Yan Receives NSF Grant to Boost Multi-GPU Supercomputer Capabilities – News

The grant will accelerate graph processing in supercomputers and provide software support for the rapid deployment of GPU supercomputers across the country.

Written by Tehreem Khan
Media contact: Alicia Rohan

Stream DaYan 2RT scrDa Yan, Ph.D., an assistant professor in the University of Alabama Department of Computer Science at the Birmingham College of Arts and Sciences, received an RII Track 4 grant from the National Science Foundation to improve the efficiency of large-scale processing in supercomputers with graphics processors.

The $275,573 grant will be used to support Yan’s project entitled Massively Parallel Graph Processing on Next-Generation Multi-GPU Supercomputers. The impact of this project will be significant considering the nation is replacing the central processing unit supercomputers with GPU supercomputers faster than ever to benefit from a significant increase in performance and energy efficiency. However, this trend creates challenges for large-scale graph processing, as users need to develop custom GPU programs for each and every graph problem. This project can potentially address this limitation by providing a unified programming framework with optimized system support.

CPUs act as a computer’s brain, giving instructions to other components of a computer telling them what to do, while GPUs were originally designed to speed up the creation of images in a frame buffer for output to a Display device are determined. General purpose computing on GPU units is the use of a GPU beyond computer graphics computations to perform computations in applications traditionally performed by the CPU with massive parallelism, including problems in graph theory.

“Building on my success in developing the T-Thinker framework to speed up basic graph operations in a CPU-rich environment, this project aims to develop a new graph-parallel framework called T-Thinker GPU that performs graph computations effectively will accelerate a GPU-rich environment,” said Yan. “At least three basic graph operations are implemented on the T-Thinker GPU: subgraph matching, dense subgraph mining, and frequent subgraph pattern mining.”

READ :  Living on the Edge: Will 2023 be the Year of Edge Computing?

The scholarship also supports two joint visits for Yan and a Ph.D. Student at Argonne National Laboratory in summer 2023 and summer 2024. The principal investigator will develop and test these GPU tools on ANL’s exascale supercomputer Aurora. Once developed, they will collaborate with subject matter scientists in ANL to deploy and use these GPU tools in scientific applications such as bioinformatics and knowledge graph searches. These efforts will greatly improve the way computer science curriculum is developed and taught.

“By bringing the current GPU programming technology learned at ANL back to the UAB Department of Computer Science, we will contribute to the parallel computing curriculum, especially modern GPU technology,” Yan said. “The courses will be able to train more students to become the GPU programming professionals that are desperately needed not only at UAB but throughout the state of Alabama.”

The effect doesn’t stop there. According to Yan, the project will bring back enough GPU experience to work with local researchers and explore the use of the proposed T-Thinker GPU framework, designed to address intensive big data problems and help create a larger Big data problem to be broken down into smaller tasks. This framework is used in applications such as third-generation DNA sequencing, visualization of large multi-omics networks, and biomedical image analysis, benefiting a variety of industries.