-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
Rzyen performance - Windows 24H2 vs 23H2 - performance bug fix
- Thread starter winjer
- Start date
lmimmfn
Member
Could have been from a second run when shader compilation is done from the first run.
lmimmfn
Member
I tried this on Windows 10 running with/without admin account, i did 2 runs for each(without admin/with admin - to remove shader compilation issues), results are from second run for without admin account & with
Conclusion: On Windows 10, If you're gaming at 1440 and above with a decent GPU and on Windows 10(cant speak for Windows 11), this is a nothing burger. I gained 1.23 % in Cyberpunk and 0-1% in Shadow of the Tomb Raider.
Those results are within margin of multiple runs without any changes.
Result: On Windows 10 = Waste of time
Setup: 7800X3D, 4070Ti, NVME, and as said Windows 10. Screen resolution was 3440x1440 so GPU is limiting rather than CPU
Conclusion: On Windows 10, If you're gaming at 1440 and above with a decent GPU and on Windows 10(cant speak for Windows 11), this is a nothing burger. I gained 1.23 % in Cyberpunk and 0-1% in Shadow of the Tomb Raider.
Those results are within margin of multiple runs without any changes.
Result: On Windows 10 = Waste of time
Setup: 7800X3D, 4070Ti, NVME, and as said Windows 10. Screen resolution was 3440x1440 so GPU is limiting rather than CPU
Cyberpunk(In game benchmark):
Without Admin:
Avg: 66.70
Min: 33.22
Max: 81.63
With Admin:
Avg: 67.52
Min: 33.79
Max 81.59
Num Of Frames: 4336
Shadow of the Tomb Raider:
Without Admin:
Min CPU Game: 246
Min CPU Render: 141
Min GPU: 96
Max CPU Game: 527
Max CPU Render: 513
Max GPU: 248
Average CPU Game: 358
Average CPU Render: 261
Average GPU: 125
95% CPU Game: 252
95% CPU Render: 141
95% GPU: 102
Shadow of the Tomb Raider:
With Admin:
Min CPU Game: 244
Min CPU Render: 141
Min GPU: 96
Max CPU Game: 522
Max CPU Render: 525
Max GPU: 248
Average CPU Game: 358
Average CPU Render: 263
Average GPU: 125
95% CPU Game: 252
95% CPU Render: 141
95% GPU: 102
Without Admin:
Avg: 66.70
Min: 33.22
Max: 81.63
With Admin:
Avg: 67.52
Min: 33.79
Max 81.59
Num Of Frames: 4336
Shadow of the Tomb Raider:
Without Admin:
Min CPU Game: 246
Min CPU Render: 141
Min GPU: 96
Max CPU Game: 527
Max CPU Render: 513
Max GPU: 248
Average CPU Game: 358
Average CPU Render: 261
Average GPU: 125
95% CPU Game: 252
95% CPU Render: 141
95% GPU: 102
Shadow of the Tomb Raider:
With Admin:
Min CPU Game: 244
Min CPU Render: 141
Min GPU: 96
Max CPU Game: 522
Max CPU Render: 525
Max GPU: 248
Average CPU Game: 358
Average CPU Render: 263
Average GPU: 125
95% CPU Game: 252
95% CPU Render: 141
95% GPU: 102
Last edited:
Yes and CPUs were also affected but only with a much lower chance that it was working, as and controlled full addresses while Intel only checked parts of the address to save performance.
But this was fixed via bios Updates yes ago. I can understand that Ms mitigated this via windows, but not deactivating it when it was fixed for a processor is really odd, especially as it was known that it costs CPU performance.
