Workshop: HPC Systems Professionals Workshop (HPCSYSPROS26)
Authors: Mateus Francisco (Unicamp) and Fabio Andrijauskas (University of California San Diego)
Abstract: High-performance computing (HPC) is a necessary tool for many research fields. However, the cluster is not the only component necessary to operate an HPC environment. This paper presents the evolution of the Coaraci High-Performance Computing infrastructureat the Gleb Wataghin Institute of Physics (IFGW), University of Campinas (UNICAMP), Brazil; The project modernized an existing academic data center through upgrades to electrical power, backup generation, UPS systems, precision cooling, airflow management, high-speed networking, workload management, monitoring, user onboarding, and technical support. Thermal improvements included cold-aisle containment, high-flow floor panels, and high-airflow rack doors.
Average Power Usage Effectiveness (PUE) decreased from 2.14 in April 2024 to 1.56 in June 2025, an observed reduction of about 27%. The optimized cooling strategy also enabled operation with two active cooling units and one standby unit. Since February 2024, Coaraci has operated 24/7 with 366 registered users, 284 of whom have executed workloads.
Back to HPC Systems Professionals Workshop (HPCSYSPROS26) Archive Listing
Back to Full Workshop Archive Listing