qsim
qsim copied to clipboard
quantumlib
[WIP] Include GPU in wheels
Fixes #430.
Since cross-platform support requires building on the manylinux image, installing NVCC on the runner is the best we can do unless we decide to push a manylinux + CUDA image to Docker Hub.
Bazel page is returning 404s, need to wait for it to come back and retry.
An inevitable side effect of building GPU support on machines that lack GPU support: we have no way to test GPU behavior in continuous integration. However, since the wheels generated by GHA are accessible from the test artifacts, it will be possible for us to spin up a GPU-enabled machine, copy over the wheels, and test before release.
If we move forward with this, the process outlined above will become a mandatory part of all GPU changes.
@95-martin-orion @sergeisakov FYI, we prepared a demo to show how to build a project upon CUDA/cuQuantum wheels: https://github.com/NVIDIA/cuQuantum/tree/main/extra/demo_build_with_wheels Let us know if it helps!
