AMD Confirms Its GPU Drivers Are Overclocking CPUs Without Asking (tomshardware.com) 73
AMD has confirmed to Tom's Hardware that a bug in its GPU driver is, in fact, changing Ryzen CPU settings in the BIOS without permission. This condition has been shown to auto-overclock Ryzen CPUs without the user's knowledge. From the report: Reports of this issue began cropping up on various social media outlets recently, with users reporting that their CPUs had mysteriously been overclocked without their consent. The issue was subsequently investigated and tracked back to AMD's GPU drivers. AMD originally added support for automatic CPU overclocking through its GPU drivers last year, with the idea that adding in a Ryzen Master module into the Radeon Adrenalin GPU drivers would simplify the overclocking experience. Users with a Ryzen CPU and Radeon GPU could use one interface to overclock both. Previously, it required both the GPU driver and AMD's Ryzen Master software.
Overclocking a Ryzen CPU requires the software to manipulate the BIOS settings, just as we see with other software overclocking utilities. For AMD, this can mean simply engaging the auto-overclocking Precision Boost Overdrive (PBO) feature. This feature does all the dirty work, like adjusting voltages and frequency on the fly, to give you a one-click automatic overclock. However, applying a GPU profile in the AMD driver can now inexplicably alter the BIOS settings to enable automatic overclocking. This is problematic because of the potential ill effects of overclocking -- in fact, overclocking a Ryzen CPU automatically voids the warranty. AMD's software typically requires you to click a warning to acknowledge that you understand the risks associated with overclocking, and that it voids your warranty, before it allows you to overclock the system. Unfortunately, that isn't happening here. Until AMD issues a fix, "users have taken to using the Radeon Software Slimmer to delete the Ryzen Master SDK from the GPU driver, thus preventing any untoward changes to the BIOS settings," adds Tom's Hardware.
Overclocking a Ryzen CPU requires the software to manipulate the BIOS settings, just as we see with other software overclocking utilities. For AMD, this can mean simply engaging the auto-overclocking Precision Boost Overdrive (PBO) feature. This feature does all the dirty work, like adjusting voltages and frequency on the fly, to give you a one-click automatic overclock. However, applying a GPU profile in the AMD driver can now inexplicably alter the BIOS settings to enable automatic overclocking. This is problematic because of the potential ill effects of overclocking -- in fact, overclocking a Ryzen CPU automatically voids the warranty. AMD's software typically requires you to click a warning to acknowledge that you understand the risks associated with overclocking, and that it voids your warranty, before it allows you to overclock the system. Unfortunately, that isn't happening here. Until AMD issues a fix, "users have taken to using the Radeon Software Slimmer to delete the Ryzen Master SDK from the GPU driver, thus preventing any untoward changes to the BIOS settings," adds Tom's Hardware.
And here the problem starts (Score:3, Insightful)
We've known for years that the peecee "architecture" is essentially untrustable. And it doesn't get better when manufacturers claim to improve security by taking owner control away.
But this is where the problem starts: That "whoops we totally reconfigured the thing to auto-overheat without permission" is easily possible and not seen as anything more than a silly boo-boo.
Both intel and AMD have simply no concept of respect for buyer ownership.
Blame the fanboys (Score:2, Insightful)
AMD probably feels it's got to keep fanboys satisfied with the hype somehow.
And this is the problem with competing based on hype, Apple being another company with this issue, eventually when you've got nothing worth hyping, you just have to start cheating instead. Case in point, Apple's M1 chips are regularly advertised by Apple as being the most performant AND most energy efficient chips on the market, but the reality is the M1 isn't even close to high end Intel/AMD offerings, it really only competes with
Re: (Score:3)
"Intel made the same mistake, it had a good run when it moved to Core, but then rested on it's laurels and ended up taking shortcuts resulting in security vulnerabilities and so forth."
You got your history wrong. We'd been warning about speculative execution exploits since the original Pentium. They only materialized recently, but as you go back, any system using speculative execution pre-fixes is vulnerable one way or another.
Re: (Score:2)
... the reality is the M1 isn't even close to high end Intel/AMD offerings ...
