Am I... not possible right, but 7601 (Naples) vs 7452 (Rome) 32 cores CPU - is there a RADICAL difference in performance? But the price difference $500-600 vs $2000-2200 is FOUR TIMES! Not sure, there's a 4 times increase... No offence, but...
The problem with Naples isn't that its slow on average, it's that there are pathological cases where the internal architecture fails to scale as expected. Looking at
@Patriot 's post for example, you would expect that a 7601 roughly matches a 7402 (the 7402 boosts higher and has a generation of architectural improvements, but 25% fewer cores), but the 7402 beats it again by roughly 25%. That's not bad, except...because the desktop parts are clocked so high, a 7402 performs like a 5950X even on workloads that scale, so you end up with a $500 CPU on a $500 motherboard which performs like a $700 desktop part with half the cores. Compounding the issue is the fact that you can't use a Naples chip as a stand-in for Rome until stock recovers, because there are virtually no platforms with simultaneous Naples and Rome support.
Naples was OK in the cloud because the hyperscalers have a different set of metrics - boost performance is not important to them because you expect your physical hosts to be fully utilized in most scenarios, and being able to scale out to tons of I/O is important because you might have a bunch of VMs hammering the disks and network at once. It was also OK for rendering work, but mostly because AMD gave you a good deal at a time when a 28C Skylake was $11K. Now that 24C Skylakes are $500 on eBay it makes a lot less sense.
It's all a little tenuous when you think about it - I'm sure part of the reason why Rome parts are so expensive is because they are "current gen", and that's still worth something to a lot of people. The 5950X is so good (especially now that 32GB UDIMMs are popular) that it's really hard to justify "small" Rome unless you are doing some crazy project with 100G ethernet and tons of passthrough.