Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[Train] Add env vars to enable Share AMD ROCR_VISIBLE_DEVICES #49346

Merged
merged 3 commits into from
Dec 19, 2024
+12 −0
Merged

[Train] Add env vars to enable Share AMD ROCR_VISIBLE_DEVICES #49346

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
9bade02
add env vars to enable AMD devices are shared
hongpeng-guo Dec 19, 2024
199f74e
Merge branch 'master' into hpguo/AMD_GPU_devices
hongpeng-guo Dec 19, 2024
be5ca42
update ROCM to ROCR
hongpeng-guo Dec 19, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions python/ray/_private/ray_constants.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -434,6 +434,7 @@ def env_set_by_user(key):
NOSET_CUDA_VISIBLE_DEVICES_ENV_VAR = "RAY_EXPERIMENTAL_NOSET_CUDA_VISIBLE_DEVICES"

CUDA_VISIBLE_DEVICES_ENV_VAR = "CUDA_VISIBLE_DEVICES"
ROCR_VISIBLE_DEVICES_ENV_VAR = "ROCR_VISIBLE_DEVICES"
NEURON_RT_VISIBLE_CORES_ENV_VAR = "NEURON_RT_VISIBLE_CORES"
TPU_VISIBLE_CHIPS_ENV_VAR = "TPU_VISIBLE_CHIPS"
NPU_RT_VISIBLE_DEVICES_ENV_VAR = "ASCEND_RT_VISIBLE_DEVICES"
Expand Down
7 changes: 7 additions & 0 deletions python/ray/train/_internal/backend_executor.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -27,6 +27,7 @@
ENABLE_SHARE_CUDA_VISIBLE_DEVICES_ENV,
ENABLE_SHARE_NEURON_CORES_ACCELERATOR_ENV,
ENABLE_SHARE_NPU_RT_VISIBLE_DEVICES_ENV,
ENABLE_SHARE_ROCR_VISIBLE_DEVICES_ENV,
RAY_TRAIN_ENABLE_STATE_TRACKING,
TRAIN_ENABLE_WORKER_SPREAD_ENV,
TRAIN_PLACEMENT_GROUP_TIMEOUT_S_ENV,
Expand Down Expand Up @@ -124,6 +125,12 @@ def __init__(
ENABLE_SHARE_NPU_RT_VISIBLE_DEVICES_ENV,
ray_constants.NPU_RT_VISIBLE_DEVICES_ENV_VAR,
),
# For AMD GPUs, they are using ROCR_VISIBLE_DEVICES env var.
ResourceConfig(
ray_constants.GPU,
ENABLE_SHARE_ROCR_VISIBLE_DEVICES_ENV,
ray_constants.ROCR_VISIBLE_DEVICES_ENV_VAR,
),
]

# Record the initialization time of BackendExecutor, which is
Expand Down
4 changes: 4 additions & 0 deletions python/ray/train/constants.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -56,6 +56,10 @@ def _get_ray_train_session_dir() -> str:
# Backend.share_cuda_visible_devices. 1 for True, 0 for False.
ENABLE_SHARE_CUDA_VISIBLE_DEVICES_ENV = "TRAIN_ENABLE_SHARE_CUDA_VISIBLE_DEVICES"

# Integer value which if set will not share ROCR accelerator visible devices
# across workers. 1 for True (default), 0 for False.
ENABLE_SHARE_ROCR_VISIBLE_DEVICES_ENV = "TRAIN_ENABLE_SHARE_ROCR_VISIBLE_DEVICES"

# Integer value which if set will not share neuron-core accelerator visible cores
# across workers. 1 for True (default), 0 for False.
ENABLE_SHARE_NEURON_CORES_ACCELERATOR_ENV = (
Expand Down
Loading