Tell that to my M1 Max laptop [geekbench.com], which is pushing up against the performance of my 16-core Mac Pro [geekbench.com] from just three years ago. That would be a laptop competing with a relatively recent, fairly high-end desktop.
Mind you, that chip is going on three years old at this point, so it's not as fast as the current top end, but that's also comparing eight high-performance cores against 16. So in terms of performance-per-core, the Apple chip is slaughtering that three-year-old high-end Intel server chip.
it really only competes with their low end - that's what energy efficiency requires, you can't have both maximum performance AND energy efficiency, you have to pick one, and Apple have chosen energy efficiency.
No, you can h
Re: (Score:2)
Geekbench sucks. Go look up how well your precious M1 does in Blender or something like that. Oh look here we go!
https://opendata.blender.org/b... [blender.org]
There's an M1 Ultra losing to a 3950X. Your M1 Max is barely better than a 3700X.
Re: (Score:2)
Geekbench sucks. Go look up how well your precious M1 does in Blender or something like that. Oh look here we go!
https://opendata.blender.org/b... [blender.org]
There's an M1 Ultra losing to a 3950X. Your M1 Max is barely better than a 3700X.
In one narrow benchmark. *rolls eyes*
The whole point of a benchmark suite like Geekbench is so that you can compare the CPU at a wide range of tasks, not just massively parallel algorithms like raytracing.
Re: (Score:2)
Narrow.
Ha ha!
HA HA HA
Okay pal.
Re: (Score:2)
And for the record
Geekbench 5 does a TERRIBLE job of showing you how the CPU performs in a range of tasks.
Re: (Score:1)
If I want a chip clocked beyond it's specifications, then I'll set that up myself, thanks.
I realize that for the pitiful losers who use computers for nothing but games, speed is everything, but some of us need the results to actually be right.
Re:And here the problem starts (Score:5, Informative)
I had a postdoc job where I was doing computational physics on a cluster that a professor had built out of GTX 480's. These things ran hot as hell and were known in the gaming world as not even stable enough for graphics half the time.
Sometimes the jobs would crash and the professor kept blaming my code -- "you must have a race condition somewhere". Nope -- the GPU's were overheating and breaking stuff (apparently this causes segfaults from memory corruption). And when I looked closer, there was more evidence of fuckyness. The GPU's spent most of their time doing iterative linear algebra solves, in which the computer converges over hundreds/thousands of iterations on the solution to a large set of linear equations. Usually the error in its current best-guess solution decreases exponentially with iteration number. But sometimes it would decrease exponentially, then in one iteration spike back up and decrease exponentially again from there.
Of course what was happening was that the GTX480's were introducing arithmetic errors that didn't cause the program to crash -- they just meant that the solver converged from the new "starting point" to a new solution. I didn't notice this at first because these solvers are fairly well tested and we tended to trust that they worked and not look under the hood too much once they were tested, but once I did ... holy hell, those GPU's were a mess.
We replaced them with GTX Titans midway through my job and magically they stopped doing that (and the code stopped crashing).
Especially with Zen (at least mine), there is little performance to be gained by pouring in more power: my laptop's CPU is nearly as fast at 15W as at 35W.
Re: And here the problem starts (Score:2)
AMD drivers for video cards have been crap since the 90's. Thats the main reason i stopped buying amd video cards.
I remember i updated the drivers for the video card and it changed the desktop resolution to 1024x768, but left the monitor resolution at 800x600 so the desktop would scroll side to side and up and down when the mouse pointer got close to the edge of the screen. Couldnt turn it off. So i went back to the stock drivers for it and it was fine.
Re: (Score:2)
Is this Windows or Linux drivers?
Re: And here the problem starts (Score:2)
These were windows drivers back in the 90's (so likely windows 98... or xp)when it was still ATI.
Re: And here the problem starts (Score:2)
Re: (Score:2)
We've known for years that the peecee "architecture" is essentially untrustable. And it doesn't get better when manufacturers claim to improve security by taking owner control away.
peecee is basically that derpy car club in the Wendy's parking lot. You want manufacturers to show owners some respect? Buy more neon light thingies.
Re: (Score:2)
But this is where the problem starts: That "whoops we totally reconfigured the thing to auto-overheat without permission" is easily possible and not seen as anything more than a silly boo-boo.
