Sylvain Jeaugey fda4362c9e Fix potential deadlock during init in multi-thread mode. ...

Make sure all calls calling cudaMalloc (including devCommSetup) are
called before the last bootstrapBarrier. That way, we avoid calls to
cudaMalloc be blocked by a NCCL kernel launched on another GPU by
another thread which completed init faster.

Resolve #623.


[ROCm/rccl commit: ecab28a7c9]

2022-09-26 02:13:10 -07:00

projects/rccl

Fix potential deadlock during init in multi-thread mode.

2022-09-26 02:13:10 -07:00

S

Description

No description provided

282 MiB

Languages

C++ 67.5%

C 20.6%

Python 6.6%

CMake 3.4%

Shell 0.6%

Other 1.1%