I'm planning to build a GPU server with two Tesla V100 cards but I'm not sure about what kind of connection to ask: SXM2 or PCIe.
Is it possible to build an SXM2-based GPU server using only two Tesla V100? Are there any issues related to not fill all the slots?
And how expensive is an SXM2-based server in comparison with a PCI-e based?
Thanks in advance,