Skip to content

use cu126 for 10 series and older GPUs #17172

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

Open
w-e-w wants to merge 1 commit into
base: dev
Choose a base branch
from
Open

use cu126 for 10 series and older GPUs #17172

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
15 changes: 14 additions & 1 deletion modules/launch_utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -347,8 +347,21 @@ def early_access_blackwell_wheels():
return f'pip install {ea_whl.get(sys.version_info.minor)}'


def get_default_torch_index_url():
"""Choose default torch index url based on GPU CUDA Compute Capability (CC)
Use cu126 for CC <= 7.2, cu128 for CC > 7.2
PyTorch have dropped support for older GPUs on their cu128 and above wheels,
For Nvidia 10 series and older GPUs (CC < 7.0) should use cu126 wheels
Since GPUs with CC <= 7.2 are considered legacy by Nvidia
ref 2025-11-07 https://developer.nvidia.com/cuda-legacy-gpus
"""
if get_cuda_comp_cap() <= 7.2:
return "https://download.pytorch.org/whl/cu126"
return "https://download.pytorch.org/whl/cu128"


def prepare_environment():
torch_index_url = os.environ.get('TORCH_INDEX_URL', "https://download.pytorch.org/whl/cu128")
torch_index_url = os.environ.get('TORCH_INDEX_URL', get_default_torch_index_url())
torch_command = os.environ.get('TORCH_COMMAND', f"pip install torch==2.7.0 torchvision==0.22.0 --extra-index-url {torch_index_url}")
if args.use_ipex:
if platform.system() == "Windows":
Expand Down