Statements (19)
| Predicate | Object |
|---|---|
| gptkbp:instanceOf |
gptkb:personal_computer
|
| gptkbp:announced |
2020
|
| gptkbp:connects_up_to |
32 GPUs
|
| gptkbp:enables |
multi-GPU communication
scalable GPU clusters |
| gptkbp:manufacturer |
gptkb:NVIDIA
|
| gptkbp:maximum_bandwidth |
900 GB/s
|
| gptkbp:part_of |
NVIDIA data center solutions
|
| gptkbp:supports |
gptkb:NVIDIA_A100_GPUs
gptkb:NVIDIA_H100_GPUs NVLink protocol |
| gptkbp:used_in |
gptkb:NVIDIA_HGX_systems
gptkb:NVIDIA_DGX_systems |
| gptkbp:usedFor |
high-performance computing
AI workloads high-speed GPU interconnect |
| gptkbp:bfsParent |
gptkb:NVIDIA_Grace_Hopper_Superchip
|
| gptkbp:bfsLayer |
7
|
| https://www.w3.org/2000/01/rdf-schema#label |
NVIDIA NVLink Switch System
|