AMD Ryzen 9000 Series "Zen 5" Tested on Windows 11 24H2 Update, Shows Significant Performance Gains (2024)

bugRight. So Zen5 still isn't significantly faster than Zen4, only more power efficient.

If you need AVX512 you take Zen5, that's pretty much it. As linux users, you already noticed that we didn't care for this batsh*t crazy show... AMD preformed better with each kernel release, even 6.11 will bring boost even for older CPUs.

Ferrum MasterWINTEL has never ceased to be, what's with the surprises?

the conspiracy is that intel had like 4x more people working on getting their stuff work optimally than amd does.

sneaky sneaky Microsoft and intel keeping amd down with their putting more effort behind it.

shame he didn’t do a reference test with an intel machine to see how that does under the patched up windows

bugRight. So Zen5 still isn't significantly faster than Zen4, only more power efficient.

So people got what they were clamoring for since years... Comment sections around the world were filled with complaints that CPU/GPU manufacturers race for MHz and forget about efficiency, and how "everyone" would forgo massive performance jumps for good efficiency boosts. Fast forward to 2024 when AMD did just that (accidentally, after announcing massive performance boosts too, and partially failing to deliver) and people are unhappy that the CPU is "just" more efficient.

30+% perf. gain in some games ... Or in other words, nearly 1/3 of perf. loss just because of Windows not being optimized properly ...
You even pay for Windows and still get this. What do you pay for, moving taskbar icons from left to center, rounding window corners, and showing ads?

W1zzard, will you re-test 9700X in games and 9950X in applications with Win11 24H2 update?
I know you have lot of work to do but it'd be really good to see the difference (former review vs. newer one).

Anyway, does this "admin bug" also affect Windows 10?

bugRight. So Zen5 still isn't significantly faster than Zen4, only more power efficient.

LittleBro30+% perf. gain in some games ... Or in other words, nearly 1/3 of perf. loss just because of Windows not being optimized properly ...
You even pay for Windows and still get this. What do you pay for, moving taskbar icons from left to center, rounding window corners, and showing ads?

Yeah but ... is there only MS (with Intel's help) to blame? It looks like AMD launched Zen 4 processors, then went into a sleep state for two years. Regardless of the fact that the same chips are used in Epyc and TR series!

Of course I'm assuming here that the crucial parts of the Windows scheduler are written by AMD/Intel/Qualcomm as some kind of "CPU drivers". MS just gets to decide if and when they are included in Windows updates. Is that not the case?

I'm a bit dissapointed. I read Windows 11 24H2. My windows did not update to that Version recently. Everyone claims to be using a newer windows Version which my box does not update itself to with the build in Windows 11 Pro "Windows Update Option".

I updated a few days ago. Yesterday I checked with the command "winver". I'm stuck at another version. Windows 11 Pro had here the problem, I had to upgrade the version manually, last time, around October 2023-February 2024. Could be the issue that I do not use a Microsoft Email adress for the windows login option, secure boot or TPM.

According to 20 minutes websearch yesterday, how to upgrade to Windows 11 Pro 24H2.

• windows insider -> which most likely needs a microsoft accoutn and you are instantly loocked out or locked in with that email adress - which is bad in my point of view
• download an ISO -> Not sure if that ISO is already available to the public -> there are different informations for that. -> still unclear if this is a safe and official iso or not.
• different websites claim, W11 24H2 may be released in September or October. I assume year 2024.

LittleBroAnyway, does this "admin bug" also affect Windows 10?

Yes - I read it somewhere. It affects different processor "versions" and w10 and w11. It is a generic hidden performance penalty bug.

It seems to be this - I have not tested it - check before you use it: www.neowin.net/news/windows-hidden-admin-account-apparently-boosting-amd-ryzen-90007000-performance/
or www.howtogeek.com/962/enable-the-hidden-administrator-account-on-windows-vista/

For those wondering, the Administrator Account on Windows is generally hidden but can be accessed by using the command:

net user administrator /active:yes
It has one of the highest privileges in the Windows environment, surpassed only by the SYSTEM Account.

kondaminthe conspiracy is that intel had like 4x more people working on getting their stuff work optimally than amd does.

Especially those bribing OEMs to purchase only INTEL CPUs... but... unfortunately that's no conspiracy.

AusWolfWhat about X3D models? What about Intel? I think this needs some more testing.

Also, what about the small minority of people who don't have a 4090? Do they see any improvement at all?

At least they're using 1080p instead of 720p for this comparison I guess.

What this mainly shows is how poor Windows is generally these days.

FoulOnWhiteInstalled 24H2 last night, shame i don't have a AM5 setup though. I don't suppose 24H2 makes any difference to a non AMD setup.

HWU did try it on Intel 14600K with like 4 games. Borderlands 3, Cyberpunk 2077 Phantom Liberty and Far Cry 6 showed no change. In Gears 5 where Zen4/5 had a big uplift, 14600K also got a 25% performance boost.
Very curious about the X3D CPUs as well though.

