Revert disabling of context tracking for Radeon (#1927)

* Revert disabling of context tracking for Radeon

Original commit df3b7e47
 `Disable context tracking for the current version. (#1839)`

* Add env variable for disabling of context tracking for Radeon

`export NCCL_DISABLE_CONTEXT_TRACKING=1` to force disable of context tracking

* Update docs/how-to/rccl-usage-tips.rst

Fix grammar, thanks @amd-jnovotny

Co-authored-by: Jeffrey Novotny <jnovotny@amd.com>

* Rename NCCL_DISABLE_CONTEXT_TRACKING -> RCCL_DISABLE_CONTEXT_TRACKING

* Revert changes in includes and rename util function

---------

Co-authored-by: Jeffrey Novotny <jnovotny@amd.com>

[ROCm/rccl commit: 07925ec027]
Esse commit está contido em:
Artem Kuzmitckii
2025-09-27 21:19:50 +02:00
commit de GitHub
commit 722b0cd579
4 arquivos alterados com 37 adições e 3 exclusões
+14
Ver Arquivo
@@ -8,6 +8,8 @@
#include "core.h"
#include "nvmlwrap.h"
#include "archinfo.h"
#include <unistd.h>
#include <stdlib.h>
@@ -191,6 +193,18 @@ bool matchIfList(const char* string, int port, struct netIf* ifList, int listSiz
return false;
}
RCCL_PARAM(DisableContextTracking, "DISABLE_CONTEXT_TRACKING", 0);
bool rcclNeedEnableContextTrack(int cuDeviceId) {
hipDeviceProp_t devProp;
if (rcclParamDisableContextTracking() == 1)
return false;
if (hipGetDeviceProperties(&devProp, cuDeviceId) != 0)
return false;
return IsArchMatch(devProp.gcnArchName,"gfx11")
|| IsArchMatch(devProp.gcnArchName,"gfx12")
|| IsArchMatch(devProp.gcnArchName,"gfx10");
}
__thread struct ncclThreadSignal ncclThreadSignalLocalInstance = ncclThreadSignalStaticInitializer();
void* ncclMemoryStack::allocateSpilled(struct ncclMemoryStack* me, size_t size, size_t align) {