A Deep Dive Into Architecting DGX Blackwell Clusters for Generative AI
As the variety and size of generative AI models and the processing needs continue to increase, the compute infrastructure needs to adapt to handle them relaibly. See how we're meeting these needs as we deploy with Blackwell architecture by exploring the philosophy and methodology behind the designs of DGX systems, network fabric, interconnect, and storage to enable optimal trade-offs among performance, power, and thermal (air-cooled and liquid-cooled). You’ll meet the architects of DGX supercomputers, Eos and Selene, and gain insights to how they're creating next-gen clusters to power your generative AI needs.