128_t vprod1x2 = wasm_i16x8_mul(vxb2, vxa1); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld128()
112 const v128_t vprod2x2 = wasm_i16x8_mul(vxb2, vxa2); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld128()
115 const v128_t vprod0x3 = wasm_i16x8_mul(vxb3, vxa0); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld128()
[all …]
|
| D | 3x4c8-minmax-wasmsimd-ld64.c | 82 const v128_t vprod0x0 = wasm_i16x8_mul(vxa0, vxb0); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld64()
85 const v128_t vprod1x0 = wasm_i16x8_mul(vxa1, vxb0); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld64()
88 const v128_t vprod2x0 = wasm_i16x8_mul(vxa2, vxb0); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld64()
93 const v128_t vprod0x1 = wasm_i16x8_mul(vxa0, vxb1); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld64()
96 const v128_t vprod1x1 = wasm_i16x8_mul(vxa1, vxb1); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld64()
99 const v128_t vprod2x1 = wasm_i16x8_mul(vxa2, vxb1); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld64()
104 const v128_t vprod0x2 = wasm_i16x8_mul(vxa0, vxb2); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld64()
107 const v128_t vprod1x2 = wasm_i16x8_mul(vxa1, vxb2); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld64()
110 const v128_t vprod2x2 = wasm_i16x8_mul(vxa2, vxb2); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld64()
115 const v128_t vprod0x3 = wasm_i16x8_mul(vxa0, vxb3); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld64()
[all …]
|
| D | 2x4c8-xw-minmax-wasmsimd.c | 70 const v128_t vprod0x0 = wasm_i16x8_mul(vxa0, vxb0); in xnn_qs8_gemm_xw_minmax_ukernel_2x4c8__wasmsimd()
73 const v128_t vprod1x0 = wasm_i16x8_mul(vxa1, vxb0); in xnn_qs8_gemm_xw_minmax_ukernel_2x4c8__wasmsimd()
78 const v128_t vprod0x1 = wasm_i16x8_mul(vxa0, vxb1); in xnn_qs8_gemm_xw_minmax_ukernel_2x4c8__wasmsimd()
81 const v128_t vprod1x1 = wasm_i16x8_mul(vxa1, vxb1); in xnn_qs8_gemm_xw_minmax_ukernel_2x4c8__wasmsimd()
86 const v128_t vprod0x2 = wasm_i16x8_mul(vxa0, vxb2); in xnn_qs8_gemm_xw_minmax_ukernel_2x4c8__wasmsimd()
89 const v128_t vprod1x2 = wasm_i16x8_mul(vxa1, vxb2); in xnn_qs8_gemm_xw_minmax_ukernel_2x4c8__wasmsimd()
94 const v128_t vprod0x3 = wasm_i16x8_mul(vxa0, vxb3); in xnn_qs8_gemm_xw_minmax_ukernel_2x4c8__wasmsimd()
97 const v128_t vprod1x3 = wasm_i16x8_mul(vxa1, vxb3); in xnn_qs8_gemm_xw_minmax_ukernel_2x4c8__wasmsimd()
|
| D | 2x4c8-minmax-wasmsimd-ld128.c | 72 const v128_t vprod0x0 = wasm_i16x8_mul(vxb0, vxa0); in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld128()
74 const v128_t vprod1x0 = wasm_i16x8_mul(vxb0, vxa1); in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld128()
77 const v128_t vprod0x1 = wasm_i16x8_mul(vxb1, vxa0); in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld128()
80 const v128_t vprod1x1 = wasm_i16x8_mul(vxb1, vxa1); in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld128()
90 const v128_t vprod0x2 = wasm_i16x8_mul(vxb2, vxa0); in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld128()
92 const v128_t vprod1x2 = wasm_i16x8_mul(vxb2, vxa1); in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld128()
95 const v128_t vprod0x3 = wasm_i16x8_mul(vxb3, vxa0); in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld128()
98 const v128_t vprod1x3 = wasm_i16x8_mul(vxb3, vxa1); in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld128()
|
| D | 2x4c8-minmax-wasmsimd-ld64.c | 70 const v128_t vprod0x0 = wasm_i16x8_mul(vxa0, vxb0); in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld64()
73 const v128_t vprod1x0 = wasm_i16x8_mul(vxa1, vxb0); in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld64()
78 const v128_t vprod0x1 = wasm_i16x8_mul(vxa0, vxb1); in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld64()
81 const v128_t vprod1x1 = wasm_i16x8_mul(vxa1, vxb1); in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld64()
86 const v128_t vprod0x2 = wasm_i16x8_mul(vxa0, vxb2); in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld64()
89 const v128_t vprod1x2 = wasm_i16x8_mul(vxa1, vxb2); in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld64()