Patiently waiting to see other benchmarks how it affects the X3D CPUs and Intel's Alder and Raptor Lake CPUs, if performance gains across the board, would be very nice. Also gotta love all the conspiracy theories popping up around the web in light of this claiming that Intel has been secretly paying microsoft to intentionally gimp the performance of AMD CPUs lol.

I am still on Windows 10 22H2 though and planning to stick with 10 until end of support because i never liked 11 to be honest, so I'll be missing out on the gains. Sadge

WirkoYeah but ... is there only MS (with Intel's help) to blame? It looks like AMD launched Zen 4 processors, then went into a sleep state for two years. Regardless of the fact that the same chips are used in Epyc and TR series!

Of course I'm assuming here that the crucial parts of the Windows scheduler are written by AMD/Intel/Qualcomm as some kind of "CPU drivers". MS just gets to decide if and when they are included in Windows updates. Is that not the case?

It's similar problem to what Intel had with launch of Alder Lake on Windows 10.
Windows was unable to utilize new architecture properly, it was treating all cores as if they were the same, sometimes putting heavy load on e-cores.
Did Intel have time to work with Microsoft prior to launch of Alder Lake? Definitely. Did Microsoft have time to modify scheduler mechanism? Definitely.
Same applies to AMD's Zen 5 launch.

What makes a difference, however, is that Zen 5 worked on Linux properly out of the box. How come?
(Alder's Lake change was a big one, so Linux needed some time to get adapted, but it did.)

Now another set of questions:
Does MS actually care that their system does not utilize their partner's newest CPU architecture properly and puts heavy load on wrong cores? No. They, maybe they don't even know.
Does MS actually care that the same hardware performs better on a free OS and sometimes even when under emulation? No. Because MS rules 'em all. What's Linucs, anyway?
Do AMD and Intel care that their hardware performs worse in Windows? Yes, they must care, because Windows is an OS which is used by vast majority of systems that utilize their CPUs.

MS is (and always have been) like: "We are the leaders, we create new things. Here are the requirements, go get used to it." Examples: Windows OS, DirectX, Copilot, MS Store, ...
But the flow may go as well the other way. Sometimes it's OS that needs to adapt. Linux prepares kernel for upcoming new interfaces and changes in architecture (mostly) in advance. And it's a free OS.

The funny thing about it is that MS is literally done without AMD's/Intel's x86. So, MS should really care, too. Zen 5 architecture will be applied to EPYCs which run their Azure and other server stuff.
If not for that, then be it their commitment because people pay for that piece-o-sheet OS.

What really makes me think that this is mostly on MS is that it's all connected to some hidden admin account with some special OS privileges.

LeshaI hope Win10 will get a similar update... And I wonder how it affects performance in applications.

Win10 updates will expire in October 25, so I doubt it will get an update.

LittleBroIt's similar problem to what Intel had with launch of Alder Lake on Windows 10.
Windows was unable to utilize new architecture properly, it was treating all cores as if they were the same, sometimes putting heavy load on e-cores.
Did Intel have time to work with Microsoft prior to launch of Alder Lake? Definitely. Did Microsoft have time to modify scheduler mechanism? Definitely.
Same applies to AMD's Zen 5 launch.

What makes a difference, however, is that Zen 5 worked on Linux properly out of the box. How come?
(Okay, Alder's Lake change was a big one, so Linux needed some time to get adapted, but it did.)

Alder Lake was not the first heterogeneous x86 CPU. Intel broke its teeth with Lakefield in this regard. While Lakefield was not widespread, it was a production CPU and Intel was heavily involved with MS trying to optimize for different cores. Lakefield was launched more than a year before Alder Lake and Intel was actively working on this even before Lakefield launch.

These are not separate developments either. I think AMD was actually first to start on the core distinction part with the Preferred cores.

Linux usually "works properly" for either of two main reasons - manufacturers directly contribute code to it or the optimization target is different. Plus, Linux is not or does not have a single optimization target. Just look at the Phoronix' Zen5 stories and the difference between Clear Linux and Ubuntu.

MS cares about being optimized enough. However when it comes to utilizing new CPU architecture properly it is very difficult for OS developer to know what exactly "properly" is :)

LeshaI hope Win10 will get a similar update... And I wonder how it affects performance in applications.

Windows 10's final version is 22H2 and support will end on Oct 2025, so i doubt they'll be bringing this kind of update.

LittleBrois that Zen 5 worked on Linux properly out of the box. How come?
(Okay, Alder's Lake change was a big one, so Linux needed some time to get adapted, but it did.)

Ever hear of Servers OS and EPYC and the importance of that?

londisteAlder Lake was not the first heterogeneous x86 CPU. Intel broke its teeth with Lakefield in this regard.

And that early MS+QC Windragon. I don't remember the name but at the time, I was confident that MS had used that heterogeneous processor to sharpen its teeth.

Ferrum MasterWINTEL has never ceased to be, what's with the surprises?

The surprise?

Show me the last time a new Windows version caused such gaming performance uplift, regardless of CPU brand.