AND SCALABILITY BY 3X
TO EXPEDITE SCIENTIFIC RESEARCH
UNIVERSITY OF BRISTOL
“Early benchmarking shows that the new system is three times faster than our previous cluster—research that used to take a month now takes a week, and what took a week now only takes a few hours. That’s a massive improvement that will be a great benefit to research at the University.”
–Dr. Christopher Woods, EPSRC research software engineer fellow, University of Bristol
More than 1,000 researchers at the University of Bristol in areas such as paleobiology, earth science, biochemistry, mathematics, physics, molecular modelling, life sciences, and aerospace engineering are benefitting from a new high-performance machine named BlueCrystal 4 (BC4). With more than 15,000 cores, BC4 is the largest UK University system by core count with a theoretical peak performance of 600 teraflops.
“Some of our researchers are looking at whole-planet modelling with the aim of trying to understand the earth’s climate, climate change, and how that’s going to evolve, while others are looking at rotary blade designs for helicopters, gene mutation, and where diseases come from,” said Dr. Christopher Woods, EPSRC research software engineer fellow, University of Bristol. “Early benchmarking shows that the new system is three times faster than our previous cluster—research that used to take a month now takes a week, and what took a week now only takes a few hours. That’s a massive improvement that will be a great benefit to research at the University.”
BC4 uses Lenovo NeXtScale compute nodes, each comprised of two 14 core 2.4 GHz Intel Broadwell CPUs with 128 GiB of RAM. It also includes 32 nodes of two NVIDIA Pascal P100 GPUs plus one GPU login node, designed into the rack by Lenovo’s engineering team to meet the specific requirements of the University.
Connecting the cluster are several high-speed networks, the fastest of which is a two-level Intel Omni-Path Architecture network running at 100Gb/s. BC4’s storage is comprised of 1PB of disk provided by DDN’s GS7K® and IME® systems running IBM’s parallel file system Spectrum Scale™.
Benchmarking capabilities of Lenovo’s HPC research center in Stuttgart has shown that BC4 is highly efficient in terms of physical footprint, while fully utilizing the 30KW per rack energy limit. Lenovo’s commitment to third-party integration has allowed the University to avoid vendor lock-in while permitting new hardware to be added easily between refresh cycles.
Dr. Christopher Woods continues: “To help with the interactive use of the cluster, BC4 has a visualization node equipped with NVIDIA Grid vGPUs; this helps our scientists to visualize the work they’re doing, so they can use the system even if they’ve not used an HPC machine before.”
Currently numbering in the hundreds, applications running on the University’s previous cluster will be replicated onto the new system, allowing researchers to create more applications and better scaling software. Applications can be moved directly onto BC4 without re-engineering.
Launched at an event on May 24 at the University of Bristol, BC4 will house more than 1,000 system users.