Statements (23)
| Predicate | Object |
|---|---|
| gptkbp:instanceOf |
AI supercomputer
|
| gptkbp:announced |
2023-05-28
|
| gptkbp:connects |
gptkb:NVLink_Switch_System
Grace Hopper Superchips |
| gptkbp:designedFor |
AI workloads
large language models generative AI |
| gptkbp:features |
gptkb:NVLink_interconnect
shared memory architecture GH200 Grace Hopper Superchips |
| gptkbp:firstCustomers |
gptkb:Google
gptkb:Microsoft gptkb:Meta |
| https://www.w3.org/2000/01/rdf-schema#label |
Nvidia DGX GH200 system
|
| gptkbp:location |
gptkb:United_States
|
| gptkbp:manufacturer |
gptkb:Nvidia
|
| gptkbp:maxGPUCount |
256
|
| gptkbp:maximumSpeed |
1 exaflop AI performance
|
| gptkbp:maxSystemMemory |
144 TB
|
| gptkbp:supports |
AI inference
AI model training |
| gptkbp:bfsParent |
gptkb:Nvidia_GH200_Grace_Hopper_Superchip
|
| gptkbp:bfsLayer |
7
|