Xi Computers has deployed a 6.3 PFLOPS AI supercomputer at the University of Wyoming’s (UW’s) Advanced Research Computing Center (ARCC). This supercomputer enhances UW ARCC’s computational capacity for scientific and AI research. Xi Computers demonstrates expertise in configuring and delivering high-performance computing clusters, utilizing Intel, AMD, and Nvidia technology.
Xi Computers has announced the deployment of a 6.3 PFLOPS AI supercomputer for the University of Wyoming’s (UW’s) Advanced Research Computing Center (ARCC). The system features a mix of CPU and GPU nodes, including 25 1U Xi BladeRAIDer units with 50 AMD Epic 9454 48-core CPU nodes, delivering 259,105 MOPS of floating-point performance. It also includes 30 4U Xi NetRAIDer cluster nodes, housing 48 Nvidia L40S GPUs, 64 Nvidia A30 GPUs, and 48 Nvidia H100 SXM GPUs. These GPUs provide peak performance of 6.3 PFLOPS, focusing on FP64 and FP32 operations.
The L40S is a more powerful variant of the L40, with a higher clock speed and more CUDA cores. This results in a higher performance per watt and a lower power consumption. The L40S is a professional graphics AIB by Nvidia, launched on October 13, 2022. It is built on the 5nm process and is based on the AD102 graphics processor. Nvidia L40S is a lower-cost version of the $32,000 Nvidia H100 AI.
The supercomputer’s nodes are connected through a 200 MHz Nvidia InfiniBand network and a 200 Gbps Ethernet network, integrated seamlessly with the university’s existing infrastructure.
The system constitutes one of the most advanced current supercomputer configurations and is intended to significantly enhance the computational capacity for scientific and AI research at UW ARCC.
The project highlights Xi’s expertise in configuring and delivering high-performance computing (HPC) clusters. Known for delivering custom-built desktops, laptops, workstations, and AI supercomputer clusters, Xi Computers designs systems using Intel, AMD, and Nvidia technology to meet the needs of various industries.