Visual Molecular Dynamics on CUDA GPUs

VMD is a molecular visualization program for displaying, animating, and analyzing large biomolecular systems using 3-D graphics and built-in scripting. VMD is distributed free of charge by Theoretical and Computational Biophysics Group at the University of Illinois, Urbana-Champaign.

Several key kernels and applications in the release version 1.8.7 of VMD take advantage of the massively parallel CUDA architecture of NVIDIA’s GPUs. These applications run 20 to 100x faster when using a NVIDIA CUDA GPU compared to running them on a CPU only and are discussed below.

You can learn the latest about GPU-acceleration of VMD and NAMD at //

Tesla Personal computer
Look for NVIDIA Tesla Personal Supercomputers from NVIDIA Tesla Preferred Providers.
where to buy

Molecular Orbital Display

Visualization of molecular orbitals (MOs) is important for analyzing the results of quantum chemistry simulations.

As shown by the results in the chart, the GPU enables the first-ever interactive animations of quantum chemistry simulation trajectories using only on-the-fly calculation.

  • CPU run-times are 10-100 seconds
  • GPU run-times are less than 0.3 seconds

The speedup of this application in VMD scales with the number of Tesla GPUs:

  • 3 Tesla GPUs achieve 323x speedup
  • 4 Tesla GPUs achieve 412x speedup

VMD Molecular Orbital Computation
Molecular Orbital Computation in VMD


Ion Placement

Coulomb-based ionization of very large structures such as viruses can require several days even using moderately sized CPU clusters.
The GPU-based ion placement tools in VMD accelerate the electrostatic potential map calculations to the point where 4 Tesla C1060 GPUs achieve 1.78 Teraflops of actual performance on this task. Using GPUs enables precomputing average fields, so that you can replace on-the-fly computations within the molecular dynamics (MD) codes.

VMD Ion Placement
Direct Coulomb Summation (69x faster)


Implicit Ligand Sampling

Implicit ligand sampling is used to find gas migration pathways inside proteins. It is used to study how O2, CO, etc molecules make their way inside proteins and is particularly useful in alternative fuel and energy research.
A ILS run that takes 2 more months on a CPU, takes only 3 days to run on a single Tesla C1060 GPU.

VMD Implicit Ligand
Implicit Ligand Sampling in VMD (20x faster)

CUDA-Acceleration in Related Verticals See Also