Known issues affecting users of GROMACS¶
Here is a non-exhaustive list of issues that are we are aware of that are affecting regular users of GROMACS.
Unable to compile with CUDA 11.3¶
Due to a bug in the nvcc compiler, it is currently not possible to compile NVIDIA GPU-enabled GROMACS with version 11.3 of the CUDA compiler. We recommend using CUDA 11.4 or newer.
Verlet buffer underestimated for inhomogeneous systems¶
The current Verlet buffer estimation code assumes that the density in the system is uniform. This leads to an underestimate of the buffer for strongly inhomogeneous systems. The temporary solution to this is to lower the verlet-buffer-tolerance parameter value by the factor between the uniform density and the local density. In the 2023 release this correction will be performed automatically.
Verlet buffer underestimated when using only r^-12 potentials¶
When only the repulsive part of the Lennard-Jones potential is used, as can be the case in coarse-grained systems, the Verlet buffer can be underestimated due to the extremely non-linear nature of the r^-12 potential. A temporary solution is to decrease the verlet-buffer-tolerance until you get a non-zero Verlet buffer. This issue will be fixed in the 2023 release.
Build is fragile with gcc 7 and CUDA¶
Different forms of gcc 7 have different behaviour when compiling test programs with nvcc. This prevents GROMACS from reliably testing compilation flags for use with nvcc. So in this case we use flags unilaterally and this could lead to compilation errors. The best way to avoid these potential problems is to use a more recent version of gcc.
SYCL build unstable when using oneAPI with LevelZero backend¶
There are multiple issues with different versions of Intel oneAPI when using the LevelZero backend.
In many cases, it works fine, and if it fails, it does so explicitly (either crash or hang), so it should be fine to experiment with.
For most cases, we recommend using OpenCL backend (the default) when running SYCL build of GROMACS on Intel GPUs.