Leo Fang

`cuda.core.ObjectCode`: Support Tile IR

1

While we can't generate Tile IR yet (#1322), we can already load it from disk/memory: https://github.com/NVIDIA/cutile-python/blob/9db6438f2beea52f9ae3632c9a6815061eb1c761/cext/tile_kernel.cpp#L124-L151

Unlike in C++ I think it'd have to be an optional keyword argument, for a few reasons - To preserve backward compatibility (of both `cuda.core` and RMM Python) - To...

Everywhere in this repo is OSS and Apache-2.0 and can accept external contributions, except for the `cuda_bindings/` directory. We should run a preliminary scan (preferably without having to `/ok to...

Blocked by #718.

Address known thread-safety issues in `cuda.core`

2

> In general, `cuda.core` has a few internal, module-level states that should be guarded by a lock. We haven't done that extensively. _Originally posted by @leofang in https://github.com/NVIDIA/cuda-python/pull/1364#discussion_r2624683212_

WIP: Support passing SMV as a kernel argument of type `cuda::std::mdspan`

1

## Description Accompanying PR: https://github.com/cupy/cupy/pull/9525 Still WIP, lots to sort out. closes #748 ## Checklist - [ ] New or existing tests cover these changes. - [ ] The documentation...

RFC: Remove `StridedMemoryView` (to somewhere else)?

2

As part of https://github.com/NVIDIA/cuda-python/issues/674#issuecomment-3656821898 it was raised that our library design should strictly follow what the underlying CUDA driver/runtime/library APIs offer (**Update**: added "library" to cover other CTK libraries such...

Using #1311 as the playground, as of commit 4011bb8e54b57b6138ce8da809bca606be4b9b21 and CI logs at https://github.com/NVIDIA/cuda-python/actions/runs/19951181011 I verified that nv-gha-runners no longer makes containers as a hard requirement for running jobs on...