We are currently running an Nvidia A100 40GB GPU on our HP DL380. Although we understand that GPU passthrough is experimental, thanks to the tutorials setup of GPU passthrough went well and the performance is good. The only issue we had to solve with this GPU was that we ran into a hard limit for PCI device memory size due to the 40GB.
For a new server we are considering moving from Intel Xeon scalable to AMD EPYC (Milan), with two Nvidia A100.
Has by chance anyone made experience with a similar setup (Nvidia A100+EPYC)? Would anyone recommend against moving from Intel to AMD for an application with GPU passthrough ?
Thanks in advance
For a new server we are considering moving from Intel Xeon scalable to AMD EPYC (Milan), with two Nvidia A100.
Has by chance anyone made experience with a similar setup (Nvidia A100+EPYC)? Would anyone recommend against moving from Intel to AMD for an application with GPU passthrough ?
Thanks in advance
Last edited: