Return to Level1Techs.com

Threadripper 2000 series /thread


#41

If we assume a reasonably sane “general purpose” VM metric of 4-8 GB per virtual machine core, you’re looking at 128-256 GB. :smiley:


#42

lol here come all the youtube vidya’s hahahha embargo must have been lifted at midnight cst lolol or i am just that far behind its been a super long day im sick im hitting the rack


#43

Steve works like a Draft horse :slight_smile:


#44

So, the Windows CPU scheduler may be to blame for poor performance… Cause the Linux benchmarks prove the WX is still very capable with a proper CPU scheduler.


#45

Like i said, not unexpected. Linux kernel is open to patches by AMD staff, and also runs on things other than x86/x64 and has had a lot more work done to make it work properly on vastly different architectures.

Windows may patch things for this or otherwise enable TR2 users to improve performance. Windows applications may also adapt for better optimisation.


#46

WHAT!? That’s madness, now I want that CPU. Time to upgrade my i7 5820K, this thing is making it look like a Pentium…

/s


#47

Indonesian overclocker, Ivan Cupa, achieved a clock frequency of 5955.4 MHz across all 32 cores and 64 threads that the 2990WX has to offer. The feat was achieved on an MSI MEG X399 Creation motherboard along with a Corsair 1500W power supply and tons of liquid nitrogen to keep temps under control. A CPU voltage of 1.45V was reported within the Ryzen Master utility but we aren’t sure if that is correct since CPU-Z and Ryzen Master utility are reporting different numbers.


#48

Damn. Did they improve their silicon or something?

I didn’t think Ryzen could get anywhere near this.


#49

I’m just as shocked!

I was under the impression that Ryzen had problems when the temperatures went sub zero that kept ln2 overclockers from getting the high numbers.


#50

I doubt that figure was bench stable


#51

I’m definitely curious about the linux gaming side performance both directly and through windows passthrough compared to directly running Windows. Would be hilarious to see the scheduler in linux give better performance through a windows vm than running baremetal.


#52

Ah, I misunderstood then. I suppose they probably would have said, explicitly, if it was.


#53

its wtftech, they tend to exaggerate.


#54

It came from a HWBot entry, which is linked in the article, 2nd paragraph, 5th word.


#55

So, does that mean it’s bench stable? I don’t really pay much attention to this stuff.


#56

image


#57

I guess its bench stable for this benchmark, not sure the intricacies of this one but it worked long enough to run it (probably more then once).


#58

that’s not a benchmark, it’s a cpu verification run.


#59

Its still a pretty insane number. And its not WCCF Tech making it up.


#60

Okay, so it just says he’s not lying about running on that frequency, but something more like PCMark would be required for “stable”?