94 const v128_t vprod0x3 = wasm_i16x8_mul(vxa0, vxb3); in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld64()
97 const v128_t vprod1x3 = wasm_i16x8_mul(vxa1, vxb3); in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld64()
|
| D | 1x4c8-minmax-wasmsimd-ld64.c | 58 const v128_t vprod0x0 = wasm_i16x8_mul(vxa0, vxb0); in xnn_qs8_gemm_minmax_ukernel_1x4c8__wasmsimd_ld64()
63 const v128_t vprod0x1 = wasm_i16x8_mul(vxa0, vxb1); in xnn_qs8_gemm_minmax_ukernel_1x4c8__wasmsimd_ld64()
68 const v128_t vprod0x2 = wasm_i16x8_mul(vxa0, vxb2); in xnn_qs8_gemm_minmax_ukernel_1x4c8__wasmsimd_ld64()
73 const v128_t vprod0x3 = wasm_i16x8_mul(vxa0, vxb3); in xnn_qs8_gemm_minmax_ukernel_1x4c8__wasmsimd_ld64()
|
| D | 1x4c8-minmax-wasmsimd-ld128.c | 60 const v128_t vprod0x0 = wasm_i16x8_mul(vxb0, vxa0); in xnn_qs8_gemm_minmax_ukernel_1x4c8__wasmsimd_ld128()
63 const v128_t vprod0x1 = wasm_i16x8_mul(vxb1, vxa0); in xnn_qs8_gemm_minmax_ukernel_1x4c8__wasmsimd_ld128()
72 const v128_t vprod0x2 = wasm_i16x8_mul(vxb2, vxa0); in xnn_qs8_gemm_minmax_ukernel_1x4c8__wasmsimd_ld128()
75 const v128_t vprod0x3 = wasm_i16x8_mul(vxb3, vxa0); in xnn_qs8_gemm_minmax_ukernel_1x4c8__wasmsimd_ld128()
|
| D | 1x4c8-xw-minmax-wasmsimd.c | 58 const v128_t vprod0x0 = wasm_i16x8_mul(vxa0, vxb0); in xnn_qs8_gemm_xw_minmax_ukernel_1x4c8__wasmsimd()
63 const v128_t vprod0x1 = wasm_i16x8_mul(vxa0, vxb1); in xnn_qs8_gemm_xw_minmax_ukernel_1x4c8__wasmsimd()
68 const v128_t vprod0x2 = wasm_i16x8_mul(vxa0, vxb2); in xnn_qs8_gemm_xw_minmax_ukernel_1x4c8__wasmsimd()
73 const v128_t vprod0x3 = wasm_i16x8_mul(vxa0, vxb3); in xnn_qs8_gemm_xw_minmax_ukernel_1x4c8__wasmsimd()
|
| /external/XNNPACK/src/qs8-igemm/gen/ |
| D | 3x4c8-minmax-wasmsimd-ld128.c | 99 const v128_t vprod0x0 = wasm_i16x8_mul(vxb0, vxa0); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld128()
101 const v128_t vprod1x0 = wasm_i16x8_mul(vxb0, vxa1); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld128()
103 const v128_t vprod2x0 = wasm_i16x8_mul(vxb0, vxa2); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld128()
106 const v128_t vprod0x1 = wasm_i16x8_mul(vxb1, vxa0); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld128()
109 const v128_t vprod1x1 = wasm_i16x8_mul(vxb1, vxa1); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld128()
112 const v128_t vprod2x1 = wasm_i16x8_mul(vxb1, vxa2); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld128()
123 const v128_t vprod0x2 = wasm_i16x8_mul(vxb2, vxa0); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld128()
125 const v128_t vprod1x2 = wasm_i16x8_mul(vxb2, vxa1); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld128()
127 const v128_t vprod2x2 = wasm_i16x8_mul(vxb2, vxa2); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld128()
130 const v128_t vprod0x3 = wasm_i16x8_mul(vxb3, vxa0); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld128()
[all …]
|
| D | 3x4c8-minmax-wasmsimd-ld64.c | 97 const v128_t vprod0x0 = wasm_i16x8_mul(vxa0, vxb0); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld64()
100 const v128_t vprod1x0 = wasm_i16x8_mul(vxa1, vxb0); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld64()
103 const v128_t vprod2x0 = wasm_i16x8_mul(vxa2, vxb0); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld64()
108 const v128_t vprod0x1 = wasm_i16x8_mul(vxa0, vxb1); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld64()
111 const v128_t vprod1x1 = wasm_i16x8_mul(vxa1, vxb1); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld64()
114 const v128_t vprod2x1 = wasm_i16x8_mul(vxa2, vxb1); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld64()
119 const v128_t vprod0x2 = wasm_i16x8_mul(vxa0, vxb2); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld64()
122 const v128_t vprod1x2 = wasm_i16x8_mul(vxa1, vxb2); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld64()
125 const v128_t vprod2x2 = wasm_i16x8_mul(vxa2, vxb2); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld64()
130 const v128_t vprod0x3 = wasm_i16x8_mul(vxa0, vxb3); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld64()
