Coursera Spring Sale
40% Off Coursera Plus Annual!
Grab it
Explore the evolving landscape of NVMe storage technology in this 35-minute conference presentation that examines how AI advancement is driving significant changes in storage infrastructure requirements. Learn about the bifurcation occurring in storage needs, where AI workloads like model training, checkpointing, and key-value cache tiering demand line-rate saturating SSDs with near-GPU and HPC attachment capabilities, while multi-stage inference, synthetic data generation, and post-training optimization require dense, high-capacity disaggregated storage solutions that are displacing traditional rotating media in datacenter nearline tiers. Discover the architectural considerations spanning both ends of this spectrum, including Gen6 performance capabilities, indirection unit selection strategies, power monitoring approaches for enhanced energy efficiency, liquid-cooled thermal design implementations, and methods for achieving high capacity through strategic form factor and packaging choices. Understand how AI applications and workloads are creating divergent storage requirements, examine the critical architectural requirements for Near GPU SSDs, and explore various approaches to managing energy efficiency in SSDs as power demands increase with Gen6 performance levels. Gain insights into how thoughtful design decisions can unlock storage systems' full potential in addressing the complex and evolving challenges presented by modern AI workloads, presented by Suresh Rajgopal from Micron Technology at SNIA SDC 2025.
