Top-of-the-Line Supercomputer Turbocharges NIH Research
Access to robust computing resources provides a critical foundation for advancing the wide variety of biomedical research taking place within the NIH’s Intramural Research Program (IRP). Whether performing molecular modeling simulations, generating whole-genome sequencing data, deducing the structures of biomolecules, or advancing drug discovery efforts, our ability to analyze large-scale biological and biomedical data strongly depends on our ability to employ computationally intensive approaches that produce interpretable results and advance translational efforts aimed at improving human health.
Given the ever-growing demand for computational resources, the NIH launched a concerted effort to augment our high-performance computing (HPC) environment in the fall of 2013. This endeavor has paid off handsomely, with the NIH going from lagging to leading in four short years.
Biowulf, our HPC resource, now has the distinction of being the first supercomputer completely dedicated to biomedical research to be listed among the Top 100 most powerful computers in the world, coming in at number 66 in the most recent Top500.org rankings released this past November. Rather than just keeping pace, the IRP now finds itself among the world’s leaders in the biomedical computing space, providing its investigators with a world-class HPC environment for advancing biomedical science.
To give you a sense of how far we’ve come since the expansion began, here are some top-level statistics regarding our HPC capabilities:
- Our computing capacity has increased by 413%, going from 18,000 to 92,300 cores. This includes 28 large-memory nodes for memory-intensive projects.
- Our data storage capacity has increased by 733%, going from 3 to 25 petabytes, and now includes leading-edge object-oriented storage technology
- Over 650 graphical processing units (GPUs) have been installed, soon to include an additional 32 high-performance NVIDIA Tesla V100 GPUs. These GPUs provide a massive performance leap over traditional CPUs, thereby accelerating complex computational projects requiring high computational throughput such as high-end simulations and deep learning.
- New 100 Gbps connections that provide a significant improvement in network speeds are now in place, allowing for fast, secure, and reliable data transfers.
The Biowulf expansion has already had a significant impact on how research is performed in the IRP, with half of all IRP investigators’ research programs actively using Biowulf to process and analyze their data. From 2015 through 2017, five percent of all IRP peer-reviewed publications were based on data that were generated or analyzed using Biowulf. And over the past four years, intramural researchers have published over 1000 peer-reviewed studies that relied on NIH’s HPC resources, nearly matching the total from the previous ten years.
These improvements were made possible by the hard work and dedication of many people from across the IRP and NIH, including the members of the NIH’s High-Performance Computing Working Group, who provide leadership on HPC strategies and initiatives to support the IRP’s scientific computing needs. The High-Performance Computing team at NIH’s Center for Information Technology (CIT) has also played a critical and central role in this endeavor, with these unsung heroes working tirelessly each and every day to support the many complex projects that utilize the HPC system and train IRP scientists to effectively leverage the massive computing power available to them.
From sequencing the genome of a parasitic worm to examining how different parts of the brain communicate with one another in autistic individuals, the cutting-edge research enabled by NIH’s HPC resources has dramatically expanded our knowledge of biology and human health and will continue to push the boundaries of modern biomedical science. For more information on the HPC expansion and how IRP scientists are taking advantage of the computational resources available to them, please visit irp.nih.gov/supercomputing.
Related Blog Posts
This page was last updated on Tuesday, May 3, 2022