GROMACS version: 2026-rc
GROMACS modification: No
Simulation for mdrun is very slow. Using WSL without GPU support. Tips for troubleshooting?
Hi! What exactly do you mean by very slow? How big is the system, and what performance are you seeing? An obvious performance gain would be to use a GPU, if you have one available.
Hello, thank you for the reply. I have 500,000 steps and it moves an average of 100 steps every 20 seconds. I am trying to simulate a 1UN3 protein molecule. I am currently using an Acer swift 3. Using 1 MPI thread and Using 8 OpenMP threads. I am not sure if there is a GPU in-built and if I can activate it if there is.
Okay, that’s indeed very slow, even for running on a laptop. In general, some good information about the performance is reported by GROMACS at the end of the log file for a simulation. There, you can find the performance in terms of ns/day for example, and also a breakdown of the time spent in different parts of the code (under “Real cycle and time accounting”). You could check for example if there is an unusually high percentage of time unaccounted for (i.e. under “Rest”).
There seem to be different versions of your laptop model, both with AMD and Intel hardware, so I can’t give specific advice. One important thing to check though is if you might be using more OpenMP threads than your CPU has hardware threads. To be safe, you could try running with -ntmpi 1 -ntomp 4 to limit the number of threads you’re using.
Regardless of the hardware vendor, you’re laptop should have a GPU available. You can try configuring GROMACS with SYCL support to utilize it (check the installation instructions). However, your the performance you’re seeing is so low even for a CPU-only run that there’s likely something else going wrong.