MIT Sloan: Lead AI Adoption Across Your Organization — Not Just Pilot It
Learn AI, Data Science & Business — Earn Certificates That Get You Hired
Overview
Google, IBM & Meta Certificates — All 10,000+ Courses at 40% Off
One annual plan covers every course and certificate on Coursera. 40% off for a limited time.
Get Full Access
Learn about an algorithmic framework for constructing direct-connect network topologies optimized for collective communications in this 14-minute conference presentation from NSDI '25. Discover how researchers from the University of Washington, Raytheon BBN Technologies, and MIT CSAIL tackle the challenge of balancing latency versus bandwidth trade-offs in network workloads through systematic topology synthesis. Explore their methodology for starting with small, optimal base topologies and iteratively scaling them to create much larger, efficient network structures. Understand how the team incorporates established large-scale graph topologies into their framework using a novel polynomial-time algorithm for generating efficient communication schedules. Examine the evaluation results from multiple testbeds and large-scale simulations that demonstrate significant performance improvements achieved through their derived topologies and communication schedules, providing practical insights for optimizing collective communication systems in distributed computing environments.
Syllabus
NSDI '25 - Efficient Direct-Connect Topologies for Collective Communications
Taught by
USENIX