Better L4 and array storage design

This commit is contained in:
Youmu Liu
2026-03-04 18:12:58 +08:00
parent 7e970352ab
commit dfa6f4ba92
103 changed files with 853 additions and 6837 deletions
@@ -182,15 +182,52 @@ SN_MK_PROC_DECL_STD(double, avx, , fma);
SN_MK_PROC_DECL_STD(double, avx, _fma, fma);
SN_MK_PROC_DECL_STD(double, 512, , fma);
SN_MK_PROC_DECL_STD(float, sse, , tiled_fma);
SN_MK_PROC_DECL_STD(float, avx, , tiled_fma);
SN_MK_PROC_DECL_STD(float, avx, _fma, tiled_fma);
SN_MK_PROC_DECL_STD(float, 512, , tiled_fma);
SN_SK_PROC_DECL_STD(float, avx, , add);
SN_SK_PROC_DECL_STD(float, 512, , add);
SN_MK_PROC_DECL_STD(double, sse, , tiled_fma);
SN_MK_PROC_DECL_STD(double, avx, , tiled_fma);
SN_MK_PROC_DECL_STD(double, avx, _fma, tiled_fma);
SN_MK_PROC_DECL_STD(double, 512, , tiled_fma);
SN_SK_PROC_DECL_STD(double, sse, , add);
SN_SK_PROC_DECL_STD(double, avx, , add);
SN_SK_PROC_DECL_STD(double, 512, , add);
SN_SK_PROC_DECL_STD(float, sse, , sub);
SN_SK_PROC_DECL_STD(float, avx, , sub);
SN_SK_PROC_DECL_STD(float, 512, , sub);
SN_SK_PROC_DECL_STD(double, sse, , sub);
SN_SK_PROC_DECL_STD(double, avx, , sub);
SN_SK_PROC_DECL_STD(double, 512, , sub);
SN_SK_PROC_DECL_STD(float, sse, , transpose);
SN_SK_PROC_DECL_STD(float, avx, , transpose);
SN_SK_PROC_DECL_STD(float, 512, , transpose);
SN_SK_PROC_DECL_STD(float, sse, , scalar_mul);
SN_SK_PROC_DECL_STD(float, avx, , scalar_mul);
SN_SK_PROC_DECL_STD(float, 512, , scalar_mul);
SN_SK_PROC_DECL_STD(double, sse, , scalar_mul);
SN_SK_PROC_DECL_STD(double, avx, , scalar_mul);
SN_SK_PROC_DECL_STD(double, 512, , scalar_mul);
SN_SK_PROC_DECL_STD(float, sse, , fill);
SN_SK_PROC_DECL_STD(float, avx, , fill);
SN_SK_PROC_DECL_STD(float, 512, , fill);
SN_SK_PROC_DECL_STD(double, sse, , fill);
SN_SK_PROC_DECL_STD(double, avx, , fill);
SN_SK_PROC_DECL_STD(double, 512, , fill);
SN_SK_PROC_DECL_STD(float, sse, , quality);
SN_SK_PROC_DECL_STD(float, sse, , fma);
SN_SK_PROC_DECL_STD(float, avx, , fma);
SN_SK_PROC_DECL_STD(float, avx, _fma, fma);
SN_SK_PROC_DECL_STD(float, 512, , fma);
SN_SK_PROC_DECL_STD(double, sse, , fma);
SN_SK_PROC_DECL_STD(double, avx, , fma);
SN_SK_PROC_DECL_STD(double, avx, _fma, fma);
SN_SK_PROC_DECL_STD(double, 512, , fma);
#pragma endregion