Hi Aditya,
Thank you for your suggestion! Having a simple CPU/GPU selection option would indeed be convenient.
Actually, GROMACS Wizard already allows you to specify GPU usage through its Settings, although you first need to have a locally installed version of GROMACS built with GPU support, as mentioned above.
Once you have such a version:
Open GROMACS Wizard > Settings, enable Use a different GROMACS version, and specify the paths to your GROMACS executable and force fields.
In the same Settings window, you can specify Additional performance parameters that will be passed to gmx mdrun. For example:
-nb gpu — run short-range nonbonded calculations on the GPU.
-pme gpu — run PME calculations on the GPU.
-gpu_id 0 — select GPU 0.
You can then run your simulations locally with GPU acceleration.
Note that GROMACS built with GPU/CUDA support can also automatically detect and use compatible GPUs, so explicitly specifying these parameters is not always necessary. However, -nb gpu is useful if you want to explicitly request GPU usage.
You can find more information here:
GROMACS Wizard — Custom GROMACS versions and performance settings
GROMACS — Getting good performance from mdrun
These settings apply to local computations. For Cloud computations, GROMACS Wizard uses a GROMACS version built for the selected computing machine, i.e., if you select a machine with a GPU, then it will use it.
Thank you again for the suggestion! We will keep in mind the possibility of making CPU/GPU selection more straightforward in the interface.