If it overheats, that's a design flaw. Every CPU has at least one (and possibly more than one) thermocouple or similar to determine CPU temperature. If the CPU overheats, it should throttle itself adequately to prevent hardware damage. The real problem here is that companies like AMD are voiding your warranty for doing something that shouldn't cause damage if their hardware is designed correctly. And I would argue that the fact that their own software is doing this silently is proof that they have concl
Re: (Score:2)
AMD's AM4 platform has prominently featured software-controllable UEFI features since x370 came out in 2017. It's the first time that any part of their software stack has had a bug like this, and it doesn't even affect that many users. I have a Radeon VII and a 3900X, and my system doesn't experience changes to my UEFI settings when use OC profiles (or undervolt profiles) for any game or application. And yes I do have the offending driver in question.
Re: (Score:2)
Please, stop using the internet.
Re: (Score:2)
That's not as easy as it sounds. He tried copying and pasting in real life, but the library charges a dime for every photocopy and they get really angry when he starts gluing things inside the new arrivals.
Re: (Score:2)
P. S. That's not glue.
Open-source drivers too? (Score:3)
Setting stuff in the BIOS without permission (Score:5, Insightful)
sounds as much of a bug as VW optimizing their diesel engines for emissions tests by mistake.
Re: Setting stuff in the BIOS without permission (Score:2)
What VW was doing was (is?) legal in Europe. For whatever reason they just assumed it would be legal in the United States as well. Boy did they get that wrong...
Re: (Score:2)
What VW was doing was (is?) legal in Europe. For whatever reason they just assumed it would be legal in the United States as well. Boy did they get that wrong...
Fraud is legal in Europe? You might want to review the fallout in the various European countries where charges were filed.
Re: Setting stuff in the BIOS without permission (Score:2)
And you might want to do a little more research.
https://euobserver.com/dieselg... [euobserver.com]
Re: (Score:2)
And you might want to do a little more research.
https://euobserver.com/dieselg... [euobserver.com]
That's from 2016.
There have been a multitude of indictments and fines [wikipedia.org] levied against those orgs over the emissions scandals in several countries. Martin Winterkorn [wikipedia.org] is currently a wanted fugitive in the US over the scandal. They are eager to [eventually] hold him accountable, because what they did was clearly not legal.
Re: (Score:2)
That's from 2016.
Yes it is! How observant. It also has something that you really should have read before going ahead and eliminating all doubt about your intelligence:
The report by the German transport ministry said “all manufacturers use defeat devices as per the definition” in the regulation.
Luckily for the manufacturers, the regulation has an exemption from the ban when “the need for the device is justified in terms of protecting the engine against damage or accident and for safe operation of the vehicle”.
The argument that the device is needed to protect the engine is used comprehensively, and is explicitly accepted by the German transport ministry. Their UK colleagues also seemed satisfied with that explanation.
So basically, EU regulators were already aware of this, and deemed it legal. Furthermore, apparently everybody was doing it there, not just VW.
Perhaps because it was legal? Oops. But it gets better! You see, the EU investigators noted this:
There are also cars on the market, for example in the United States, which have an emissions control system that does not have to be switched off. That, Klinger writes, “proves that the use of the switch-off device is not absolutely necessary in order to protect the engine from damage”.
And now, back to your response...
There have been a multitude of indictments and fines [wikipedia.org] levied against those orgs over the emissions scandals in several countries. Martin Winterkorn [wikipedia.org] is currently a wanted fugitive in the US over the scandal. They are eager to [eventually] hold him accountable, because what they did was clearly not legal.
Umm...you do know that you're trying to argue against me because I said it was legal in E
Re: (Score:2)
mainly just that they lied about it to the customers, therefore causing harm to other VW stakeholders
That's the fraud. I don't agree so much with the simple "therefore causing harm to other VW stakeholders" - they caused harm to their customers through their lies. You seem to be quite dismissive of this. Naturally, fraud is extremely difficult to prosecute as you need to prove intent, and the whole thing about maliciousness and incompetence generally having the same symptoms, but I don't think it should be ignored.
But, go ahead and keep smoking that "gotcha!" joint if you want.
