[rocm][ray] Fix: Unify Ray device visibility handling across CUDA and ROCm (#33308)
Signed-off-by: Kourosh Hakhamaneshi <kourosh@anyscale.com>
This commit is contained in:
committed by
GitHub
parent
192ad4648b
commit
2f6d17cb2f
@@ -60,6 +60,11 @@ def test_ray_runtime_env(monkeypatch: pytest.MonkeyPatch):
|
||||
runtime_env = {
|
||||
"env_vars": {
|
||||
"TEST_ENV_VAR": "test_value",
|
||||
# In future ray versions, this will be default, so when setting a
|
||||
# task or actor with num_gpus=None/0, the visible devices env var
|
||||
# won't be overridden resulting in no GPUs being visible on a gpu
|
||||
# machine.
|
||||
"RAY_ACCEL_ENV_VAR_OVERRIDE_ON_ZERO": "0",
|
||||
},
|
||||
}
|
||||
|
||||
|
||||
Reference in New Issue
Block a user