Windows 11 scores varied wildly

[David (PassMark)]

Just to add to the pile of woe. I was also reading today that Win11 slows down SSDs in some circumstances.

[David (PassMark)]

Yes, we support Win11.

Can you post the individual 2D results. Both before and after installing Win11.

[Manjunatha]

Update: I ran the test again a couple hours later. The results for CPU are still great but my 2D score seems to have halved (and consistently is now half of what it was before between runs)! How strange....

I wish I had the breakdown of my 1215 2D score from earlier to compare.....oops

Please confirm PassMark 10 support on Windows 11 OS ?

We are getting 28% low score for 2D graphics mark and 13 % low total score compared to windows 10 OS
Low score observed after updating KB patch and driver for AMD CPU

[GameRoof]

Update: I ran the test again a couple hours later. The results for CPU are still great but my 2D score seems to have halved (and consistently is now half of what it was before between runs)! How strange....

I wish I had the breakdown of my 1215 2D score from earlier to compare.....oops

[GameRoof]

KB5006746 from Windows Update + AMD Chipset Driver 3.10.08.506 have fully resolved the anomalous Prime Number results! Memory score seems to be consistently a bit better also. So results are back in line with where they should be now.

Yippee!

[mp5allstar]

Just recently updated my Windows 11 machine last night, and it looks like it is back to normal! To everyone - I modified my CPU curves and maximum thermals (75 degrees) so that is why my 5900X score is "only" 38200.

[CerianK]

Agreed, It may be worth noting that security features may respond to various usage scenarios in seemingly odd ways, especially when layered on other issues.
While not security software related, I have seen Windows Update (on 10) trigger on dozens of PCs with attached Ethernet devices that are not Internet connected, creating many 1000s of Event Log entries (related to update failures) per day. The PC is no longer very responsive, at least in the case where an HDD is in use.

[David (PassMark)]

CerianK,
That is kind of a different issue. That second article was a discussion about how new security features (slightly) hurt performance. And it seems this is expected and a deliberate trade off.

The issue initially reported here was just a bug in the was the cache was managed.

[CerianK]

Additional relevant article from Tom's Hardware:
https://www.tomshardware.com/news/wi...-hvci-security

[CerianK]

Possibly relevant Tom's discussions:
https://www.tomshardware.com/news/am...5-patch-coming (Edit: David already linked)
https://www.tomshardware.com/how-to/...vbs-windows-11

[David (PassMark)]

Yep.

And here is the official documentation from AMD for the issue
https://www.amd.com/en/support/kb/faq/pa-400

Turns out this forum post was the the 2nd report of the issue worldwide.

[mp5allstar]

If this was true (Win11 hurts CPU & disk performance) it would both be unexpected and pretty big news in the IT world. I suspect there must be some sensible explanation for this. e.g. different power saving setting, additional Win11 background activities for a few hours, like rebuilding the Windows search index.

We run some tests today to see what our results are. We only have one Win11 machine at the moment. So it won't be conclusive.

Looks like it did become pretty big news then huh?

[David (PassMark)]

Or maybe it is this issue
https://www.tomshardware.com/news/am...5-patch-coming

"First, the measured and functional L3 latency [in Win11] can increase by ~3X, meaning you can see the impact with measurement utilities and that it results in a real degradation of performance in applications"

Or this one,

"Additionally, AMD's "preferred core" feature, which directs single-threaded applications to the fastest core on the chip, also might not work as expected. This would primarily impact performance in lightly-threaded applications. AMD says this performance reduction may be more noticeable in chips with eight or more cores and a 65W or higher TDP rating".

So many bugs to choose from, I don't know which one to pick.

That x3 cache performance difference does sound very very similar to what was measured above however.

[David (PassMark)]

AMD are saying they know what the cause is, but can't tell us due to NDAs. Sigh.

But we are suspecting it might be the new Microsoft's Virtualization-Based Security (VBS) and Hypervisor-Protected Code Integrity (HVCI) features in Win11. There are some details here about turning it off
https://docs.microsoft.com/en-us/win...code-integrity

Update: We testing some of these security options today, (again on a very limited number of machines) and it doesn't seem to be the cause of such a big difference in Prime number result.

[David (PassMark)]

Alexatkin, what hardware are you using and was the slow down due to a lower score for the prime number test?