Re: Setting stuff in the BIOS without permission (Score:2)
No, that's just you trying to do a Putin to save face. The cheat device, or the "bug" in question, was totally legal. It's not my fault you can't read.
Re: (Score:2)
You talking out of your ass, or lying to seem cool?
VW execs were indicted in German courts by German prosecutors for what was discovered by the US.
What they did was nowhere near legal in Europe.
Re: Setting stuff in the BIOS without permission (Score:2)
You do know that German regulators were already aware that this was going on, right?
https://euobserver.com/dieselg... [euobserver.com]
The argument that the device is needed to protect the engine is used comprehensively, and is explicitly accepted by the German transport ministry. Their UK colleagues also seemed satisfied with that explanation.
Re: (Score:2)
Honestly, it smells like the regulators were corrupt.
They believed that the rules allowed for defeat devices to protect the engine, and thus.... no possible trouble could be had by using one. Even if it was provably not used to protect the engine.
Their assertion was proved false as soon as charges were filed on the matter in Germany.
So you are correct that the German Transportation Ministry was caught fucking the choirboy, but that does not mean it was legal.
Re: Setting stuff in the BIOS without permission (Score:2)
No, it really wasn't a case of that. EU wide regulation very explicitly permitted it. The regulators were operating within the law. Notice it wasn't just the UK and Germany, basically every EU country. There really is no argument at all for protecting the engine. There never was. Why did the EU legislators put that there? I have no idea. If you read that article I linked, you can note how the investigators were looking at the case of cars in the US, and how, other than VW, nobody else was actually doing thi
Re: (Score:2)
You're invoking the behavior of an autocratic regime that ruled by decree to determine what constitutes "law" in Germany today. You couldn't gaslight harder if you tried.
What they did was illegal. The nominal "protection of the engine" poppycock was allowed, but in the US it was proved that the device was indeed emissions defeat hardware.
It was shown that it specifically detect that a test was being ran, and altered engine outputs to defeat the test.
In the UK, th
Re: (Score:2)
You're invoking the behavior of an autocratic regime that ruled by decree to determine what constitutes "law" in Germany today. You couldn't gaslight harder if you tried.
I'll just leave this here.
https://www.britannica.com/top... [britannica.com]
What they did was illegal.
I don't think you really understand what you're even arguing against here. I never argued that VW hasn't broken any laws related to this whole situation. What I'm saying is that the cheat device, in and of itself, was legal. You have not shown anything to contradict that. I know it really upsets your Putin style world view where your favorites can't possibly do anything wrong in your eyes, but nevertheless, here we are, and you're a victim of your
Re: (Score:2)
I'll just leave this here.
I wouldn't have bothered, because the differences in the various law systems don't even factor in here, and your hand-waving claims of otherwise don't change that.
I don't think you really understand what you're even arguing against here. I never argued that VW hasn't broken any laws related to this whole situation. What I'm saying is that the cheat device, in and of itself, was legal. You have not shown anything to contradict that. I know it really upsets your Putin style world view where your favorites can't possibly do anything wrong in your eyes, but nevertheless, here we are, and you're a victim of your own propaganda.
And I'm saying that every jurisdiction that looked at the cheat device said it was not legal.
"A" cheat device isn't inherently illegal. Using one to defeat emissions regulations is.
You wrote:
What VW was doing was (is?) legal in Europe. For whatever reason they just assumed it would be legal in the United States as well. Boy did they get that wrong...
This is inherently incorrect. You're trying to weasel out of what you wrote.
Re: (Score:2)
I wouldn't have bothered, because the differences in the various law systems don't even factor in here, and your hand-waving claims of otherwise don't change that.
Okay, Vladimir.
And I'm saying that every jurisdiction that looked at the cheat device said it was not legal.
You would only be correct if you were speaking about jurisdictions outside of the EU. But we weren't talking about those, were we? I showed you already, and you yourself even linked one time, that European regulators had already deemed them to be legal. I don't think you realize just how big of a conspiracy theory you're creating out of this if every EU regulator was fucking the Choir boy. One of them, maybe. But all of them? Let me guess, you also believe 9/11 was an inside job and that the
Re: (Score:2)
Actually, in addition to my earlier comment, after I dug a little deeper it actually turns out that the VW execs, as well as an Audi exec, were prosecuted not because they broke any laws against faking emissions and exceeding the standards (which, when it comes to diesel vehicles, are actually very low standards compared to the US -- add that to the fact that the EU effectively wasn't even enforcing them.) The reason they were prosecuted was actually due to what amounts to financial crimes. Embezzlement was
Re: (Score:2)
That is a lie. [apnews.com]
What you stated, is indeed bullshit.
