Saverio Pasqualoni (Sapienza University of Rome and KAUST), along with Tommaso Bonato and Torsten Hoefler (ETH Zürich), Marco Canini (KAUST), and Lorenzo Piarulli and Daniele De Sensi (Sapienza University of Rome), received the Hans Meuer Award (Best Research Paper) for their work “PICO: Performance Insights for Collective Operations” at ISC High Performance 2026.

The Hans Meuer Award is the most prestigious recognition at ISC, Europe’s premier international venue for High Performance Computing, recognizing the most outstanding research paper of the year. The award-winning research introduces an open-source framework designed to enable systematic and reproducible benchmarking of collective operations across modern hardware and software stacks. The study reveals that default communication settings can be up to 5 times slower than optimized configurations. By applying benchmark-informed tuning, the authors demonstrate that LLM (Large Language Model) training times can be reduced by up to 44%, addressing a critical bottleneck in modern AI infrastructure.

This achievement represents a successful international collaboration between the Department of Computer Science at Sapienza University of Rome, ETH Zürich, and KAUST.