-
-
Notifications
You must be signed in to change notification settings - Fork 12.7k
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.
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
cudaPackages: keep unsupported attributes, mark unsupported, explain why #266475
Comments
Is this currently blocking any packages/PRs? It looks like #266115 has already been merged |
No, not blocking anything. Just had this thought since I’ve seen this pattern around as I’ve been rewriting |
More pros of
|
I just had
torch /torchvision doesn't actually use tensorrt by default.
I'm pretty sure |
Some more reflection on the TRT situation: it's different from We only enable With TRT, there are cudatoolkit releases for which TRT isn't a thing (yet). We couldn't even assign the broken/fake derivation any meaningful version. However, I still don't feel like omitting the attribute is a good thing, because it breaks eval so easily. If we leave out |
A fair number of tools we use in
cudaPackages
depend on specific versions of CUDA. Sometimes this is a strict dependency due to API changes or other breakages; other times, it's merely because it is not known whether the code is forward (or backward!) compatible.The approach that I (@ConnorBaker) have taken so far in
cudaPackages
has been that when a package is unavailable for some release of CUDA, or is known to be incompatible, it is not present in that CUDA release's package set (e.g.,cudaPackages_12_2
).However, this limits the flexibility of downstream users and avoids a larger pattern: attempting to use newer or older versions of dependencies.
@SomeoneSerge raised this issue in a comment on a PR about
cuda-samples
, which we use as a sort of sanity-check, allowing us to make sure that thecudatoolkit
builds: #266115 (comment)For context:
cudaPackages.cuda-samples
has a release for each recent release of CUDA except for 11.7. I made the decision to excludecuda-samples
fromcudaPackages_11_7
be wrapping it inoptionalAttrs
.This issue proposes two mechanisms in abstract:
cuda-samples
release (11.6) withcudaPackages_11_7
cuda-samples
release (11.8) withcudaPackages_11_7
cudaPackages
even if compatibility is not guaranteed by NVIDIA's support matrixTaken together, these mechanisms would allow users to attempt to build things which would otherwise not be possible (e.g., using a newer release of CUDNN which might support a specific version of CUDA, even if it is not listed in the support matrix).
It's worth keeping in mind that support matrices (specifically as we're talking about them here, with respect to NVIDIA's software) are essentially guaranteeing that the listed combinations work. It does not guarantee that combinations outside those enumerated will be broken, only that they are unsupported.
@SomeoneSerge @samuela thoughts?
The text was updated successfully, but these errors were encountered: