| 2 Oct 2024 |
 Kevin Mittman (UTC-7) | In reply to @ss:someonex.net
Kevin Mittman Hi! Do you know how dcgm uses cuda and why it has to link several versions? See libdcgm_cublas_proxy${cudaMajor}.so | 18:34:06 |
| Kevin Mittman (UTC-7) | In reply to @connorbaker:matrix.org
Kevin Mittman: does NVIDIA happen to have JSON (or otherwise structured) versions of their dependency constraints for packages somewhere, or are the tables on the docs for each respective package the only source? I'm working on update scripts and I'd like to avoid the manual stage of "go look on the website, find the table (it may have moved), and encode the contents as a Nix expression" Not really ... wishlist for future. Which product / component is this? | 18:46:04 |
| Kevin Mittman (UTC-7) | SomeoneSerge (utc+3): seems like reply got stuck in a thread | 18:46:54 |
| 3 Oct 2024 |
 connor (he/him) | In reply to @justbrowsing:matrix.org
Not really ... wishlist for future. Which product / component is this? That particular request was born out of frustration with TensorRT.
Any idea why the support matrix for TensorRT says only CUDNN 8.9.7 is supported (https://docs.nvidia.com/deeplearning/tensorrt/support-matrix/index.html) but the 24.09 container is shipping it with CUDNN 9.4 (https://docs.nvidia.com/deeplearning/frameworks/support-matrix/index.html)? | 05:21:35 |
| Kevin Mittman (UTC-7) |
- TRT 8.x depends on cuDNN 8.x (last release was 8.9.7)
- TRT 10.x has optional support for cuDNN (not updated for 9.x)
- The DL frameworks container image is more generic
| 17:09:43 |
| 4 Oct 2024 |
| connor (he/him) | I know that when packaging TRT (any version) for Nixpkgs, it autopatchelf flags a dependency on cuDNN, so we need to link against it.
Does TRT 10.x not work with cuDNN 9.x at all, or is it not an officially supported combination?
onnxruntime, for example, says for CUDA 11.8 to use TRT 10.x with cuDNN 8.9.x, and with CUDA 12.x to use TRT 10.x with cuDNN 9.x. The latter combination wasn’t in the support matrix, so I was surprised.
For the DL frameworks container, does that mean TRT comes without support for cuDNN since it’s not an 8.9.x release, that it’s not officially supported (per the TRT support matrix), or something else? | 15:21:59 |
