Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Discover how the Keysight AI (KAI) Data Center Builder solution supports every phase of AI data center design and deployment in this 13-minute technical presentation. Learn how this tool provides actionable data to improve performance and increase reliability of AI clusters by streamlining design processes, optimizing resource allocation, and enhancing overall efficiency of AI infrastructures. Alex Bortok, Lead Product Manager for AI Data Center Solutions at Keysight Technologies, explains how KAI Data Center Builder tests backend networks that GPUs use for data exchange during model training and inference, distinguishing it from front-end network testing solutions. Explore the product's capabilities in emulating AI workloads, benchmarking network infrastructure, fine-tuning performance, reproducing production issues, and planning for new designs without purchasing expensive GPU hardware for testing. The presentation highlights two key applications: Collective Benchmarks for zooming in on single transactions between thousands of GPUs, and Workload Emulation which accounts for the relationship between computation and data movement. Recorded live in Santa Clara on April 25, 2025 as part of AI Infrastructure Field Day, this talk provides valuable insights for professionals working with AI infrastructure optimization.