VW wasn't raided by German Federal Police for "Financial Crimes".
Re: Setting stuff in the BIOS without permission (Score:2)
Holy fuck do you even read your own links?
The 71-year-old Winterkorn and the others, whose names were not released, face six months to 10 years in prison if convicted of aggravated fraud involving serious losses. Other charges include unfair competition and breach of trust.
Prosecutors said the defendants could also be forced to forfeit sales bonuses ranging from around 300,000 euros to 11 million euros ($340,000 to $12.45 million).
Winterkorn is already under indictment in the U.S. on charges of fraud and conspiracy to violate the Clean Air Act and could get up to 20 years in prison. But he cannot be extradited from Germany to the U.S.
Look at the difference between what he was charged with in the US, and then compare that with what he was charged with in Germany. Notice how that in no way relates to air pollution in Germany, where it does in the United States?
My original post said that the so called "bug" was legal. Not kind of legal, just fucking legal. If it was illegal, they probably would have charged him with that, but they didn't. I already know what was going through your hamster wheel whe
Re: (Score:2)
Fraud, unfair competition, and breach of trust are not a "financial crime".
At least now we know why you're pushing this narrative.
There is no criminal punishment for "violating emissions regulations".
It is tort. The criminal proceedings (in every jurisdiction they occurred under) involve the fraud of attesting that people were following regulations.
Had the "bug" been legal, there would have been no fraud.
You are inventing a narrative here. Lax enforcement is not legality, and every single EU
Re: (Score:2)
I did.
Fraud, unfair competition, and breach of trust are not a "financial crime".
Do you know the difference between perjury and fraud? Do you know what unfair competition does to harm others? Do you know what breach of trust even is? (Hint: It doesn't involve that time you told your mother that you were going fishing, but you were really doing crack.)
Don't answer, I already know: No, no, and no. Here's a hint though: They all involve some kind of illegal personal gain at somebody else's expense. In this case, it's all for financial gain.
But of course, your Russian instincts will take ov
Re: (Score:2)
Meh accidentally hit submit too early.
Had the "bug" been legal, there would have been no fraud.
No, the fraud is actually telling regulators and consumers that their cars meet one set of standards, when in reality they do not, in order to gain from it. That would also apply if, for example, they stated the cars were zero emissions, but they weren't, even though there were no defeat devices involved.
So I'll state it yet again: The defeat software itself was legal. What was not legal was failing to disclose that. Whole different situation from the US: The existence
Re: (Score:2)
Everything is a conspiracy to you isn't it.
Re: (Score:2)
That's deliberate, and has been the case since x370 launched in 2017. What's accidental is the Ryzen Master module in the in the Radeon Software module doing it without prompting.
Great! (Score:2)
If they overclock my CPU I want them! Because I sure couldn't manage :D.
Re:Great! (Score:4, Interesting)
If you want to overclock but not void your warranty, just install this driver. If your CPU gets bricked you can legitimately say you didn't enable overclocking in the BIOS.
Re: (Score:2)
Not really legitimately, after all he did it intentional - albeit by (ab)using an AMD driver.
And he can not prove it, that the driver did it, and not he himself manually.
Re: (Score:2)
If your CPU gets bricked you can legitimately say
If your USD $567 Ryzen 9 5850X CPU gets bricked you can legitimately say...
But who would care besides your wallet?
"voids your warranty" (Score:2)
Single point of failure (Score:2)
Users with a Ryzen CPU and Radeon GPU could use one interface to overclock both.
Once again we see, "Just because we can do this doesn't mean we should do this."
Whoever thought it would be a good idea to combine two disparate systems must have been a programmer because no logical person would have done this. It's like cars with electronic dashes which are linked to everything else in the car so when it has problems your car has problems.
