Compiling gromacs-2023.2 on MS azure cluster

User discussions
1

GROMACS version: 2023.2
GROMACS modification: No

Hi,
I just tried to install gmx 2023.3 on an MS azure cluster instance, as in:
cmake … -DGMX_BUILD_OWN_FFTW=ON -DGMX_GPU=CUDA -DCMAKE_INSTALL_PREFIX=/datadrive/local/gromacs-2023.2-bin

what I get is the error msg included below.
Any help would be highly appreciated!
cheers
Michael

CMake Error: The following variables are used in this project, but they are set to NOTFOUND.
Please set them or make sure they are set and tested correctly in the CMake files:
CUDA_cufft_LIBRARY (ADVANCED)
linked by target “testutils” in directory /datadrive/local/gromacs-2023.2/src/testutils
linked by target “libgromacs” in directory /datadrive/local/gromacs-2023.2/src/gromacs

CMake Warning at /datadrive/local/cmake-3.27.0-rc5-linux-x86_64/share/cmake-3.27/Modules/FindCUDA.cmake:2012 (add_library):
Cannot generate a safe runtime search path for target libgromacs because
files in some directories may conflict with libraries in implicit
directories:

runtime library [libgomp.so.1] in /usr/lib/gcc/x86_64-linux-gnu/9 may be hidden by files in:
  /opt/nvidia/hpc_sdk/Linux_x86_64/22.3/compilers/lib

Some of these libraries may not be found correctly.
Call Stack (most recent call first):
cmake/gmxManageNvccConfig.cmake:314 (cuda_add_library)
src/gromacs/CMakeLists.txt:180 (gmx_cuda_add_library)

the installed nvidia driver is: 470.103.01
CUDA Version: 11.4

2

Hi,

It appears that you are trying to to use the NVIDIA HPC SDK which is not officially supported in the current release (as a compiler). Can you try to use a standalone CUDA toolkit instead?

Cheers,
Szilárd