Question "Ryzen Burnout? AMD Board Power Cheats May Shorten CPU Lifespan" - Tom's

[UsandThem]

https://www.tomshardware.com/news/ryzen-burnout-amd-board-power-cheats-may-shorten-cpu-lifespan

Unbeknownst to you, your motherboard may be silently killing your Ryzen processor faster than expected. HWinfo introduced a new feature today that the vendor says exposes that some X570 motherboard vendors are clandestinely misreporting key measurements to AMD's Ryzen processors, thus boosting performance. Unfortunately, this tactic is similar to overclocking. It results in higher power draw and more heat production, thus potentially killing Ryzen chips sooner than expected – but all without the user's knowledge.

"I'd like to stress that despite this exploit is essentially made possible by something AMD has included in the specification, the use of this exploit is not something AMD condones with, let alone promotes. Instead they have rather actively put pressure on the motherboard manufacturers, who have been caught using this exploit," The Stilt added.

At least there is a free HWinfo tool (linked to at the end of the article) that lets people see if their motherboard behaves this way.

 

[tamz_msc]

That's weird no OC but 999W is ok?

Let's imagine a very bad scenario where the cpu would be hit by some stupid 'power virus' that someone will make in the future, does that mean the cpu is left unprotected from the motherboard?

999W PL2 doesn't mean that the CPU will consume that much.

 

[DrMrLordX]

So is this it? can someone answer me please with the AnandtechForum(R) IfICanTakeYouSeriously(TM) approach ?
thanks
/off

Honestly? Set everything to bone stock default, including your LLC (Auto) and get rid of any voltage offsets. And then run HWiNFO64 beta and see what it says about your %. The OP has an article that shows you what data field to look for, and where to get the beta. Then you will know.

Once again: PL/tau configurability is made *FOR OEMS*.

Since you continue to misapprehend the entire issue at hand, there is no sense in discussing this matter with you. Go have fun making disingenuous arguments with someone else.

 

[MrTeal]

So, if I'd reading this correctly, a number less that 100 means the MB is underreporting the package power, and a number over 100 means it's overreporting the package power.

If that's the case, I'm getting 122% average under Cinebench R15, and 108% under Prime95 blend at @ 60°C temp. I wonder if that's leaving performance on the table?
Seems workload matters quite a bit. With the recommended R20, my board X570-P reads 95.6% average. Nothing to see here.

 

[RetroZombie]

The usual comment is that the chipset pulls more. Or RAM. Or something.

And it does, the problem is right here:



The X570 motherboards by having 'two' duplicated chipsets, the idle power draw is much higher than it should, so on full power the same occurs, right?
Just to see the difference a potato A320 does ~22W idle and one X570 will do ~38W.

 

[Iron Woode]

So, if I'd reading this correctly, a number less that 100 means the MB is underreporting the package power, and a number over 100 means it's overreporting the package power.

If that's the case, I'm getting 122% average under Cinebench R15, and 108% under Prime95 blend at @ 60°C temp. I wonder if that's leaving performance on the table?
Seems workload matters quite a bit. With the recommended R20, my board X570-P reads 95.6% average. Nothing to see here.

which is identical to mine.

 

[Feld]

Just tested my setup (in sig) with everything at default BIOS settings (9/9/2019 BIOS). It stayed right around 31% deviation the entire time running Prime95 blend for 45 minutes. Looks like the Asrock Taichi X570 is guilty of juicing the CPU. I don't know how much power the CPU specifically is drawing, but as reported by my UPS the difference between system idle and system load during that Prime95 run was 116 W. Though for a 105 W TDP CPU on water with a ~31% reporting deviation, that difference between idle and load seems smaller than I would expect.

The attached screenshot is from just before I stopped Prime95.

 

[Phynaz]

Remember when The Stilt used to post technical information here?

Then the AMD fanboys chased him off.

Fanboys are a pestilence.
AMD's fanboys unfortunately are more ignorant, toxic and feeble-minded than all of the rest combined.

Sad to see that this site has gone down the drain as well.

Goodbye.

So you get an old "fanboy post" from the past to be able to use it yourself, when you know its not allowed?


esquared
Anandtech Forum Director

 

[maddie]

Remember when The Stilt used to post technical information here?

Then the AMD fanboys chased him off.

And your point is?

 

[Markfw]

And your point is?

He doesn't really have a point. He is trolling. Let him rant... Won't be for long.

 

[Arkaign]

Well, this seems like something that is very good to come to light so it can be stamped out where it can be, and it does seem like more recent bios updates for some mobos have reduced or eliminated this situation.

It's interesting to me that it follows a similar trend with Zen2 overall : AMD did a masterful job of making them run REALLY well at efficient stock settings. Most OC attempts, added voltage, trying to get clock speeds to rise end in either extremely little gains, split gains (better MT, worse ST or vice versa), or in some cases performance regression, as the sensitive and dense 7nm dies tend to hotspot and bog when unwisely pushed.

