Join the discussion
Question 5/31
An architect is designing a distributed training solution using HPE Private Cloud AI. The solution involves multiple HPE ProLiant DL380a servers, each with four NVIDIA H100 GPUs.
Review the logical network topology:
```
[HPE ProLiant Server 1] \
- GPU 0 <--> GPU 1 (Internal)
- GPU 2 <--> GPU 3 (Internal)
\
[NVIDIA Spectrum Switch (RoCE)] <--> [HPE GreenLake for File Storage]
/
[HPE ProLiant Server 2] /
- GPU 0 <--> GPU 1 (Internal)
- GPU 2 <--> GPU 3 (Internal)
```
Which interconnect technology facilitates the highest-bandwidth communication path labeled "(Internal)" for scaling the training job across the GPUs within a single server?
Review the logical network topology:
```
[HPE ProLiant Server 1] \
- GPU 0 <--> GPU 1 (Internal)
- GPU 2 <--> GPU 3 (Internal)
\
[NVIDIA Spectrum Switch (RoCE)] <--> [HPE GreenLake for File Storage]
/
[HPE ProLiant Server 2] /
- GPU 0 <--> GPU 1 (Internal)
- GPU 2 <--> GPU 3 (Internal)
```
Which interconnect technology facilitates the highest-bandwidth communication path labeled "(Internal)" for scaling the training job across the GPUs within a single server?
Add Comments
