Vladislav Zhurba
Vladislav Zhurba
I see that you're trying to interop between Driver's `cuda.CUdeviceptr` type and the Runtime's `cudart.cudaMemset` API. While there already exists some interop in Runtime, this does looks like a good...
Resolved with latest [CUDA Python 12.5.0](https://github.com/NVIDIA/cuda-python/releases/tag/v12.5.0) release. Closing.
I don't think the file had called it out explicitly before. I've now added it to requirements.txt if platform_system is Windows. This change will get picked up in the next...
Fixed with [12.6](https://github.com/NVIDIA/cuda-python/releases/tag/v12.6.0) release, closing.
Here's my investigation report. **Version 1.** This is a minimized version of `cuCtxGetDevice` to judge the speed of light. CUdevice is created by user and we skip returning the error...
> just to create a tuple Here's one more version: **Version 5.** No tuples, but create a new CUdevice ``` def cuCtxGetDevice(): cdef CUdevice device = CUdevice() err = cydriver.cuCtxGetDevice(device._pvt_ptr)...
I can apply the fix in autogen and push a commit to this PR.
I'll need https://github.com/NVIDIA/cuda-python/pull/544 to be merged first for this to be a clean commit though.
The only useful docs from 11.8.x are the API References and I can see the benefit of adding a versioned doc like `11.8.5` so that these become web accessible. Backporting...
Fixed with [12.6](https://github.com/NVIDIA/cuda-python/releases/tag/v12.6.0) release, closing.