Supported Pytorch added for Maxwell.

[plushplank]

Description

Simplified my previous PR and added cu126 Pytorch installation for Maxwell GPUs dynamically as cu128 is unsupported. No other code is touched.

--no-cache-dir is used for my variable to bypass memory restrictions.

Screenshots/videos:

Checklist:

• I have read contributing wiki page
• I have performed a self-review of my own code
• My code follows the style guidelines
• My code passes tests

[w-e-w]

I made a PR as well

• use cu126 for 10 series and older GPUs #17172
  I've only saw your PR when I was posting my PR

---

--no-cache-dir is used for my variable to bypass memory restrictions.

what memory restrictions your system RAM?
how much do you have?

[plushplank]

I made a PR as well

• use cu126 for 10 series and older GPUs #17172
  I've only saw your PR when I was posting my PR

--no-cache-dir is used for my variable to bypass memory restrictions.

what memory restrictions your system RAM? how much do you have?

16GB DDR3

[plushplank]

I made a PR as well

• use cu126 for 10 series and older GPUs #17172
  I've only saw your PR when I was posting my PR

--no-cache-dir is used for my variable to bypass memory restrictions.

what memory restrictions your system RAM? how much do you have?

I can edit my code to include pascal as well:

def cu126():
"""For Maxwell GPUs"""
cc = get_cuda_comp_cap()
if 5.0 <= cc < 7.0:
torch_command = os.environ.get('TORCH_COMMAND', f"pip install --no-cache-dir torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu126")
return torch_command

return None

def prepare_environment():
torch_index_url = os.environ.get('TORCH_INDEX_URL', "https://download.pytorch.org/whl/cu128")
torch_command = os.environ.get('TORCH_COMMAND', f"pip install torch==2.7.0 torchvision==0.22.0 --extra-index-url {torch_index_url}")
torch_command = cu126() or os.environ.get('TORCH_COMMAND', f"pip install torch==2.7.0 torchvision==0.22.0 --extra-index-url {torch_index_url}")

[plushplank]

Closed for a pr exists by @w-e-w.