AMD Isolates Windows 11 and Windows 10 Performance Stuttering Issues to fTPM

[btarunr]

Does it take ages for the taskbar calendar and notification center to load on your Windows 11 PC powered by an AMD Ryzen processor? Notice random stutters in performance? Chances are, the lag is caused not due to user-interface bugs by Microsoft, but hardware. AMD discovered that certain Ryzen-powered Windows 11 and Windows 10 PCs experience intermittent performance stutters when running with fTPM (firmware TPM) enabled.

The performance stutter is caused due to background memory transactions between Windows and the fTPM, to authenticate an action, as the fTPM serves the function of a hardware root of trust. Since the fTPM is part of the UEFI firmware that resides on the SPI flash EEPROM chip, the performance stutter is caused due to fTPM-related memory transactions with this chip.



AMD issued an immediate workaround, as well as announced that it's working on a fix. As a workaround, you can switch from fTPM to a discrete TPM module (or dTPM), which uses the TPM 2.0 header on your motherboard. dTPMs such as the one pictured above, have been selling on Amazon for anywhere between $50-100. Be absolutely sure to disable Bitlocker before switching between fTPM and dTPM, if you have it enabled. Or you could just wait for AMD's fix, which will be distributed by motherboard or OEM vendors, as UEFI firmware updates.

AMD expects firmware updates with the fix to start coming out around May 2022. These will use the AGESA V2 ComboPI 1.2.0.7 (or later) microcode. The latest version of AGESA in distribution is 1.2.0.6b.

View at TechPowerUp Main Site | Source

 

[Cutechri]

Cannot wait to get Nova Lake in however many years time. Tired of seeing too much of this stuff from AMD.

I'm not buying a scalped dTPM just to fix this, AMD. Get your shit together.

 

[Ruru]

Nothing to complain here, works like a charm. R5 3600, B550 & Win11 Pro.

 

[tabascosauz]

The same reddit thread/youtube video where we were all introduced to the cause of this lovely bug stated that users have not had much luck with discrete TPMs either..........

But at least AMD acknowledges it now (I mean, the video evidence is pretty clear). So we frustrated owners won't just sit there talking to ourselves like a bunch of bozos while other Ryzen owners chime in with their "no problems here on my PC, so this bug doesn't exist".

When was the last time AGESA actually worked without crippling performance or removing PBO functionality? AGESA 1202? 1203 from mid-2021? If nothing else, hoping the 5800X3D launch blesses us with a good AGESA 1207. Or maybe not.

 

[ThrashZone]

Hi,
Interesting some people say they have stuttering and some say they have none.
Guess we see whom was right

 

[Cutechri]

I definitely have many to complain about. This is an issue that is prevalent on my machine. Not to mention all the other issues AMD have been screwing my head with.

 

[ThrashZone]

Hi,
Driver issues is one of the main reasons I've never gone amd systems.

 

[Cutechri]

Not really the drivers and moreso hardware issues. My old 3900X rebooting at idle due to not handling stock VDDG CCD/IOD voltages well is sure fun...

 

[ThrashZone]

Hi,
Drivers/ firmware not to far apart

 

[Cutechri]

hoping the 5800X3D launch blesses us with a good AGESA 1207. Or maybe not.

Yeah... I'm avoiding 1.2.0.5 and .6 after everything I've heard about them. Hope 1.2.0.7 is good, because I will be updating in order to fix this annoying fault. If not, then well, I'd have to use BIOS flashback on my board for the first time ever.

 

[samum]

Meanwhile, those of us that bypassed TPM at install have no problem with this.

 

[tabascosauz]

Yeah... I'm avoiding 1.2.0.5 and .6 after everything I've heard about them. Hope 1.2.0.7 is good, because I will be updating in order to fix this annoying fault. If not, then well, I'd have to use BIOS flashback on my board for the first time ever.

I think you can still disable fTPM on 11? All that it ever did for me was require me to change my password at the next boot. Admittedly I do use a MS account for login and do not use Bitlocker, both of which will probably change how it behaves. Obviously 10 is easy, but I think it also requires a password change if you installed with TPM on.

 

[Cutechri]

I think you can still disable fTPM on 11? All that it ever did for me was require me to change my password at the next boot. Admittedly I do use a MS account for login and do not use Bitlocker, both of which will probably change how it behaves. Obviously 10 is easy, but I think it also requires a password change if you installed with TPM on.