There's a reason systems should be kept separate, and this is a shining
Re: (Score:2)
Re: (Score:2)
Since they have dedicated software for overclocking (AMD's Ryzen Master), why would anyone add this to a device driver instead of allowing the standalone software to interact with both the CPU and GPU overclocking? They say they wanted it to be a single location for users to manage overclocking, but chose to add it to a device driver SMH.
Nothing to see here (Score:5, Informative)
The driver should *not* be changing your BIOS settings without consent.. but this article is trying way too hard to scaremonger over a non-issue.. Precion Boost Overclocking is generally the *default* config on AMD motherboards and does not void any warranty. (That's what gives you the "boost Clock" advertised right on the base CPU Specs. The warranty voiding (which really doesn't) only applies to people manually making changes to the votages and what not.
Re: (Score:3)
The driver should *not* be changing your BIOS settings without consent.. but this article is trying way too hard to scaremonger over a non-issue.. Precion Boost Overclocking is generally the *default* config on AMD motherboards and does not void any warranty. (That's what gives you the "boost Clock" advertised right on the base CPU Specs. The warranty voiding (which really doesn't) only applies to people manually making changes to the votages and what not.
Why are you bringing facts into a nerd fight? We want pillory companies whose products we lust after. It's our constitutional right, per Ben Franklin, at lest that's what I read on the internet.
Re: (Score:2)
Re: (Score:2)
Precion Boost Overclocking is generally the *default* config on AMD motherboards and does not void any warranty.
False. PBO does not default to "enabled" on any motherboard, and enabling it will cause the warranty void warning to come up in BIOS. PBO defaults to "Auto" in BIOSes, and Auto means that it's off by default but can be triggered by software, software such as Ryzen Master which will also tell you that enabling PBO voids your warranty.
AMD is clear about that in it's Terms of Service.
In the future if you're going to try and post facts I suggest you look them up first.
Re: (Score:2)
Even worse, using PBO can cause systems to
Re: (Score:3)
Because Precision Boost Overdrive enables operation of the processor outside of specifications and in excess of factory settings, use of the feature invalidates the AMD product warranty and may also void warranties offered by the system manufacturer or retailer. GD-128
So much misinformation is +5 on this shithole site because fanboys will +1 anything that fits their fucking preconceptions.
Re: (Score:2)
The upmods aren't really the problem, it's the down mods who don't follow the moderation guidelines.
I couldn't disagree more.
Promoting misinformation is harmful.
lots of motherboards have PBO on by default, if that voids warranties and AMD doesn't tell MB makers to put disclaimers on their packaging, AMD should be culpable.
No, they don't.
They do however have it set to "Auto", which as The Internet (TM) has observed, doesn't seem to do anything. You have to explicitly set it to Enabled to get PBO.
Re: (Score:2)
I couldn't disagree more. Trying to hide misinformation instead pointing out how it's wrong only leads to them believing their conspiracy is right. Since it's impossible to stop the spread of misinformation, the only chance of minimizing its damage is through reasoned arguments. What you're advocating is no better than security through obscurity.
Moderation is a double-edged sword.
Ultimately, both of our points are right- and that's simply how it has to be.
What's good for the goose is good for the gander, and also what's bad for the goose is bad for the gander.
I think perhaps what we've stumbled upon here is simply a fundamental limitation of moderation in general.
Re: (Score:2)
As many others have opined, PBO isn't on by default.
Re: The Turbo Button. Stop freaking out. (Score:2)
Re: (Score:2)
Re: (Score:2)
PBO is a feature. However, AMD explicitly says it will void your warranty if it's turned on.
Sauce. [amd.com]
Because Precision Boost Overdrive enables operation of the processor outside of specifications and in excess of factory settings, use of the feature invalidates the AMD product warranty and may also void warranties offered by the system manufacturer or retailer. GD-128
I think probably tons of people turn that shit on (and I probably would too), but either way it's not OK for them to do it for you without big warnings (which you get if you turn it on via explicit avenues)
Minimizing that makes you a shith
Already overclocked? (Score:1)
Re: (Score:2)
If you went into the UEFI and overclocked by hand, this bug can overwrite some of your settings and produce instability, which is how it was first noticed. It does not affect every machine with both an AMD CPU and dGPU.