Searched refs:vec_idx (Results 1 – 3 of 3) sorted by relevance
| /petsc/src/mat/impls/aij/seq/aijperm/ |
| H A D | aijperm.c | 286 __m256i vec_idx, vec_ipos, vec_j; in MatMult_SeqAIJPERM() local 381 vec_idx = _mm256_loadu_si256((__m256i const *)&aj[ipos]); in MatMult_SeqAIJPERM() 383 vec_x = _mm512_i32gather_pd(vec_idx, x, _MM_SCALE_8); in MatMult_SeqAIJPERM() 389 vec_idx = _mm256_loadu_si256((__m256i const *)&aj[ipos]); in MatMult_SeqAIJPERM() 391 vec_x = _mm512_mask_i32gather_pd(vec_x, mask, vec_idx, x, _MM_SCALE_8); in MatMult_SeqAIJPERM() 418 vec_idx = _mm256_i32gather_epi32(aj, vec_ipos, _MM_SCALE_4); in MatMult_SeqAIJPERM() 420 vec_x = _mm512_i32gather_pd(vec_idx, x, _MM_SCALE_8); in MatMult_SeqAIJPERM()
|
| /petsc/src/mat/impls/sell/seq/ |
| H A D | sell.c | 29 #define AVX512_Mult_Private(vec_idx, vec_x, vec_vals, vec_y) \ argument 31 vec_idx = _mm256_loadu_si256((__m256i const *)acolidx); \ 33 vec_x = _mm512_i32gather_pd(vec_idx, x, _MM_SCALE_8); \ 36 #define AVX2_Mult_Private(vec_idx, vec_x, vec_vals, vec_y) \ argument 38 vec_idx = _mm_loadu_si128((__m128i const *)acolidx); /* SSE2 */ \ 39 vec_x = _mm256_i32gather_pd(x, vec_idx, _MM_SCALE_8); \ 331 __m256i vec_idx; in MatMult_SeqSELL() local 336 __m128i vec_idx; in MatMult_SeqSELL() local 371 AVX512_Mult_Private(vec_idx, vec_x, vec_vals, vec_y); in MatMult_SeqSELL() 383 AVX512_Mult_Private(vec_idx, vec_x, vec_vals, vec_y); in MatMult_SeqSELL() [all …]
|
| /petsc/src/mat/impls/aij/seq/ |
| H A D | aij.h | 627 __m256i vec_idx; in PetscSparseDensePlusDot_AVX512_Private() local 632 vec_idx = _mm256_loadu_si256((__m256i const *)aj); in PetscSparseDensePlusDot_AVX512_Private() 634 vec_x = _mm512_i32gather_pd(vec_idx, x, _MM_SCALE_8); in PetscSparseDensePlusDot_AVX512_Private() 644 vec_idx = _mm256_mask_loadu_epi32(vec_idx, mask, aj); in PetscSparseDensePlusDot_AVX512_Private() 646 vec_x = _mm512_mask_i32gather_pd(vec_x, mask, vec_idx, x, _MM_SCALE_8); in PetscSparseDensePlusDot_AVX512_Private()
|