[AMD][CI] Fix test_custom_allreduce for A100 testgroup (#34735)
Signed-off-by: Ryan Rock <ryan.rock@amd.com>
This commit is contained in:
@@ -33,6 +33,7 @@ def graph_allreduce(
|
|||||||
):
|
):
|
||||||
with monkeypatch.context() as m:
|
with monkeypatch.context() as m:
|
||||||
m.delenv("CUDA_VISIBLE_DEVICES", raising=False)
|
m.delenv("CUDA_VISIBLE_DEVICES", raising=False)
|
||||||
|
m.delenv("HIP_VISIBLE_DEVICES", raising=False)
|
||||||
device = torch.device(f"cuda:{rank}")
|
device = torch.device(f"cuda:{rank}")
|
||||||
torch.cuda.set_device(device)
|
torch.cuda.set_device(device)
|
||||||
init_test_distributed_environment(tp_size, pp_size, rank, distributed_init_port)
|
init_test_distributed_environment(tp_size, pp_size, rank, distributed_init_port)
|
||||||
@@ -92,6 +93,7 @@ def eager_allreduce(
|
|||||||
):
|
):
|
||||||
with monkeypatch.context() as m:
|
with monkeypatch.context() as m:
|
||||||
m.delenv("CUDA_VISIBLE_DEVICES", raising=False)
|
m.delenv("CUDA_VISIBLE_DEVICES", raising=False)
|
||||||
|
m.delenv("HIP_VISIBLE_DEVICES", raising=False)
|
||||||
device = torch.device(f"cuda:{rank}")
|
device = torch.device(f"cuda:{rank}")
|
||||||
torch.cuda.set_device(device)
|
torch.cuda.set_device(device)
|
||||||
init_test_distributed_environment(tp_size, pp_size, rank, distributed_init_port)
|
init_test_distributed_environment(tp_size, pp_size, rank, distributed_init_port)
|
||||||
|
|||||||
Reference in New Issue
Block a user