Closed ViktorM closed 1 year ago
Well, the message Loaded runtime CuDNN library: 8.3.2
tells you the problem: JAX was trying to load CuDNN, but found a really old version (8.3) when it tried.
Usually this means that you have already loaded an older CuDNN into your process (e.g., importing PyTorch before importing JAX is one common way this can happen, since PyTorch usually bundles an older CuDNN). The other way it can happen is that an older CuDNN is first in your LD_LIBRARY_PATH
, but I think that's unlikely to be the case here since you used the pip
installation of JAX.
Try searching for other CuDNN installations (files name *cudnn*
) on your system?
This is probably not something we can fix from the JAX end.
same issue, it seems because the current version of pytorch does not support cudnn8.8 or higher version
@chaojiewang94 You might do well to install a CPU-only version of PyTorch, if the goal is to use that in the context of a GPU-using JAX program.
Description
pip install --upgrade "jax[cuda12_pip]" -f https://storage.googleapis.com/jax-releases/jax_cuda_releases.html
jax installation using the command above is not working with the latest Brax: https://github.com/google/brax
For some reason it complains on runtime CuDNN version:
What jax/jaxlib version are you using?
jax 0.4.8
Which accelerator(s) are you using?
RTX 4090
Additional system info
Python 3.9 Ubuntu 22.04
NVIDIA GPU info
+---------------------------------------------------------------------------------------+ | NVIDIA-SMI 530.30.02 Driver Version: 530.30.02 CUDA Version: 12.1 | |-----------------------------------------+----------------------+----------------------+ | GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC | | Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. | | | | MIG M. | |=========================================+======================+======================| | 0 NVIDIA GeForce RTX 4090 On | 00000000:01:00.0 Off | Off | | 30% 44C P8 29W / 450W| 1535MiB / 24564MiB | 7% Default | | | | N/A | +-----------------------------------------+----------------------+----------------------+
+---------------------------------------------------------------------------------------+ | Processes: | | GPU GI CI PID Type Process name GPU Memory | | ID ID Usage | |=======================================================================================| | 0 N/A N/A 1285367 G ...viktorm/Downloads/Telegram/Telegram 72MiB | | 0 N/A N/A 1288565 G /usr/lib/xorg/Xorg 861MiB | | 0 N/A N/A 1288676 G /usr/bin/gnome-shell 99MiB | | 0 N/A N/A 1289827 G ...AAAAAAAACAAAAAAAAAA= --shared-files 86MiB | | 0 N/A N/A 1290862 G ...893162563,891623453882694089,131072 412MiB | +---------------------------------------------------------------------------------------+