The Vera Rubin CPU/GPU Coherent Memory Architecture. Different kind of Inferencing tasks require specifiy AI Chip and memory designs. In Inferencing, there is no one-size-fits-all (HBM) approach. NVIDIA is addressing these requirements with their Vera Rubin Architecture.
Source: NVIDIA
Feb 19
at
7:36 AM
Relevant people
Log in or sign up
Join the most interesting and insightful discussions.