Probably. I'd rather try out the official fix though

 

[eidairaman1]

No performance issues on a 5800 b550 Steel Legend using ftpm

 

[DemonicRyzen666]

Mean while, pirates win again. All jokes a side I don't like the idea of TPM
I've heard that some new games are to be locked to a specific PC meaning you can't install it on any other.

 

[ThrashZone]

No performance issues on a 5800 b550 Steel Legend using ftpm

Hi,
Maybe asus did some mojo to the bios

 

[zlobby]

Odd, no issues on anything from 2700U (mobile) to 5xxx (desktop). All have virtualization security enabled.

I finally took a leap of faith and migrated most personal machines on Win 11. So far, so good.

 

[AnarchoPrimitiv]

Hi,
Drivers/ firmware not to far apart

I'd really like to know what these "issues" are, I built a 2700x/5700x system in 2019 and I've literally, and I mean literally, never had a single issue, not one single black screen, not one single crash, not one single issue, and I mean that in the most literal sense. I have numerous games installed, use various content creator software, and have never had an issue, am I just lucky?

 

[EatingDirt]

The same reddit thread/youtube video where we were all introduced to the cause of this lovely bug stated that users have not had much luck with discrete TPMs either..........

But at least AMD acknowledges it now (I mean, the video evidence is pretty clear). So we frustrated owners won't just sit there talking to ourselves like a bunch of bozos while other Ryzen owners chime in with their "no problems here on my PC, so this bug doesn't exist".

When was the last time AGESA actually worked without crippling performance or removing PBO functionality? AGESA 1202? 1203 from mid-2021? If nothing else, hoping the 5800X3D launch blesses us with a good AGESA 1207. Or maybe not.

People on reddit are a special type. However, that being said, the TPM isn't an issue on every configuration (like mine). So, if it doesn't happen on every instance of people using TPM with a Ryzen CPU, it can be tough for people on a forum to pointpoint the exact cause unless that forum happens to be full of hardware/software engineers.

 

[tabascosauz]

People on reddit are a special type. However, that being said, the TPM isn't an issue on every configuration (like mine). So, if it doesn't happen on every instance of people using TPM with a Ryzen CPU, it can be tough for people on a forum to pointpoint the exact cause unless that forum happens to be full of hardware/software engineers.

To be fair, I'd usually wholeheartedly agree. Reddit is full of "experts" who "troubleshoot" by not troubleshooting.

In this case, the fTPM-related stuttering is very distinctive. It does not affect anything 99% of the time; it usually lasts only a few seconds, and it's easy to miss if you're not continuously focused on your PC (ie. gaming, media, audio). But when it happens it is application-agnostic, very noticeable across the board, and it's nigh impossible to confuse it with any other sort of hardware/software issue.

In the first few months I was clean installing all the time back and forth between 10 and 11, trying to figure out the L3 problem and how Win 11 scheduler clocked in games. Ran a bunch of different boards in that period with a lot of B-die tinkering. The only constant was the fTPM problem, and the difference is indisputable once aware of what's causing it. Detailed this a bunch of times elsewhere so I won't go into it, but if there's just one time that some random Redditor actually made some sense, it's this one.

 

[Cutechri]

I'd really like to know what these "issues" are, I built a 2700x/5700x system in 2019 and I've literally, and I mean literally, never had a single issue, not one single black screen, not one single crash, not one single issue, and I mean that in the most literal sense. I have numerous games installed, use various content creator software, and have never had an issue, am I just lucky?

Congrats. You didn't have issues, doesn't mean the rest of us didn't either. This point is moot from the get go.

 

[mechtech]

Well at least it can be fixed.

Cannot wait to get Nova Lake in however many years time. Tired of seeing too much of this stuff from AMD.

I'm not buying a scalped dTPM just to fix this, AMD. Get your shit together.

If it’s in the firmware. Wouldn’t that also be tied to ami/award/whoever makes motherboard bios also??

 

[ThrashZone]

Hi,
Yeah isn't this just a micro code like intel does ?

 

[Wirko]

The performance stutter is caused due to background memory transactions between Windows and the fTPM, to authenticate an action, as the fTPM serves the function of a hardware root of trust.

Am I the only one wondering what these "actions" are?

 

[Cutechri]

If it’s in the firmware. Wouldn’t that also be tied to ami/award/whoever makes motherboard bios also??

idk. All I know is Intel users aren't having to deal with this