We are using GPU on our AWS RHEL 7 platform.
We are facing an issue - unable to load library ‘libcuda.so.1’.
We are new to GPUs. We have Tesla K80 on AWS EC2 RHEL 7, which we want to run in a container.
Nvidia-smi output is working fine: NVIDIA-SMI (screenshot below)
Driver version = 470.94
CUDA version = 11.4
We have tried to install nvidia-docker, nvidia-container-toolkit and tried few hacks as per documentation online. However, we are unable to resolve this issue.
We want to use it with MATLAB’s parallel computing toolbox. At this point, it would be great if we can get some support from NVIDIA directly.
This is bit urgent for us.
Thanks
libcuda (the cuda driver) usually gets installed alongside the driver, how did you install it?