Tweet
You guys do a lot of interesting Xeon versus Eypc benchmarks on this site. Have you ever considered doing a proper database test, for instance HammerDB / TPC-C? So far, no one has ever really compared the database performance of Xeon versus Epyc. Databases are an example of "Non- NUMA aware" applications and possibly EPYC's weak point because of it's MCM architecture. However, it has never been properly measured how well it copes running a TPC-C workload (http://www.hammerdb.com/docs/ch03s05.html)