Von Hugh
Gold Member
The 5800X3D seems to be a gift that keeps on giving:
(https://www.hardwareluxx.de/index.p...rsage-zen-5-mit-windows-11-24h2-getestet.html)
| Gaming performance | W11 23H2 → KB5041587 | note |
|---|---|---|
| Ryzen 9 9950X | +8.8% | 720p +DLSS/FSR avg fps, average 8 game tests |
| Ryzen 7 9700X | +12.1% | 720p +DLSS/FSR avg fps, average 8 game tests |
| Ryzen 5 9600X | +12.0% | 720p +DLSS/FSR avg fps, average 8 game tests |
| Ryzen 7 7800X3D | +6.2% | 720p +DLSS/FSR avg fps, average 8 game tests |
| Ryzen 5 7600X | +7.8% | 720p +DLSS/FSR avg fps, average 8 game tests |
| Ryzen 7 5800X3D | +16.9% | 720p +DLSS/FSR avg fps, average 7 game tests |
(https://www.hardwareluxx.de/index.p...rsage-zen-5-mit-windows-11-24h2-getestet.html)
Last edited:
Xdrive05
Member
BTW, for anyone using 24H2, consider removing Recall.
First time I tried to turn it off, I got a strange error and the window closed. Then I tried a second time and it worked.
YMMV, people!
Rossco EZ
Member
Said it before, the 1080ti of CPU’s
Unknown Soldier
Member
Trogdor1123
Member
Wonder which games they used?
winjer
Gold Member
Optimierte Sprungvorhersage: Zen 5 mit Windows-Update KB5041587 getestet - Hardwareluxx
Optimierte Sprungvorhersage: Zen 5 mit Windows 11 24H2 getestet.
www.hardwareluxx.de
Trogdor1123
Member
Some of those jumps are incredible. I’m really happy with my 5800x3d right now. I ran a 2500k for like decade (and my kid is using it now) so I’d love to get a length like that out of it tooOptimierte Sprungvorhersage: Zen 5 mit Windows-Update KB5041587 getestet - Hardwareluxx
Optimierte Sprungvorhersage: Zen 5 mit Windows 11 24H2 getestet.
Unknown Soldier
Member
Out of curiosity, I downloaded this 23H2 update KB5041587 and ran a benchmark nobody runs
FFXIV Dawntrail Benchmark
Before:
After:
So...about a 3.1% performance uplift. Hey, free is free I guess. Not gonna complain about free
FFXIV Dawntrail Benchmark
Before:
After:
So...about a 3.1% performance uplift. Hey, free is free I guess. Not gonna complain about free
winjer
Gold Member
I tried out a few UE4 games, with Windows 24H2 and found some really good performance gains.
For context, UE4 usually has it's performance limited by a couple of main threads, while the rest of the threads only get minor workloads.
Days Gone, original Outer Worlds, Ghostwire Tokyo, Destroy all humans 2, got 10-20 extra fps.
In the Outer Worlds, in the Byzantium plaza, it used to tank performance into the 60-70s fps. But now it stays in the 80-100 fps range.
My 5800X3D just got a new life.
For context, UE4 usually has it's performance limited by a couple of main threads, while the rest of the threads only get minor workloads.
Days Gone, original Outer Worlds, Ghostwire Tokyo, Destroy all humans 2, got 10-20 extra fps.
In the Outer Worlds, in the Byzantium plaza, it used to tank performance into the 60-70s fps. But now it stays in the 80-100 fps range.
My 5800X3D just got a new life.
Last edited:
winjer
Gold Member
Out of curiosity, I downloaded this 23H2 update KB5041587 and ran a benchmark nobody runs
FFXIV Dawntrail Benchmark
Before:
After:
So...about a 3.1% performance uplift. Hey, free is free I guess. Not gonna complain about free
Do you have VBS off?
Try going to Settings, Core Isolation and turn off Memory Integrity.
So it's a 5% increase in the games they tested. That's still better than nothing. And Hardwareluxx seemed to get better results on the same Windows patch.
Optimierte Sprungvorhersage: Zen 5 mit Windows-Update KB5041587 getestet - Hardwareluxx
Optimierte Sprungvorhersage: Zen 5 mit Windows 11 24H2 getestet.
www.hardwareluxx.de
Edit: HUB confirms they were were testing with VBS off in both cases, so it was still apples to apples.
