StreamHPC has carried out several successful parallel-computing projects in molecular dynamics since 2012. Below are a few examples of our work in bioinformatics, chemistry and meteorology.
GROMACS is one of the fastest molecular dynamics software packages on the market. To broaden the user base that can benefit from the processing power of modern GPUs, we ported GROMACS from CUDA to OpenCL and further optimised the code for use with AMD FirePro accelerators. The resulting performance is on a par with that of the original CUDA code but without the restriction of being bound to a specific parallel computing hardware. GROMACS is used world-wide by over 5000 research centers, from simulating molecular docking to examining the hydrogen bonds in a falling water drop. Read more…
For the university of Stanford, we further optimised a part of TeraChem, a general purpose quantum chemistry software designed to run on NVIDIA GPU architectures. Our work resulted in adding an extra 70% performance to the already optimised CUDA code.
For the University of Manchester, we developed a high-performance implementation of the UNIFAC group contribution model for their research on atmospheric aerosol particles. Where an OpenMP implementation of the original single-threaded code got the run time down from 32 to about 10 seconds on a quad-core CPU, we eventually brought it down to 0.062 seconds using OpenCL on a Xeon Phi accelerator – a speedup of 160x over OpenMP. Read more…
