diff options
| author | Alex James <git@alextjam.es> | 2024-11-27 11:54:29 -0600 |
|---|---|---|
| committer | Alex James <git@alextjam.es> | 2024-11-27 11:54:29 -0600 |
| commit | 4f6eafbb37cb0434f9cf310d96509f04ba4f634b (patch) | |
| tree | a9084a5af3643fa861bd8fc65d2502026ee1b8df /pkgs/development/python-modules/termplotlib/gnuplot-subprocess.patch | |
| parent | a3a67865fbbd5309e4f90d48b68a05da26434e18 (diff) | |
tiny-cuda-nn: fix a couple of build issues
The CUDA architecture detection code in tiny-cuda-nn's Torch bindings
convert CUDA capabilities to `int`s, which breaks with 9.0a. Filter
9.0a from the CUDA capabilities passed to TCNN_CUDA_ARCHITECTURES. Also
patch the Torch bindings to replace the hardcoded `-std=c++14` argument
to nvcc with `-std=c++17` for CUDA 11+ to fix the build.
Diffstat (limited to 'pkgs/development/python-modules/termplotlib/gnuplot-subprocess.patch')
0 files changed, 0 insertions, 0 deletions