youmuliu/stgsharp
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Working on L4, now on swiss. FMA_TILE is re arranged.

Browse Source
This commit is contained in:
Youmu Liu
2026-03-03 18:02:31 +08:00
parent ae1a49d6fb
commit 91e136f81f
12 changed files with 1083 additions and 477 deletions
Download Patch File Download Diff File Expand all files Collapse all files
StgSharp.Native/sn-head/sn_intrinsic.h
+6 -6
View File
@@ -54,12 +54,12 @@ typedef union scalar_pack {
MAT_KERNEL(T) * mat_1; /* Offset 8: 2nd matrix kernel base */ \
MAT_KERNEL(T) * mat_2; /* Offset 16: 3rd matrix kernel base */ \
void *scalar; /* Offset 24: scalar pack */ \
int32_t i; /* Offset 32: row index of the result tile */ \
int32_t j; /* Offset 36: column index of the result tile */ \
int32_t k; /* Offset 40: column index of the result tile */ \
int32_t common_i; /* Offset 44: row index of the result tile */ \
int32_t common_j; /* Offset 48: column index of the result tile */ \
int32_t common_k; /* Offset 52: column index of the result tile */ \
int32_t x; /* Offset 32: row index of the result tile */ \
int32_t y; /* Offset 36: column index of the result tile */ \
int32_t z; /* Offset 40: column index of the result tile */ \
int32_t common_x; /* Offset 44: row index of the result tile */ \
int32_t common_y; /* Offset 48: column index of the result tile */ \
int32_t common_z; /* Offset 52: column index of the result tile */ \
uint32_t mask; /* Offset 56: mask for task schedualing, not useful in C side */ \
uint32_t reverse; /* Offset 60: reserved for future use */ \
scalar_pack padding; /* Offset 64: padding to make the struct size 128 bytes */ \