I think I said the RAM usage from the DBs would be low, but if I am training a LLM or other large model, the RAM usage could balloon. Plus there is a chance the RAM usage from in-memory DBs might spike as well, depending on what I am working on.heh, that's a big change from "the ram usage will be pretty low" a few posts up. if your requirements are 512GB RAM and 16 lanes of nvme then yes, you are forced onto an enterprise platform. insofar that you accept performance is bad anyway compared to the desktop platforms, Skylake-SP is not a bad choice - peak performance will be about 20% worse than Milan, but the monolithic dies help idle power and you get x16 GPU, x16 nvme, and x8 HBA all at full speed
Interesting about the Skylake-SP, but I the EPYC and the H12SSL have already shipped, so no going back there