Uses torch.version.cuda to compile CUDA extensions

[syed-ahmed]

Fixes: #2152

Tested this by installing a cpu only version of PyTorch and then running pip install -e . -vvv. "PyTorch GPU support is not available. Skipping compilation of CUDA extensions" is in the logs, confirming that the previous behavior is preserved.

Using pip 25.1 from /usr/local/lib/python3.12/dist-packages/pip (python 3.12)                 
Non-user install because site-packages writeable                                              
Created temporary directory: /tmp/pip-build-tracker-a0n8qmb2                                  
Initialized build tracking at /tmp/pip-build-tracker-a0n8qmb2                                 
Created build tracker: /tmp/pip-build-tracker-a0n8qmb2                                        
Entered build tracker: /tmp/pip-build-tracker-a0n8qmb2                                        
Created temporary directory: /tmp/pip-install-tn_td1mv                                        
Created temporary directory: /tmp/pip-ephem-wheel-cache-iflb4k6r                              
Obtaining file:///workspace/ao                                                                
  Added file:///workspace/ao to build tracker '/tmp/pip-build-tracker-a0n8qmb2'               
  Running setup.py (path:/workspace/ao/setup.py) egg_info for package from file:///workspace/a
o                                                                                             
  Created temporary directory: /tmp/pip-pip-egg-info-j4p5drqz                                 
  Running command python setup.py egg_info                                                    
  INFO:root:running egg_info                                                                  
  INFO:root:creating /tmp/pip-pip-egg-info-j4p5drqz/torchao.egg-info                          
  INFO:root:writing /tmp/pip-pip-egg-info-j4p5drqz/torchao.egg-info/PKG-INFO                  
  INFO:root:writing dependency_links to /tmp/pip-pip-egg-info-j4p5drqz/torchao.egg-info/depend
ency_links.txt                                                                                
  INFO:root:writing requirements to /tmp/pip-pip-egg-info-j4p5drqz/torchao.egg-info/requires.t
xt                                                                                            
  INFO:root:writing top-level names to /tmp/pip-pip-egg-info-j4p5drqz/torchao.egg-info/top_lev
el.txt                                                                                        
  INFO:root:writing manifest file '/tmp/pip-pip-egg-info-j4p5drqz/torchao.egg-info/SOURCES.txt
'                                                                                             
  INFO:root:reading manifest file '/tmp/pip-pip-egg-info-j4p5drqz/torchao.egg-info/SOURCES.txt
'                                                                                             
  INFO:root:adding license file 'LICENSE'                                                     
  INFO:root:writing manifest file '/tmp/pip-pip-egg-info-j4p5drqz/torchao.egg-info/SOURCES.txt
'                                                                                             
  PyTorch GPU support is not available. Skipping compilation of CUDA extensions

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/ao/2163

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❗ 1 Active SEVs

There are 1 currently active SEVs. If your PR is affected, please view them below:

• CI workflows being skipped on PR

✅ No Failures

As of commit 18ad551 with merge base 43a65e2 ():
💚 Looks good so far! There are no failures yet. 💚

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[msaroufim]

Should be fine to merge if CI is green, in the meantime there does seem to be a lint issue, we use ruff

[syed-ahmed]

Thanks @msaroufim for the lint!