First Windows 11 Patch Tuesday Makes Ryzen L3 Cache Latency Worse, AMD Puts Out Fix Dates

[Wirko]

Dilbert said something half-related.

 

[zlobby]

If you install a new OS when its first released then you deserve all the misery you get imo.

Although I agree with the general notion that everyone is responsible for their own actions (free will implications aside and yes, I see the irony here), I think it's pretty much safe to say that when you pay a hefty sum for a product, you expect the product to be at least half-decent.

 

[Why_Me]

Although I agree with the general notion that everyone is responsible for their own actions (free will implications aside and yes, I see the irony here), I think it's pretty much safe to say that when you pay a hefty sum for a product, you expect the product to be at least half-decent.

If you're new to the computer world then ya I can see someone purchasing a new version of Windows that was just released. The rest of us know better.

 

[R-T-B]

If you're new to the computer world then ya I can see someone purchasing a new version of Windows that was just released. The rest of us know better.

Yeah, you don't pay for this upgrade, for starters.

At least... erm... not conventionally... uhoh... bleh... awwshititshappeningagain...

*SoulSwallowedByTelemetry*

 

[Mussels]

To the people not getting the timeline of how this is really weird:

1. It was reported long ago
2. It was fixed long ago... in the dev channel. BEFORE Win 11's release date.
3. The first official build, brought the bug back.
4. The first official patch made it even worse
5. Alder lake reviews must use W11 (Launch date is November 4)
6. reviews take time. I'd have to ask w1zz, but at least a week of testing for such a big launch sounds about right for him.

Look at how w1zz compares results for years of hardware, by keeping the same testing rig. Throw in a mandatory OS change and he has to retest EVERYTHING

What needs to happen is for the official, working patch to come out before the reviewers start testing alder lake, or all the alder lake reviews will have terrible AMD results.

There seems to be enough time for this to happen, but a lot of us are just wondering if it actually will - many reviewers are doing it as a job with no real passion, and would easily run the earlier gen intels and AMD's as early as possible and just save the results

 

[lexluthermiester]

But the major responsibility has to lie with Microsoft anyway since it's their product and these AMD cpu's are not exactly new.

Yeah, pretty much this.

or Imaginary

And this.

This is pisspoor logic

Exactly!

I'm not buying it.

On this particular issue? I don't care. I WAS thinking the same thing until I dug deeper. This unfortunate problem is coincidence, not conspiracy.

I'm normally a critic of conspiracy theories but this one stinks... bad.

You of all people here should know that digging a little deeper will often reveal the actual reality of the circumstances.

What needs to happen is for the official, working patch to come out before the reviewers start testing alder lake, or all the alder lake reviews will have terrible AMD results.

This. No reviewer can be expected to do an objective set of tests while there is a known flaw in the testing OS. However, as testing between Windows 10 and Windows 11 will need to be done for comparisons, testing in Windows 10 can be done without delay as the flaw in Windows 11 does not translate to 10, at least ATM.

 

[RJARRRPCGP]

5. Alder lake reviews must use W11 (Launch date is November 4)

That's one day before my birthday, LOL! I really hoped that the first patch-Tuesday fixed it. Yawn...

 

[R-T-B]

This unfortunate problem is coincidence, not conspiracy.

I've been following this since the initial reports. It's hardly proof in the literal sense but it does seemingly stink to high heavens (of course such a statement is by nature, individual opinion). I guess the proof will be in when the reviews are made. FWIW, I hope you are right, and will be eager to admit I was wrong if it all goes smoothly. Let's both pray/hope for that!

 

[lexluthermiester]

It's hardly proof in the literal sense but it does seemingly stink to high heavens

I'll agree. The timing is very sus. But once the technical details are taken into account, it becomes clear that it is just unfortunate timing devoid of any agenda. I'm certain that Intel might try to take advantage of it, but then again there is potential for that to bite them in the bum so it's likely that they will simply ride it out.

 

[bug]

...

I'm normally a critic of conspiracy theories but this one stinks... bad.

...

Especially since Windows has a long history of seamless updates. Right?

But yeah, restricting support to a handful of CPUs and then failing to support even those properly is a new milestone.

 

[stimpy88]

On this particular issue? I don't care. I WAS thinking the same thing until a dug deeper. This unfortunate problem is coincidence, not conspiracy.

You of all people here should know that digging a little deeper will often reveal the actual reality of the circumstances.


This. No reviewer can be expected to do an objective set of tests while there is a known flaw in the testing OS. However, as testing between Windows 10 and Windows 11 will need to be done for comparisons, testing in Windows 10 can be done without delay as the flaw in Windows 11 does not translate to 10, at least ATM.

Is Alder Lake, and it's 2 different core types supported in Windows 10? I'm sure Intel said it is not, and has to be used with Windows 11...

I've been following this since the initial reports. It's hardly proof in the literal sense but it does seemingly stink to high heavens (of course such a statement is by nature, individual opinion). I guess the proof will be in when the reviews are made. FWIW, I hope you are right, and will be eager to admit I was wrong if it all goes smoothly. Let's both pray/hope for that!

I worry that many reviewers do not know, or even care about the background on the MS/AMD bugs, and the Intel shills will just not publish which build of Windows 11 they used for testing. We all know that the day 1 reviews will decide the corporate marketing/public perception of Alder Lake, and day 2 "corrections" and "disclaimers" will be ignored by 99% of the media and Intel itself. The damage will be done.

It would be Cricket of MS to patch this bug before Intel sends out the CPU's and Motherboards to the reviewers. But I personally will ignore any review which does not state which build of Windows 11 was used for the benchmarks.

If Alder Lake still manages to beat the currently available Ryzens after they have been patched, then well done to Intel on the win. Then early next year, I will day 1 purchase a Ryzen 3D, and thank Intel for pushing the market forward.

 

[lexluthermiester]

Is Alder Lake, and it's 2 different core types supported in Windows 10?

They will be.

I'm sure Intel said it is not, and has to be used with Windows 11...

Myth. Not everyone is going to switch to 11. Some will continue to use 10 and Intel will need to develop drivers for the new CPU's.

But yeah, restricting support to a handful of CPUs and then failing to support even those properly is a new milestone.

Right? It's hitting an all new level of incompetence... Granted, these are seemingly honest mistakes. But they are mistakes that could have and should have been detected through in-house testing. Allowing them to be released to the public is the incompetent action, not that they exist..

 

[DeathtoGnomes]

not that they exist..

/offtopic

 

[stimpy88]

They will be.

Myth. Not everyone is going to switch to 11. Some will continue to use 10 and Intel will need to develop drivers for the new CPU's.

Myth? I have not seen any evidence of MS updating support for Intels upcoming big.LITTLE CPU's in Windows 10. I just read their Partner Development Manager say that it was a Win11 feature.

"When Microsoft was interviewed on the subject, Mehmet Iyigun – Partner Development Manager, said:

Throughout the development cycle of Windows 11, my team worked with our colleagues at Intel to update and optimize our next operating system to make the most of the Performance Hybrid architecture and Thread Director in particular.
With Thread Director feedback, the Windows 11 thread scheduler is much smarter in dynamically choosing the most appropriate core based on the workload to get the best power and performance."

Now I'm sure support will probably be ported to Windows 10 eventually, as it's basically the same OS, but MS have said nothing about having Win10 support this before or even at the Alder Lake release.

Do you know something the rest of us don't?

 

[lexluthermiester]

Myth? I have not seen any evidence of MS updating support for Intels upcoming big.LITTLE CPU's in Windows 10.

You also didn't see them announce updated support for each gen of Ryzen nor 8th, 9th, 10th &11th gen Core series CPU's from Intel, but they did. Those updates will come in an update as per usual.

Do you know something the rest of us don't?

Maybe? Common sense really..

 

[bug]

You also didn't see them announce updated support for each gen of Ryzen nor 8th, 9th, 10th &11th gen Core series CPU's from Intel, but they did. Those updates will come in an update as per usual.


Maybe? Common sense really..

I will laugh so hard if at the end of the day, this all turns out to be yet another botched driver from AMD.

 

[zlobby]

You also didn't see them announce updated support for each gen of Ryzen nor 8th, 9th, 10th &11th gen Core series CPU's from Intel, but they did. Those updates will come in an update as per usual.


Maybe? Common sense really..

I'd really like to emphasize the 'not supported at this time' part of the compatibility check message. I believe all will be backported to Win 10 at some point.

 

[bug]

I'd really like to emphasize the 'not supported at this time' part of the compatibility check message. I believe all will be backported to Win 10 at some point.

We really don't know what Microsoft wants at this time. Of course they use the most vague phrasing possible, they don't want users running for the pitchforks. But keep in mind Win11 has little to nothing going for it, Microsoft will want to create as much differentiation as possible (or as much as they can get away with). At the same time, they have relented about bigger issues before (see DX12 on Win7).

 

[Luminescent]

That... literally has nothing to do with this?

It does, they focused on other things like security more and less on hardware optimizations.
Better hardware support is not really the goal of Windows 11, the goal is better data collection and security against other nations getting access to the tools.

 

[ThrashZone]

We really don't know what Microsoft want at this time. Of course they use the most vague phrasing possible, they don't want users running for the pitchforks. But keep in mind Win11 has little to nothing going for it, Microsoft will want to create as much differentiation as possible (or as much as they can get away with). At the same time, they have relented about bigger issues before (see DX12 on Win7).

Hi,
Indeed I haven't noticed anything on the positive side on 11
11 needs the same tweaks as 10 did plus more

MS also activated it's torrent features to turn peoples machines and internet bandwidth into their update servers once again same as they did on win-10 early on
Wouldn't doubt if win-10 update delivery settings aren't changed too lol

 

[R-T-B]

the goal is better data collection and security against other nations getting access to the tools.

Yeah, no. And still offtopic. The topic is a Ryzen l3 cache performance issue.

 

[tygrus]

Windows 11 should still be listed as Beta because that's what it is for another year or more.

 

[lexluthermiester]

I will laugh so hard if at the end of the day, this all turns out to be yet another botched driver from AMD.

While that would be amusing, that's just not the problem this time. The more I dig into it(I find things like this fascinating) the more it looks like really low-level kernel nitty-gritty code. One little problem can make the whole thing go wonky. What microsoft is doing is a new way of addressing core scheduling that requires reworking the associated processes like branch prediction and data caching. It's an astrophysics level of complicated.

I think on this one we all should cut microsoft and AMD some slack. They'll get it fixed, but they NEED to be careful and careful takes time.

Windows 11 should still be listed as Beta because that's what it is for another year or more.

While I'd agree to an extend. 11 feels more polished than 10 did after it's beta program. There aren't many issues to fix. The AMD problem is the exception rather than the rule with this release of Windows.

 

[bug]

While that would be amusing, that's just not the problem this time. The more I dig into it(I find things like this fascinating) the more it looks like really low-level kernel nitty-gritty code. One little problem can make the whole thing go wonky. What microsoft is doing is a new way of addressing core scheduling that requires reworking the associated processes like branch prediction and data caching. It's an astrophysics level of complicated.

I think on this one we all should cut microsoft and AMD some slack. They'll get it fixed, but they NEED to be careful and careful takes time.

Yeah, I suspected the new scheduler that has to deal with big.LITTLE is the culprit. I just thought that would be funny since everybody seems hellbent to burn Microsoft to the stake for messing the their preciousss...

 

[lexluthermiester]

Yeah, I suspected the new scheduler that has to deal with big.LITTLE is the culprit.

Nope. This has nothing to do with Intel. AMD's CCX Complex/Infinity Fabric works in a way that is very different and unrelated to how Intels big/little scheme works. Each have CPU Kernel driver code and they are mutually exclusive.

The Kernel driver code for each CPU does not and can not interfere with the code of another.