Last edited:
Unknown Soldier
Member
winjer
Gold Member
Me too, it's pointless for normal users and hurts performance.
Try other games. I found that several UE4 games get some decent performance improvements.
Nikodemos
Member
MS added a bunch of brute-force fixes for hardware security loopholes/exploits, which unfortunately have the immediate side-effect of impacting performance.
winjer
Gold Member
Yes.
FewRope
Member
Time to upgrade to W11 then, the update is not yet officially released right?
winjer
Gold Member
Wait for October-November, that's when 24H2 will be released into the stable channel.
I got some really nice performance gains in many games, but it also has some bugs. I had a couple of crashes. My wallpaper keeps disappearing and repapering with every reboot. BBR2 protocol doesn't work with Steam.
And I saw some people complaining that Star Wars Outlaws doesn't work on this OS.
Last edited:
StereoVsn
Member
God sometimes I can’t stand MS. And half of my day job is on MS platform. The other half is on Amazon’s side and I would rather work with them all day, every day.
Unknown Soldier
Member
TL;DW: Windows 11 is a piece of shit, you can literally fresh install it multiple times and get different performance on Ryzen for reasons which are still completely unknown
Last edited:
Unknown Soldier
Member
Steve at HUB mentioned also there may be a bug where VBS (Memory Integrity) stays turned on and running even when you turn it off, I will eventually circle back and disable SVM (Virtualization) in BIOS which forces VBS off and run this benchmark again
Type_Raver
Member
Whats interesting is seeing that Windows 10 compared to all the versions of 11.
Anyone know how Zen 3 compares with win11 H24 vs win10?
winjer
Gold Member
I never encountered that bug. But Windows 11 has so many, it's very plausible.
Forget about silicon lottery. Now it's all about Windows install lottery.
ap_puff
Member
ap_puff
Member
Where's the 24H2 patch sony
winjer
Gold Member
It wasn't a problem with the CPUs of the AGESA. It was an issue about how the latency was being reported to the software.
AMD Ryzen 9000 inter-core latency significantly reduced with new AGESA 1202 - VideoCardz.com
Intercore latency Ryzen 9000 issue has been reportedly resolved with new AGESA Latency has dropped from 180 ns to 75 ns. One of the issues with the Ryzen 9000 series, as discovered by reviewers, was the relatively high latency between cores. Up until AGESA 1.2.0.1, the latency was measured at...
videocardz.com
ap_puff
Member
If nothing changed besides reporting I'm not sure why people are reporting improvements in benchmark scoresIt wasn't a problem with the CPUs of the AGESA. It was an issue about how the latency was being reported to the software.
AMD Ryzen 9000 inter-core latency significantly reduced with new AGESA 1202 - VideoCardz.com
Intercore latency Ryzen 9000 issue has been reportedly resolved with new AGESA Latency has dropped from 180 ns to 75 ns. One of the issues with the Ryzen 9000 series, as discovered by reviewers, was the relatively high latency between cores. Up until AGESA 1.2.0.1, the latency was measured at...
Last edited:
winjer
Gold Member
What reports of improvements in benchmarks?
marquimvfs
Member
Benchmarks didn't changed, they are oscillating a few percent. What changed was only the latency readings.
marquimvfs
Member
I agree, to some extent. I've read reports that linux latency report wasn't as bad as windows. To me, that is an extension of the problem that AMD was facing with microsoft. Obviously, they could have nailed that before launch, but, I'm still glad that they're fixing things fast enough. And the fixes are improving things instead of tanking performance.
Last edited:
ap_puff
Member
I saw some people reporting 300-500 improvement in cinebench scores on the AMD subreddit. Which is small but nonzero
Last edited:
Insane Metal
Member
Basically, the 9700X is identical (or marginally slower) than the 14700K, whilst using just 40-50% of the power. Also, if you're just intersted in gaming just forget both and go for the 7800X3D.
Last edited:
Insane Metal
Member
Insane value