[all …]
|
| D | 2x4c8-minmax-wasmsimd-ld128.c | 85 const v128_t vprod0x0 = wasm_i16x8_mul(vxb0, vxa0); in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld128()
87 const v128_t vprod1x0 = wasm_i16x8_mul(vxb0, vxa1); in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld128()
90 const v128_t vprod0x1 = wasm_i16x8_mul(vxb1, vxa0); in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld128()
93 const v128_t vprod1x1 = wasm_i16x8_mul(vxb1, vxa1); in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld128()
103 const v128_t vprod0x2 = wasm_i16x8_mul(vxb2, vxa0); in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld128()
105 const v128_t vprod1x2 = wasm_i16x8_mul(vxb2, vxa1); in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld128()
108 const v128_t vprod0x3 = wasm_i16x8_mul(vxb3, vxa0); in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld128()
111 const v128_t vprod1x3 = wasm_i16x8_mul(vxb3, vxa1); in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld128()
|
| D | 2x4c8-minmax-wasmsimd-ld64.c | 83 const v128_t vprod0x0 = wasm_i16x8_mul(vxa0, vxb0); in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld64()
86 const v128_t vprod1x0 = wasm_i16x8_mul(vxa1, vxb0); in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld64()
91 const v128_t vprod0x1 = wasm_i16x8_mul(vxa0, vxb1); in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld64()
94 const v128_t vprod1x1 = wasm_i16x8_mul(vxa1, vxb1); in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld64()
99 const v128_t vprod0x2 = wasm_i16x8_mul(vxa0, vxb2); in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld64()
102 const v128_t vprod1x2 = wasm_i16x8_mul(vxa1, vxb2); in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld64()
107 const v128_t vprod0x3 = wasm_i16x8_mul(vxa0, vxb3); in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld64()
110 const v128_t vprod1x3 = wasm_i16x8_mul(vxa1, vxb3); in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld64()
|
| D | 1x4c8-minmax-wasmsimd-ld64.c | 69 const v128_t vprod0x0 = wasm_i16x8_mul(vxa0, vxb0); in xnn_qs8_igemm_minmax_ukernel_1x4c8__wasmsimd_ld64()
74 const v128_t vprod0x1 = wasm_i16x8_mul(vxa0, vxb1); in xnn_qs8_igemm_minmax_ukernel_1x4c8__wasmsimd_ld64()
79 const v128_t vprod0x2 = wasm_i16x8_mul(vxa0, vxb2); in xnn_qs8_igemm_minmax_ukernel_1x4c8__wasmsimd_ld64()
84 const v128_t vprod0x3 = wasm_i16x8_mul(vxa0, vxb3); in xnn_qs8_igemm_minmax_ukernel_1x4c8__wasmsimd_ld64()
|
| D | 1x4c8-minmax-wasmsimd-ld128.c | 71 const v128_t vprod0x0 = wasm_i16x8_mul(vxb0, vxa0); in xnn_qs8_igemm_minmax_ukernel_1x4c8__wasmsimd_ld128()
74 const v128_t vprod0x1 = wasm_i16x8_mul(vxb1, vxa0); in xnn_qs8_igemm_minmax_ukernel_1x4c8__wasmsimd_ld128()
83 const v128_t vprod0x2 = wasm_i16x8_mul(vxb2, vxa0); in xnn_qs8_igemm_minmax_ukernel_1x4c8__wasmsimd_ld128()
86 const v128_t vprod0x3 = wasm_i16x8_mul(vxb3, vxa0); in xnn_qs8_igemm_minmax_ukernel_1x4c8__wasmsimd_ld128()
|
| /external/XNNPACK/src/qs8-gemm/ |
| D | MRx4c8-wasmsimd.c.in | 86 const v128_t vprod${M}x${N} = wasm_i16x8_mul(vxb${N}, vxa${M});
90 const v128_t vprod${M}x${N+1} = wasm_i16x8_mul(vxb${N+1}, vxa${M});
110 const v128_t vprod${M}x${N} = wasm_i16x8_mul(vxa${M}, vxb${N});
|
| /external/XNNPACK/src/qs8-igemm/ |
| D | MRx4c8-wasmsimd.c.in | 94 const v128_t vprod${M}x${N} = wasm_i16x8_mul(vxb${N}, vxa${M});
98 const v128_t vprod${M}x${N+1} = wasm_i16x8_mul(vxb${N+1}, vxa${M});
118 const v128_t vprod${M}x${N} = wasm_i16x8_mul(vxa${M}, vxb${N});
|
| /external/XNNPACK/src/qs8-dwconv/ |
| D | unipass-wasmsimd-mul16.c.in | 59 … const v128_t vprod${K}x${ABC[C:C+8]} = wasm_i16x8_mul(vi${K}x${ABC[C:C+8]}, vk${K}x${ABC[C:C+8]});
134 … const v128_t vprod${K}x${ABC[0:8]} = wasm_i16x8_mul(vi${K}x${ABC[0:8]}, vk${K}x${ABC[0:8]});
|
| /external/gemmlowp/fixedpoint/ |
| D | fixedpoint_wasmsimd.h | 114 return to_int16x8_v128_t(wasm_i16x8_mul(a.v, b.v));
|
| /external/llvm-project/clang/lib/Headers/ |
| D | wasm_simd128.h | 736 static __inline__ v128_t __DEFAULT_FN_ATTRS wasm_i16x8_mul(v128_t __a, in wasm_i16x8_mul() function
|