Fix pk_int4 cast and add pk_int4 dtype in ck tile (#1854)

* Fix pk_int4 cast and add pk_int4 dtype in ck tile

* fixes

* Improvements

* fix typo
This commit is contained in:
Bartłomiej Kocot
2025-02-04 10:32:07 +01:00
committed by GitHub
parent 9c5b2f3936
commit 9ee69dd297
12 changed files with 406 additions and 73 deletions

View File

@@ -144,6 +144,10 @@
#define CK_TILE_USE_AMD_BUFFER_ATOMIC_ADD_INTEGER 1
#endif
#ifndef CK_TILE_USE_PK4_LAYOUT_SHUFFLE
#define CK_TILE_USE_PK4_LAYOUT_SHUFFLE 1
#endif
// buffer atomic add: floating point
#ifndef __HIP_DEVICE_COMPILE__ // for host code
#define CK_TILE_USE_AMD_BUFFER_ATOMIC_ADD_FLOAT 1