Dear all,
I've been having trouble installing MELD on my workstation for a few days. I'm having a problem with CUDA compatibility. I'm on ubuntu 22.04 with CUDA 12.0. My graphic card is a NVIDIA GeForce RTX 3080, driver 525.60.3/NVIDIA Container Toolkit CLI version 1.13.4/Docker version 24.0.4).
Indeed, I read on the github that the version taken into account by MELD is only prior to CUDA 11.2, but this version is not available for ubuntu 22.04. Are there any solutions for using MELD on recent versions of ubuntu? Also is it absolutely necessary to have two or more GPUs to run MELD ?
Dear all, I've been having trouble installing MELD on my workstation for a few days. I'm having a problem with CUDA compatibility. I'm on ubuntu 22.04 with CUDA 12.0. My graphic card is a NVIDIA GeForce RTX 3080, driver 525.60.3/NVIDIA Container Toolkit CLI version 1.13.4/Docker version 24.0.4).
Indeed, I read on the github that the version taken into account by MELD is only prior to CUDA 11.2, but this version is not available for ubuntu 22.04. Are there any solutions for using MELD on recent versions of ubuntu? Also is it absolutely necessary to have two or more GPUs to run MELD ?
sudo docker run --rm --runtime=nvidia --gpus all nvidia/cuda:11.6.2-base-ubuntu20.04 nvidia-smi this command works but this one : from https://catalog.ngc.nvidia.com/orgs/hpc/containers/meld to test our setup ,
sudo docker run -it --gpus all nvcr.io/hpc/meld:200930-0.4.15 python -m simtk.testInstallation
return : OpenMM Version: 7.4.2 Git Revision: Unknown There are 3 Platforms available: 1 Reference - Successfully computed forces 2 CPU - Successfully computed forces 3 CUDA - Error computing forces with CUDA platform CUDA platform error: Error launching CUDA compiler: 256
nvcc fatal : Value 'sm_86' is not defined for option 'gpu-architecture'
Median difference in forces between platforms:
Reference vs. CPU: 1.9668e-06
I hope someone can help me solve this problem. Thanks. Damien