<div>Hi Jörg,</div>This problem seems unrelated to CP2K...<div>Have you tried to google that message?</div><div>I found some entries:</div><div><br></div><div>https://stackoverflow.com/questions/41409842/ubuntu-16-04-cuda-8-cuda-driver-version-is-insufficient-for-cuda-runtime-vers<br></div><div>https://stackoverflow.com/questions/52346957/cudagetdevice-failed-status-cuda-driver-version-is-insufficient-for-cuda-run<br></div><div>https://forums.developer.nvidia.com/t/error-cuda-driver-version-is-insufficient-for-cuda-runtime-version/66006<br></div><div><br></div><div>Any many others.... It seems a problem with the driver installation, people recommend to uninstall/reinstall...</div><div><br></div><div>Alfio</div><div><br><br></div><div class="gmail_quote"><div dir="auto" class="gmail_attr">Il giorno lunedì 2 novembre 2020 alle 00:23:32 UTC+1 sassy ha scritto:<br/></div><blockquote class="gmail_quote" style="margin: 0 0 0 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;">Dear all,
<br>
<br>over the weekend I had access to a GPU test machine so I could compare the
<br>performance between a RTX-6000 and a A100 GPU card.
<br>
<br>I have installed the following CUDA drivers:
<br>$ nvidia-smi
<br>NVIDIA-SMI 450.80.02 Driver Version: 450.80.02 CUDA Version: 11.0
<br>
<br>I am using GCC-9.3.0 with CUDA-11.0.2
<br>
<br>I am using the toolchain with the following commands:
<br>
<br>$ ./install_cp2k_toolchain.sh -j 12 --no-check-certificate --libint-lmax=6 --
<br>with-cmake=system --with-libxc --with-libint --with-fftw=system --with-
<br>openblas=system --with-libxsmm --with-parmetis --with-superlu --with-quip --
<br>with-plumed --with-gsl --with-libvdwxc --with-spglib --with-hdf5 --with-spfft
<br>--with-scalapack=system --gpu-ver=V100 --with-elpa=no --with-sirius=no
<br>
<br>(I had problems with the ELPA and SIRIUS build, hence I disabled them).
<br>
<br>Running the test jobs, *all* of them fail with:
<br>
<br>"CUDA driver version is insufficient for CUDA runtime version"
<br>
<br>The box has both cards installed and I am masking one with the
<br>CUDA_VISIBLE_DEVICES
<br>environment variable.
<br>
<br>I am puzzled of where I am getting wrong. Yesterday I tried it with both, the
<br>A100 and the RTX-6000 card using most of the above modules being installed on
<br>the system via EasyBuild, today I only tested the RTX-6000 card.
<br>
<br>Does anybody has a good idea? I don't have the box for much longer, as I said,
<br>it is only a test environment I got access granted to.
<br>
<br>Thanks for your help!
<br>
<br>All the best from a mild London
<br>
<br>Jörg
<br>
<br>
<br>
<br></blockquote></div>