Torsten Hoefler

Torsten Hoefler is a Professor at ETH Zurich, renowned for pioneering contributions to High-Performance Computing (HPC) and its application to Artificial Intelligence (AI). His work on scalable network design, parallel algorithms, and Message Passing Interface (MPI) advancements has revolutionized supercomputing and AI infrastructure. Hoefler chairs key MPI working groups and co-developed foundational concepts like "3D parallelism" that underpin modern AI systems.

Hoefler's research spans interconnection networks, congestion avoidance protocols (e.g., Slim Fly, PERCS), and performance modeling. His innovations in nonblocking collective operations (Iallreduce, Iallgather) power distributed deep learning frameworks. He also leads benchmarking and reproducibility initiatives that set global HPC research standards.

• ACM Prize in Computing (2024): Recognized for enabling AI's computational scale through HPC breakthroughs
• ACM Gordon Bell Prize (2019): For quantum transport simulations mapping transistor heat distribution

Hoefler's work impacts millions via large-language model training (e.g., ChatGPT) and reduces data center cooling costs through thermal modeling. He leads teams at ETH Zurich's Scalable Parallel Computing Laboratory, translating theoretical advances into industry-adopted technologies.