Case Study: Ghent University cuts AI experiment time from nearly seven hours to 40 minutes with Supermicro GPU servers

Supermicro’s new 10U GPU server using NVIDIA HGX-2 GPU baseboards helps leading Belgian research team improve calculation times by 10X which leads to faster AI development

Ghent University's IDLab research group required a substantial increase in computational power to maintain its leadership in AI research. Their existing infrastructure could not keep up with demand, and they needed an extremely high-speed GPU server capable of running at least six simultaneous jobs, with most completing within 40 minutes. This challenge necessitated a tenfold performance improvement.

Supermicro provided a solution with its new 10U GPU server powered by two NVIDIA HGX-2 boards, each with eight NVIDIA V100 GPUs. The implementation, aided by the flexibility to use the lab's own software, cut experiment times from nearly seven hours down to 40 minutes. This tenfold performance increase allowed Ghent University's data scientists to conduct research much faster by running multiple iterations quickly to converge on optimal solutions.

Ghent University

Brecht Vermeulen

Head of the Research Infrastructure iLab.t