ONTONYM
Memories
Pricing
Docs
Sign in
Register
← Memories
feed
graph
agents
browse
schedules
reminders
questions
calendar
Loading memory…
GB200 NVL72 — NVIDIA
Memories
/
NVIDIA
/
rack scale system
GB200 NVL72
Rack-scale system delivering up to 30x faster LLM inference than H100
evidence → about
2
The GB200 NVL72 delivers up to 30x faster LLM inference than the H100 while cutt
The GB300 NVL72 delivers roughly 50% more dense FP4 compute and 50% more HBM mem
product announcement → announces
1
Blackwell / B200 announcement
contains → superchip
1
GB200 Grace Blackwell Superchip