windows 11 has been a downgrade for me hope it gets better.

[BotZero]

I cant replicate the scores i had on windows 10.

[Diogenes49]

I tried it and i went back to windows 10

[Supafly]

Tried 11 for a day and returned to 10. Thankfully I made a backup image of 10 before I installed 11 so it was a quick switch back. Not changing till I absolutely have to.

[Migz - DH]

For me, when I upgraded from Win 10 to Win 11 using the in-place update process, I discovered a small decrease in performance.

BUT, later, I completely reinstalled for unrelated reasons as a fresh installation instead of an update, and my performance was then the same as with Win 10.

I don't know why a fresh installation would be different from an in-place update/upgrade, but for me, it was.

Since then, Win 11 has been fine for me performance wise.

[Supafly]

Originally posted by Migz - DH:

I don't know why a fresh installation would be different from an in-place update/upgrade, but for me, it was.

Bugs happen which is why a fresh OS install gets recommended to fix a lot of issues. Sometimes a Windows updates can cause a bug during the update. I've been know on the odd occasion, when I've encountered a bug after a Windows update and nothing else on my system has changed, I've used a System Restore Point to before the update then redid the exact same update.

Issue didn't happened a second time, just a buggy update not the specific update itself

[Arokhantos]

Windows 11 is clearly breaking 3dmark currently its as if overclocking does nothing for performance but then i overclock on external windows drive that has no updates or usual software score go's from 13k to 17k and now upon clean install its stick at 15k
either 3dmark is broken or i am being trolled.

[UL_Jarnis]

If you have two OS installs and you run the same application in both and get very different results, you should probably consider the most likely explanation being that there is something wrong with one of the OS installs.

3DMark is quite mature application and in all honesty it is the one job of the OS to provide a stable, compatible platform for running any and all applications made for that platform, without breaking anything between versions. This is not an easy job, and I do not envy Microsoft in any way, but they are a big business, they should be able to handle the stress.

Unfortunately currently they seem to use a model where end users get to be part of the OS beta test, so if you are using cutting edge versions (insider preview or very recently updated "live" version of latest OS) you may end up finding issues. Please do not automatically assume that software that has existed for more than ten years and that is actively supported is at fault when Microsoft ships something new.

Also 3DMark exists to measure system performance. If it gives you a measurement that is low, I guess the options are:

A: There is something wrong with the hardware
B: There is something wrong with the OS or drivers
C: There is something wrong with system settings
D: The benchmark somehow has forgotten how to measure the performance (and in case of 3DMark, this is "render 3D graphics, measure framerate, do some elementary school level math to calculate a score based on the framerate")

I would humbly suggest that options A, B or C are the issue in 99%+ of the cases.

[Arokhantos]

I usually assume option E its always user error but i cant figure it out, it also go's down each run getting lower and lower as if benchmark thinks that its running on a 6900 XT instead of a 7900 XTX cos i can see my framerate drop by 5 or 10 between runs as if its ignoring prefered cores and running on the slowest core.
edit; not user error clearly 3Dmark is broken scores only going down each run.

[UL_Jarnis]

3DMark does not decide which cores it runs on. OS scheduler job to decide. It queries how many logical cores there are, then leaves one unused to ensure GPU driver is not starved and then multithreads on the rest. OS decides what goes where. And in all honestly, the CPU load is low. In Time Spy It does have the drawback that if you have a LOT of cores, the overhead from having a lot of threads can cause the performance to be lower if everything else is equal, but the effect is not large. Just enough to make you wonder why faster CPU can cause lower graphics score using same GPU.

Note that this is specific to Time Spy and older tests. Port Royal and Speed Way do not have this issue.

[BotZero]

the difference is small but its a negative difference.

[Arokhantos]

Just went and clean installed windows 11 barebones again only to score higher again without updates, something really wrong, im not sure whats causing it yet every time i think i found out install more apps i normally run it go's back to its normal score which is 15,5k at stock heck 16k on unleashed bios and 17k almost when doing undervolt and overclock, while all of this makes no difference on my current install consistently 13,5k to 14k score usually 14k

Im gonna test speedway soon see if its also effected i do not think it is, cos score is always consistent

[Gelfan_Dothea]

Run it on Windows 7. Then you'll really be pissed.

[BotZero]

Originally posted by Gelfan_Dothea:

Run it on Windows 7. Then you'll really be pissed.

7 had a good run i stopped using it around 2017

[Gelfan_Dothea]

Originally posted by Akrios:

Originally posted by Gelfan_Dothea:

Run it on Windows 7. Then you'll really be pissed.

7 had a good run i stopped using it around 2017

lol I stopped using it three months ago. If it wasn't for Classic Shell I'd still be using it.

[BotZero]

I clean installed 11 and my scores are still lower. I found the biggest difference in Firestrike where the physics score was 8% better in december 2022 with windows 10 on the exact same hardware.

[BoGy1980 MzegdietsasgedeesleestM]

well, i can say i'm not using win11 because i just don't like microsofts apple-path which they took with W11 (i want my taskbar on the SIDE, not on the bottom where it takes room away from the already smaller side of or widescreen-screens; with a vertical taskbar you can do a lot more with a lot less real-estate)

anyways, when i upgraded my system with a new gpu, 2nd ssd and monitor last black friday, i first did a run with the installed os with the new hardware, and 3DMark told me that I was in the top 15% or the resultas of this configuration. Then i did a complete reinstall of windows 10, brought it to the same "state" i had before (removing all junk MS wants u to have like telemetry, Antivirus, Firewall, ...) and then run a test again... This time 3DMark told me i was in the TOP 1%, and everything was running STOCK SPEEDS! (It's a Strix OC card, so already has a small Factory OC). The old win10 actually was a multi-upgraded windows 7 pro (to 8, to 10), so it had a lot of "garbage" over all those years but still performed quite well (never use any "performance improvement" tools, please, don't... I never did and my system always ran smooth and fast. Another thing i could advice to do (ONLY IF YOU HAVE 32GB or MORE ram) is disabling windows paging (so disable pagefile). You'll notice a big speedbump but lose the ability to hibernate the system (which i should advice not to use btw, as your desktop just re-uses the OLD session)

If windows 12 won't allow me back to put the taskbar where I want (and not where apple/MS decide in my place where it has to be), then W10 is going to be the last Windows i've used... and yes, i'm well experienced enough in linux, runs on all my systems except my daily driver pc, so my laptops are running Linux since 2005, same for my homeserver)