| 25 Nov 2025 |
 connor (he/him) | while building every combination of TensorRT 10.x and CUDA release on x86_64-linux to verify I could remove cudnn and nvrtc as buildInputs I prodded Claude to make the optuna+helion thing I mentioned earlier: https://github.com/ConnorBaker/helion/blob/claude/enhance-helion-autotuner-01V2JcV61tYZPi7y6cuZpE54/helion/autotuner/optuna_search.py
It doesn't work with the version of Helion we have in Nixpkgs because of some API changes to parallel_benchmarking but it seems like what I wanted. Haven't had the chance to mess with it though :l | 04:50:08 |
| connor (he/him) | https://github.com/NixOS/nixpkgs/pull/464947 | 15:01:41 |
| connor (he/him) | https://github.com/NixOS/nixpkgs/pull/464957 | 15:52:14 |
| connor (he/him) | https://github.com/NixOS/nixpkgs/pull/465047 | 20:23:29 |
 Gaétan Lepage | https://github.com/NixOS/nixpkgs/pull/450587 | 20:24:55 |
| 26 Nov 2025 |
| connor (he/him) | SomeoneSerge (back on matrix) I updated https://github.com/NixOS/nixpkgs/pull/459416 please merge | 02:57:01 |
| connor (he/him) | 🤦♂️ | 15:37:52 |
| connor (he/him) | https://github.com/onnx/onnx-tensorrt/pull/1043 moved from the pycuda to cuda python package and documented it nowhere I've been able to find | 15:38:34 |
| connor (he/him) | Kevin Mittman (UTC-7): the entry for --verbose seems to be for --no-dry-run, which doesn't have an entry: https://docs.nvidia.com/cuda/cuda-compiler-driver-nvcc/#verbose-v | 18:12:52 |
| connor (he/him) | well, pycuda's tests exposed a few issues with how NVCC works
notably that it dies if GCC isn't on the path -- it doesn't even use the path modifications from nvcc.profile | 18:38:12 |
| connor (he/him) | so this is now a larger PR https://github.com/NixOS/nixpkgs/pull/465047 | 18:38:20 |
| connor (he/him) | I'm going to split out the NVCC changes to a different PR and make pycuda a stacked PR | 18:51:51 |
| connor (he/him) | :F | 18:51:53 |
| connor (he/him) | https://github.com/NixOS/nixpkgs/pull/465329 | 18:57:02 |
| 28 Nov 2025 |
 yorik.sar | Finally got around to doing this: https://github.com/NixOS/nixos-homepage/pull/1903 - connor (burnt/out) (UTC-8) SomeoneSerge (back on matrix)Gaétan Lepage feel free to approve :) | 15:01:30 |
| yorik.sar | I guess I’ll make another one for nixos-cuda.org :) | 15:05:22 |
| yorik.sar | Here it is: https://github.com/nixos-cuda/nixos-cuda.org/pull/2 | 15:12:12 |
| 29 Nov 2025 |
 matthewcroughan | Is there anybody that knows how you add cuda stuff to a rust build? | 01:54:22 |
| matthewcroughan | I have a rust build that is silently doing nothing when cuda isn't present, and not indicating in logs when it is being used or isn't | 01:54:37 |
| matthewcroughan | Leads to https://github.com/alvr-org/ALVR/issues/2792 | 01:55:02 |
 SomeoneSerge (back on matrix) | idk is it even about cuda? Any failing dlopens? @matthewcroughan:defenestrate.it: | 04:30:26 |
| matthewcroughan | I'm not sure, what do you think based on the issue? | 16:12:46 |
| matthewcroughan | Apparently people can get it to work by steam-run'ing the stuff they provide via github releases | 16:13:09 |
| SomeoneSerge (back on matrix) | Id_debuuuuuuuuuuuug | 16:49:33 |
| matthewcroughan | Yeah I can also get it working by steam-run'ing the github release | 17:01:30 |
| matthewcroughan | compiling onnxruntime takes forever though | 17:01:35 |
| matthewcroughan | so I'm doing that whilst compiling what I think could be a fix | 17:01:42 |
| 1 Dec 2025 |
| connor (he/him) | Make sure you’re setting config.cudaCapabilities explicitly to speed up compilation | 12:39:34 |
| connor (he/him) | Packaging onnx/onnx-tensorrt/onnxruntime and getting them working for a range of cuda/cudnn/tensorrt releases is maddening oh my god:
- Fix for GCC 11/NVCC's template processing https://github.com/nixos-cuda/cuda-legacy/commit/79c3026f1a761125b6faa609c5f9e56e723a4568
- https://github.com/nixos-cuda/cuda-legacy/issues/6 since ONNX Runtime needs GCC > 11.1 and I need it to build with 11.4
- I'm fairly certain I managed to get a build failure from ONNX Runtime with CUDA 11.8 because a reference to
cuda_nvcc was held onto, so ONNX Runtime failed the build
| 21:32:54 |
| 2 Dec 2025 |
| connor (he/him) | Okay NVCC from CUDA 11.4 still dies horrifically when using newer versions of glibc with GCC 11 so that’ll stay on GCC 10.
Also it seems onnxruntime 1.22.2 needs CCCL newer than what 11.4 provides, so supporting onnxruntime on 11.4 means finding an older version which works :l | 01:21:58 |
