Linux Scaling Benchmarks With The AMD Threadripper 2990WX In Various Workloads

[Spring]

You have been quoted by PCMag. The reviewer, though, apparently on a quest to support his headline (which makes no mention of Windows flaws), uses a Microsoft product (Visual Studio) to see whether or not another Microsoft product (Windows) is hampering the 2990. This reviewer also doesn't bother to do any Linux tests himself nor does he, after pointing out that your 7 Zip version is older code, bother to check to see if the more recent 7 Zip code could be an issue.

https://www.pcworld.com/article/3298...ce.html?page=2

How do you like this by the way:

Cutting memory bandwidth just kills performance of the Core i9 (blue) but oddly the Threadripper's performance doesn't bump up when two of the dies are switched off.

[chithanh]

How do you like this by the way:

It just goes on to show how the author has no clue. Right after that, he writes:

The result is again confusing

Well duh, it is because we are not looking at a memory bandwidth limited scenario in the 2990WX here.

In fact, from all those tech writers who claim that 2990WX latency/memory bandwidth limitations hurt performance, I have yet to come across a single one who demonstrated such a scenario. Most were only showing applications which don't scale or are not sufficiently NUMA aware (e.g. Euler CFD), or operating system shortcomings (e.g. 7-zip).

If people had actually run tests like 32-core Epyc vs. 32-core Threadripper, or heck even 1950X in dual-channel vs. quad-channel, they would have noticed where actual limitations are.