A statement that is both true and wrong. One of the biggest blunders from Intel was the JCC bug, see https://www.intel.com/content/dam/s...mitigations-jump-conditional-code-erratum.pdf Also affected are the Cascade Lake Xeons. What they did there was fix this in microcode. Performance cost: 0% to 8% according to Phoronix. Depends on your workload, 3-5% should be gone no matter what.Neat, digging further it does appear that some of the Cascade Lake Xeons (e.g. 5220) share a stepping with the ES2's so it should just work. Also bodes well for the ES2's, the fact that some of them made it to production would imply there are no showstopper bugs.
I looked really hard and well-intentioned at ES/QS Xeons but man, all those bugs and all those flaws costing 5% here and 3% there, made me rethink and look at EPYC instead. A Platinum 8180 QS would have to fall to 200 USD per piece to make me reconsider.
My 2 euro cents