If you run molecular dynamics simulations with GROMACS on your local machine, you’ve probably wondered how to get the best performance without overloading your system. The GROMACS Wizard in SAMSON includes a performance settings panel that helps you customize how GROMACS makes use of your computer’s hardware — with surprising control and clarity.
This blog post introduces a lesser-known but highly useful part of the GROMACS Wizard: local performance tuning through thread control and runtime parameters.
Why Performance Tuning Matters 🧪
Molecular modelers working on complex systems often perform simulations that run for hours, or even days. Performance tuning can speed up simulations, improve stability, and prevent your OS from becoming unresponsive by managing how many threads are used during simulations.
The GROMACS team states that their software already tries to make good use of your hardware by default, but there is still room for manual optimization — especially when you are juggling multiple applications or need repeatable performance benchmarks for publications or reproducibility.
Automatic Thread Control (with a Safety Net)
When you install GROMACS through the GROMACS Wizard, the system is conservative by design: it sets the number of threads to a value lower than the number of logical cores on your system. This prevents the simulation from monopolizing your CPU and keeps your OS responsive.
If you want more control, here’s the behavior based on your input:
- 0 threads: GROMACS guesses the maximum number of available threads.
- n threads < available cores: Uses
-pin onto pin threads to specific cores, helping performance and reducing interruptions.
This parameter can be changed using the Settings button at the top of the GROMACS Wizard:
Going Deeper: Additional Parameters
Advanced users might want to modify ntmpi (MPI threads), ntomp (OpenMP threads), or other options like pme (handling of Particle Mesh Ewald), or maxh (maximum hours of a run).
Although these parameters aren’t always necessary, setting them explicitly can make your runs more predictable and efficient — especially on shared environments or when managing multiple datasets in succession.
Note that these performance settings only apply to local jobs. For cloud jobs, SAMSON handles optimization differently and user-specified performance parameters are not used.
Start Simple, Iterate, and Monitor
If you’re new to tuning, start by running a small simulation with the default settings. Then increase the number of threads incrementally while observing CPU usage and total computation time. Avoid setting the number of threads too close to the number of available logical cores, especially if you’ll be using your machine for other tasks simultaneously.
For more information on the available parameters and advanced optimization strategies, the GROMACS Manual provides an in-depth guide.
To learn more about performance tuning and other capabilities of the GROMACS Wizard in SAMSON, check out the complete documentation: https://documentation.samson-connect.net/tutorials/gromacs-wizard/settings/.
SAMSON and all SAMSON Extensions are free for non-commercial use. You can download SAMSON at https://www.samson-connect.net.