floraachy/OpenBLAS
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[WIP] Refactor the driver code for direct SGEMM (#2782)

Browse Source 
Move "direct SGEMM" functionality out of the SkylakeX SGEMM kernel and make it available
(on x86_64 targets only for now) in DYNAMIC_ARCH builds
* Add  sgemm_direct targets in the kernel Makefile.L3 and CMakeLists.txt
* Add direct_sgemm functions to the gotoblas struct in common_param.h
* Move sgemm_direct_performant helper to separate file
* Update gemm.c  to macros for sgemm_direct to support dynamic_arch naming via common_s,h
* (Conditionally) add sgemm_direct functions in setparam-ref.c
This commit is contained in:
Martin Kroeker
2020-08-19 14:51:09 +02:00
committed by GitHub
parent 2c72972570
commit 75eeb265d7
10 changed files with 107 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
common_param.h
View File

@@ -175,6 +175,11 @@ BLASLONG (*ismin_k) (BLASLONG, float *, BLASLONG);
int (*ssymv_L) (BLASLONG, BLASLONG, float, float *, BLASLONG, float *, BLASLONG, float *, BLASLONG, float *);
int (*ssymv_U) (BLASLONG, BLASLONG, float, float *, BLASLONG, float *, BLASLONG, float *, BLASLONG, float *);
#ifdef ARCH_X86_64
void (*sgemm_direct) (BLASLONG, BLASLONG, BLASLONG, float *, BLASLONG , float *, BLASLONG , float * , BLASLONG);
int (*sgemm_direct_performant) (BLASLONG M, BLASLONG N, BLASLONG K);
#endif
int (*sgemm_kernel )(BLASLONG, BLASLONG, BLASLONG, float, float *, float *, float *, BLASLONG);
int (*sgemm_beta )(BLASLONG, BLASLONG, BLASLONG, float, float *, BLASLONG, float *, BLASLONG, float *, BLASLONG);