Sometimes their is a serious difference in Enterprise hardware for Data Center use, and using it for things like gaming. Many Enterprise computers use registered memory, which under some uses can be slower. This is because the amount of memory and ECC is more important that the last percent of speed.
Further, some Enterprise hardware is designed for multi-threaded uses, not maximum single threaded performance. So results can be mixed.
Now, using HPC, (High Performance Computing), like an AMD Thread Ripper is something different. For a while, both Intel and AMD paused / stopped producing CPUs for desktop HPC. But, AMD returned with new Thread Ripper models;
Now it does use registered memory, though it is newer DDR5.
Here is an example of Enterprise verses other. When I worked at Sun Microsystems about 2007 we got some new servers, the T2000. This was Sun's, (and the worlds), first heavy multi-threaded CPU. It had 8 cores with 4 threads per core for a total of 32 CPU threads.
One of those T2000s was setup as a compile server. The users complained that their fast Sun desktops could compile code in 5 minutes, where the T2000 took 20 minutes. I then asked the user to do 2 compiles on each and give me back the times. The Sun desktop did 2 compiles in 10 minutes, (twice as long), but the T2000 with it's MANY threads took 20 minutes, (same time), FOR TWO!
So while the T2000 was not a speed daemon, nor had some of the fancy desktop features, in multi-threaded roles it ruled the data center.