Torsten Hoefler

Examining the true capabilities and limits of quantum computers today.

Canary improves allreduce performance by adapting to network congestion dynamically.

A new algorithm enhances communication speed in high-performance computing.

Learn how OSMOSIS allocates network resources fairly in modern data centers.

SMaRTT offers a fast solution to congestion problems in machine learning data transfers.

LLAMP assesses network latency tolerance for high-performance computing applications effectively.

A new platform enhancing data processing using smart NICs.

Falcon and DST improve speed and efficiency in vector search systems.

SMaRTT improves network performance for AI workloads using fast response and effective routing.

New techniques improve large language model performance and efficiency.

The Quad GH200 supercomputer optimizes data flow for enhanced performance.

A new approach enhances data sharing among distributed AI models.

Exploring communication methods within multi-GPU supercomputers for enhanced performance.

This study presents techniques to improve loop performance in programming.