| /external/XNNPACK/src/qs8-gemm/gen/ |
| D | 3x8c8-minmax-avx2.c | 131 const __m256i vpermute_mask = _mm256_set_epi32(7, 3, 6, 2, 5, 1, 4, 0); in xnn_qs8_gemm_minmax_ukernel_3x8c8__avx2() local
132 __m256i vacc0x01234567 = _mm256_permutevar8x32_epi32(vacc0x02461357, vpermute_mask); in xnn_qs8_gemm_minmax_ukernel_3x8c8__avx2()
133 __m256i vacc1x01234567 = _mm256_permutevar8x32_epi32(vacc1x02461357, vpermute_mask); in xnn_qs8_gemm_minmax_ukernel_3x8c8__avx2()
134 __m256i vacc2x01234567 = _mm256_permutevar8x32_epi32(vacc2x02461357, vpermute_mask); in xnn_qs8_gemm_minmax_ukernel_3x8c8__avx2()
|
| D | 3x8c8-xw-minmax-avx2.c | 127 const __m256i vpermute_mask = _mm256_set_epi32(7, 3, 6, 2, 5, 1, 4, 0); in xnn_qs8_gemm_xw_minmax_ukernel_3x8c8__avx2() local
128 __m256i vacc0x01234567 = _mm256_permutevar8x32_epi32(vacc0x02461357, vpermute_mask); in xnn_qs8_gemm_xw_minmax_ukernel_3x8c8__avx2()
129 __m256i vacc1x01234567 = _mm256_permutevar8x32_epi32(vacc1x02461357, vpermute_mask); in xnn_qs8_gemm_xw_minmax_ukernel_3x8c8__avx2()
130 __m256i vacc2x01234567 = _mm256_permutevar8x32_epi32(vacc2x02461357, vpermute_mask); in xnn_qs8_gemm_xw_minmax_ukernel_3x8c8__avx2()
|
| D | 2x8c8-xw-minmax-avx2.c | 107 const __m256i vpermute_mask = _mm256_set_epi32(7, 3, 6, 2, 5, 1, 4, 0); in xnn_qs8_gemm_xw_minmax_ukernel_2x8c8__avx2() local
108 __m256i vacc0x01234567 = _mm256_permutevar8x32_epi32(vacc0x02461357, vpermute_mask); in xnn_qs8_gemm_xw_minmax_ukernel_2x8c8__avx2()
109 __m256i vacc1x01234567 = _mm256_permutevar8x32_epi32(vacc1x02461357, vpermute_mask); in xnn_qs8_gemm_xw_minmax_ukernel_2x8c8__avx2()
|
| D | 2x8c8-minmax-avx2.c | 111 const __m256i vpermute_mask = _mm256_set_epi32(7, 3, 6, 2, 5, 1, 4, 0); in xnn_qs8_gemm_minmax_ukernel_2x8c8__avx2() local
112 __m256i vacc0x01234567 = _mm256_permutevar8x32_epi32(vacc0x02461357, vpermute_mask); in xnn_qs8_gemm_minmax_ukernel_2x8c8__avx2()
113 __m256i vacc1x01234567 = _mm256_permutevar8x32_epi32(vacc1x02461357, vpermute_mask); in xnn_qs8_gemm_minmax_ukernel_2x8c8__avx2()
|
| D | 1x8c8-xw-minmax-avx2.c | 87 const __m256i vpermute_mask = _mm256_set_epi32(7, 3, 6, 2, 5, 1, 4, 0); in xnn_qs8_gemm_xw_minmax_ukernel_1x8c8__avx2() local
88 __m256i vacc0x01234567 = _mm256_permutevar8x32_epi32(vacc0x02461357, vpermute_mask); in xnn_qs8_gemm_xw_minmax_ukernel_1x8c8__avx2()
|
| D | 1x8c8-minmax-avx2.c | 91 const __m256i vpermute_mask = _mm256_set_epi32(7, 3, 6, 2, 5, 1, 4, 0); in xnn_qs8_gemm_minmax_ukernel_1x8c8__avx2() local
