Run GROMACS on Slurm-backed HPC at WECORE: high-throughput biomolecular dynamics for proteins, lipids, and nucleic acids, with non-bonded work offloaded to GPU and multi-node MPI where it pays.
GROMACS is a molecular dynamics engine written for biomolecular systems and tuned for throughput.
Its non-bonded kernels are hand-optimized per architecture, and its GPU offload path is mature enough that a well-configured single node often outperforms a naive multi-node run.
Run folding studies and conformational sampling long enough for the slow motions to show up.
Build and equilibrate bilayer systems, then run membrane proteins in a realistic lipid environment.
Estimate protein–ligand binding free energies from perturbation windows, each queued as an independent job.
Run nucleic-acid systems and the long production campaigns that ensemble methods depend on.
Queue many independent replicas rather than one enormous job, which is the shape most GROMACS work takes. Multi-node MPI is there for the single system that genuinely needs it.
GPU nodes carry the non-bonded kernels, and a well-configured single node often beats a naive multi-node run. The useful question is how many replicas finish per day, not how fast one step is.
