Statements (29)
| Predicate | Object |
|---|---|
| gptkbp:instanceOf |
gptkb:personal_computer
|
| gptkbp:announced |
2018
|
| gptkbp:architecture |
switching fabric
|
| gptkbp:bandwidthPerPort |
25 GB/s
|
| gptkbp:connects |
multiple GPUs
|
| gptkbp:enables |
multi-GPU communication
scalable GPU clusters all-to-all GPU communication |
| https://www.w3.org/2000/01/rdf-schema#label |
NVSwitch
|
| gptkbp:manufacturer |
gptkb:NVIDIA
|
| gptkbp:purpose |
high-speed GPU interconnect
|
| gptkbp:successor |
direct NVLink connections
|
| gptkbp:supports |
gptkb:NVLink
|
| gptkbp:technology |
gptkb:NVLink_fabric
|
| gptkbp:USB |
18
|
| gptkbp:usedFor |
high-performance computing
AI workloads data center acceleration |
| gptkbp:usedIn |
gptkb:NVIDIA_Eos_supercomputer
gptkb:NVIDIA_HGX_systems gptkb:NVIDIA_DGX_systems gptkb:NVIDIA_DGX-2 gptkb:NVIDIA_DGX_A100 gptkb:NVIDIA_DGX_H100 gptkb:NVIDIA_Selene_supercomputer AI supercomputers |
| gptkbp:bfsParent |
gptkb:NVLink
gptkb:英伟达 |
| gptkbp:bfsLayer |
6
|