Eric Shi

Looks like related changes were merged in a810b842fe82d3e0aa99871ca4f75df9ba892b4e a long time ago. Thanks for the suggestions!

I'll also mention that the [nvmath-python](https://github.com/NVIDIA/nvmath-python/) library already exposes cuDSS bindings: https://docs.nvidia.com/cuda/nvmath-python/latest/bindings/cudss.html but I haven't tried them myself.

Hi, I think there are some small modifications I can make to the build script to support your use case better, but in the meantime you can give this a...

Hey @phphuc612, I looked at the Packman docs some more and found something that should work unlike my original comment. In this approach, I have a directory on my computer...

Added in eddb998a01a55e711d692a4a62003f18f238bd31

@etaoxing Thanks for reporting the bug. @AnkaChan made a fix for this incorporating your suggestions: 77c3bda1e051e853c1e7012e3668184dfb7c97c8 Please give it a try (should already be in the nightly build), thanks!

64b85f91c6ea75723c867bba5cb4b083653ac8fa reduced the time to run the test by about 300 secs Leaving a note for the future: On my system there were some remaining modules that took a long...

> For reference the same kernel takes about 5s to compile on my machine (CUDA 13.0 as well, but sm_86) > > ``` > Module __main__._test_cell_lookup__itp_domainDefGeo_NodalField_Tet_0f54a1c3 2b967c1 load on device...

@daedalus5 ran into it during a merge request today, so the failure is rare but present on our CI/CD ``` ====================================================================== test_tile_cholesky_cholesky_multiple_rhs_cpu (warp.tests.tile.test_tile_cholesky.TestTileCholesky.test_tile_cholesky_cholesky_multiple_rhs_cpu) ---------------------------------------------------------------------- Traceback (most recent call last): File...

Second case has been fixed by @eric-heiden in 3c6388890e7a18c0164504b68b53ccb60859cbc0