NVIDIA restarts Rubin CPX inference chip project
The Rubin CPX targets over 50% of AI inference workload for prefill processing at lower cost, addressing long-context model demands with 1.34TB HBM memory per tray.
IT之家Source
#Rubin CPX
1 stories · 0 in 7 days · 0 in 30 days
Stories, products and papers tagged "Rubin CPX", newest first.
Sep 1