Leo Fang
Q: What is the current error/traceback if the underlying CTK does not support an option? I assume it says "invalid option"?
Hi @trixirt I am from NVIDIA and our team owns CCCL & CUDA Python. CCCL is a single project under which Thrust/CUB/libcudacxx are unified. We encourage you to just consider...
I left my view from the array API & CUDA perspective in https://github.com/data-apis/array-api/issues/654#issuecomment-1731612375. From the CuPy perspective, not taking a CuPy ndarray for `repeats` may be a bit counterintuitive, I'd...
We can't upload it to PyPI yet because of the size limit (https://github.com/pypi/support/issues/3904). I can check with @ethanhs next week and see if we can use pypi.nvidia.com as a temporary...
Sorry for my late response. I did discuss with the cuSPARSELt team and they have reasons to not publish to pypi.nvidia.com. Fortunately this is moot now that the size-bump request...
Update: We've released cuSPARSELt v0.6.2 wheels on PyPI (https://pypi.org/project/nvidia-cusparselt-cu12/), would you be able to give it a try @gdaisukesuzuki?
@Kairoven We have discussed with the NVIDIA RAPIDS team. We've noticed this account and several others have jointly started opening a large quantity of issues in the last 24-48 hours....
blocked by https://github.com/prefix-dev/pixi/issues/4139 and https://github.com/prefix-dev/pixi/issues/4303
@kmaehashi let me know what you think!
Blocked by #162. (For some reason GitHub created the issues in the wrong order...)