Leo Fang
Feel free to open a PR, Vyas! It should be fine to remove those for which we've raised warnings for years, especially it should be safe in v14. We do...
IIRC Vyas wants to keep this issue open. Let me do it.
Yes of course. Feel free to give it a shot.
(FWIW this is the reason why we want to have a general solution #451 🙂)
To make the situation worse, CUDA 11 and CUDA 12 conda packages have different layouts: https://github.com/conda-forge/cudatoolkit-feedstock/blob/0f01f6edfe22e77ac068daa46505711a1481272e/recipe/meta.yaml#L681-L683 (see docs in https://github.com/conda-forge/cuda-feedstock/tree/main/recipe for more info on the "new" layout)
Among the 3.4 us run time, a rough capture of major components is - calling CuPy's `__dlpack_device__()` and `__dlpack__()`: 1.23 us - constructing Python objects like shape, strides, and dtype:...
cc @NaderAlAwar for vis
It'd be interesting to see if lazily populating the attributes of `StridedMemoryView` could help reduce the construction time. A lot of times we don't need the full information, just pointer...
Hi @ricor07 👋 Thank you for your interest in contributing to the repo! Unfortunately -- I hate to say this -- while this repository is "source open," the current (NVIDIA,...
@copilot please work on this issue. We need all markdown files under each module's docs/source/ directory to be converted to ReST. If there is a way to enforce that new...