Use Windows C: drive for Conda & building

[jakirkham]

Checklist

• Used a personal fork of the feedstock to propose changes
• Bumped the build number (if the version is unchanged)
• Reset the build number to 0 (if the version changed)
• Re-rendered with the latest conda-smithy (Use the phrase @conda-forge-admin, please rerender in a comment in this PR for automated rerendering)
• Ensured the license file is being packaged.

[conda-forge-admin]

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipe/meta.yaml) and found it was in an excellent condition.

I do have some suggestions for making it better though...

For recipe/meta.yaml:

• ℹ️ The recipe is not parsable by parser conda-recipe-manager. The recipe can only be automatically migrated to the new v1 format if it is parseable by conda-recipe-manager.

_{This message was generated by GitHub Actions workflow run https://github.com/conda-forge/conda-forge-webservices/actions/runs/13611228201. Examine the logs at this URL for more detail.}

[jakirkham]

Exploring changing Windows to use C: drive to help with memory usage

However we are running out of disk space on CI. Snippet of the failure below:

CONDA_BLD_PATH=C:\\bld\\; Setting pagefile size to 16GB on D:
Parameters:
8589934592
17179869184
D:
PageFile: 8589934592 / 17179869184 bytes for \Device\HarddiskVolume3\pagefile.sys
Could Not Find C:\Windows\System32\msmpi.dll
Could Not Find C:\Windows\System32\msmpires.dll
Downloading CUDA version 11.8 installer from https://developer.download.nvidia.com/compute/cuda/11.8.0/local_installers/cuda_11.8.0_522.06_windows.exe
Expected MD5: 894c61ba173d26dc667e95ee734d3c5a
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed

  0     0    0     0    0     0      0      0 --:--:-- --:--:-- --:--:--     0
  4 3062M    4  148M    0     0   226M      0  0:00:13 --:--:--  0:00:13  226M
 13 3062M   13  401M    0     0   242M      0  0:00:12  0:00:01  0:00:11  242M
 21 3062M   21  644M    0     0   242M      0  0:00:12  0:00:02  0:00:10  242M
 29 3062M   29  888M    0     0   243M      0  0:00:12  0:00:03  0:00:09  243M
 37 3062M   37 1144M    0     0   245M      0  0:00:12  0:00:04  0:00:08  245M
 45 3062M   45 1393M    0     0   246M      0  0:00:12  0:00:05  0:00:07  249M
 53 3062M   53 1652M    0     0   248M      0  0:00:12  0:00:06  0:00:06  250M
 62 3062M   62 1903M    0     0   248M      0  0:00:12  0:00:07  0:00:05  251M
 70 3062M   70 2153M    0     0   248M      0  0:00:12  0:00:08  0:00:04  252M
 78 3062M   78 2414M    0     0   249M      0  0:00:12  0:00:09  0:00:03  253M
 85 3062M   85 2612M    0     0   245M      0  0:00:12  0:00:10  0:00:02  243M
 93 3062M   93 2868M    0     0   245M      0  0:00:12  0:00:11  0:00:01  242M
 94 3062M   94 2886M    0     0   245M      0  0:00:12  0:00:11  0:00:01  240M
curl: (23) Failure writing output to destination
Problem downloading installer...
Could not install CUDA

As the Windows install of the CUDA 11 Toolkit downloads it in the current working directory, it lands on the D: drive (as that's where repos are cloned). The pagefile is also set to the opposite drive as CONDA_BLD_PATH, which means D: is used here. As D: is already a smaller drive, having the CUDA Toolkit installer and the pagefile on D: runs out of space. Thus the build crashes

Think the fact the CUDA 11 Toolkit is downloaded onto D: is just incidental. Given the CUDA 11 Toolkit is installed on C:, it would make more sense to download it there. Plus we can probably cutdown size by using the network installer

Trying these changes in PR: conda-forge/conda-forge-ci-setup-feedstock#380