92 __m256i vacc0x01234567 = _mm256_permutevar8x32_epi32(vacc0x02461357, vpermute_mask); in xnn_qs8_gemm_minmax_ukernel_1x8c8__avx2()
|
| /external/XNNPACK/src/qs8-igemm/gen/ |
| D | 3x8c8-minmax-avx2.c | 148 const __m256i vpermute_mask = _mm256_set_epi32(7, 3, 6, 2, 5, 1, 4, 0); in xnn_qs8_igemm_minmax_ukernel_3x8c8__avx2() local
149 __m256i vacc0x01234567 = _mm256_permutevar8x32_epi32(vacc0x02461357, vpermute_mask); in xnn_qs8_igemm_minmax_ukernel_3x8c8__avx2()
150 __m256i vacc1x01234567 = _mm256_permutevar8x32_epi32(vacc1x02461357, vpermute_mask); in xnn_qs8_igemm_minmax_ukernel_3x8c8__avx2()
151 __m256i vacc2x01234567 = _mm256_permutevar8x32_epi32(vacc2x02461357, vpermute_mask); in xnn_qs8_igemm_minmax_ukernel_3x8c8__avx2()
|
| D | 2x8c8-minmax-avx2.c | 126 const __m256i vpermute_mask = _mm256_set_epi32(7, 3, 6, 2, 5, 1, 4, 0); in xnn_qs8_igemm_minmax_ukernel_2x8c8__avx2() local
127 __m256i vacc0x01234567 = _mm256_permutevar8x32_epi32(vacc0x02461357, vpermute_mask); in xnn_qs8_igemm_minmax_ukernel_2x8c8__avx2()
128 __m256i vacc1x01234567 = _mm256_permutevar8x32_epi32(vacc1x02461357, vpermute_mask); in xnn_qs8_igemm_minmax_ukernel_2x8c8__avx2()
|
| D | 1x8c8-minmax-avx2.c | 104 const __m256i vpermute_mask = _mm256_set_epi32(7, 3, 6, 2, 5, 1, 4, 0); in xnn_qs8_igemm_minmax_ukernel_1x8c8__avx2() local
105 __m256i vacc0x01234567 = _mm256_permutevar8x32_epi32(vacc0x02461357, vpermute_mask); in xnn_qs8_igemm_minmax_ukernel_1x8c8__avx2()
|
| /external/XNNPACK/src/qs8-igemm/ |
| D | MRx8c8-avx2.c.in | 110 const __m256i vpermute_mask = _mm256_set_epi32(7, 3, 6, 2, 5, 1, 4, 0); local
112 __m256i vacc${M}x01234567 = _mm256_permutevar8x32_epi32(vacc${M}x02461357, vpermute_mask);
|
| /external/XNNPACK/src/qs8-gemm/ |
| D | MRx8c8-avx2.c.in | 111 const __m256i vpermute_mask = _mm256_set_epi32(7, 3, 6, 2, 5, 1, 4, 0);
113 __m256i vacc${M}x01234567 = _mm256_permutevar8x32_epi32(vacc${M}x02461357, vpermute_mask);
|
| /external/XNNPACK/src/qs8-dwconv/gen/ |
| D | up32x9-minmax-avx512skx-mul32.c | 42 const __m256i vpermute_mask = _mm256_set_epi32(7, 3, 5, 1, 6, 2, 4, 0); in xnn_qs8_dwconv_minmax_ukernel_up32x9__avx512skx_mul32() local
219 …HIJKLMNOPQRSTUV = _mm256_permutevar8x32_epi32(vout0123GHIJ89ABOPQR4567KLMNCDEFSTUV, vpermute_mask); in xnn_qs8_dwconv_minmax_ukernel_up32x9__avx512skx_mul32()
|
| /external/XNNPACK/src/qs8-dwconv/ |
| D | unipass-avx512skx-mul32.c.in | 43 const __m256i vpermute_mask = _mm256_set_epi32(7, 3, 5, 1, 6, 2, 4, 0); local
126 …{ABC[C+24:C+28]}${ABC[C+4:C+8]}${ABC[C+20:C+24]}${ABC[C+12:C+16]}${ABC[C+28:C+32]}, vpermute_mask);
|