[gfx950] Make bypassing __threadfence the default for multinode. (#1947)
* Gate based on ROCM version, safe for ROCm 7.0.2 and beyond.
* Updates naming to gfx9CheapFenceOff since we use this for gfx942 and gfx950. Thanks Nilesh.
* Add info logging statement to NCCL_INIT to print whether enabled when INFO logging is enabled.
[ROCm/rccl commit: c70f5b4621]
Dieser Commit ist enthalten in:
committet von
GitHub
Ursprung
6d151d4e21
Commit
455d516dc4
@@ -875,7 +875,7 @@ public:
|
||||
patBarrier();
|
||||
}
|
||||
if(collWork){
|
||||
skip_fence = !collWork -> gfx942CheapFenceOff;
|
||||
skip_fence = !collWork -> gfx9CheapFenceOff;
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
In neuem Issue referenzieren
Einen Benutzer sperren