Mobo makers screwing this up did favors to absolutely nobody. I wish I knew about this before. I've had 3600 builds that ran cool under DH12s, and 3600 builds that ran into the 80s under the same HSFs, which made no sense to me at the time because it was always just stock without voltage, PBO, OC etc adjustments. Now I know it was probably some rogue Mobo tomfoolery to blame.

Fixing this will be a great thing for those affected by it. AMD not to blame here, old Mobo MFG tricks at work trying to get an upper hand probably, yet in this case it was counterproductive.

 

[DrMrLordX]

@Feld

Interesting that your board is guilty. Do you have the ability to test points on your board with a multimeter?

 

[BTRY B 529th FA BN]

Could the quality of the VRM and the design make a difference in the return values? I ended up choosing a board with a true 14+2 VRM (no doublers) Gigabyte's X570 Aorus Xtreme. From my returns I'm getting 90%

 

[lobz]

Remember when The Stilt used to post technical information here?

Then the AMD fanboys chased him off.





So you get an old "fanboy post" from the past to be able to use it yourself, when you know its not allowed?


esquared
Anandtech Forum Director

Even though I admire his technical knowledge, that doesn't make him any more qualified to make that statement than any of us about anybody here. Clearly he was frustrated and chose to let the steam out this way. You literally chose the worst moment of a respected forum member and used it to underline your agenda, never even caring about putting him in the worst light possible. That's about on the same level of fallacy as when Tom's using The Stilt's name and publishing an article stating that AMD motherboards are killing CPUs left and right...

 

[DrMrLordX]

Could the quality of the VRM and the design make a difference in the return values? I ended up choosing a board with a true 14+2 VRM (no doublers) Gigabyte's X570 Aorus Xtreme. From my returns I'm getting 90%

I'm guessing not. At least not based on what The_Stilt posted.

 

[RetroZombie]

Remember when The Stilt used to post technical information here?

Then the AMD fanboys chased him off.

Is Dr. Ian Cutress a fanboy to you?

«AMD doesn’t necessarily need to comment on the matter, as this is an issue with the motherboard manufacturers. Users might want to probe their motherboard manufacturer, and ask for a BIOS update.»

999W PL2 doesn't mean that the CPU will consume that much.

Reading your reply i think it's even worst than i though.
It's not only the cpu that is left unprotect from the motherboard, it's also the motherboard that is unprotected from the cpu because of that!
And toms writing an article about amd boards behavior, here's the major intel 'crime'! I wonder why intel is not stopping this!

 

[rbk123]

No doubt Mobo manufacturers have already ceased doing this since Bob Swan said the industry needs to stop using CPU benchmarks. Good ol' Bob to the rescue.

 

[Feld]

@Feld

Interesting that your board is guilty. Do you have the ability to test points on your board with a multimeter?

No, unfortunately I don't have a multimeter.

 

[BTRY B 529th FA BN]

No doubt Mobo manufacturers have already ceased doing this since Bob Swan said the industry needs to stop using CPU benchmarks. Good ol' Bob to the rescue.

Bob was like "We talkin about Benchmarks. Benchmarks? Is that what we talkin about? Benchmarks? Man we talkin about benchmakrs"

 

[DAPUNISHER]

Gamers Nexus backs up the doc's conclusion. Concludes by stating your CPU would be more affected by moving across country, slightly higher ambient room temp, that kind of thing. That any degradation after 3 years would be undetectable outside of a lab setting.

 

[Arkaign]

Gamers Nexus backs up the doc's conclusion. Concludes by stating your CPU would be more affected by moving across country, slightly higher ambient room temp, that kind of thing. That any degradation after 3 years would be undetectable outside of a lab setting.

This jibes with my experience, where apparently affected setups were pushing out unexpectedly high levels of heat, but protective systems to prevent damage were still operating normally. The stupidest thing about this situation is that there was no real benefit to the trickery, it just wastes power and adds heat for no gain.

Not dangerous really, just incredibly dumb. At least the fixes thanks to this being uncovered will roll out and fix the issue, which already seemed fixed on many boards already.

 

[Hayabusa Rider]

So my tuf gaming x570 with a 3700x shows a minimum of 218%. What does that mean?

 

[Elfear]

So my tuf gaming x570 with a 3700x shows a minimum of 218%. What does that mean?

Are you sure you followed the recommended testing parameters? That seems way outside the range of what is possible if tested correctly.

 

[Arkaign]

So my tuf gaming x570 with a 3700x shows a minimum of 218%. What does that mean?

My understanding is that only a reading during the all core load such as CB20 run will offer useful results. A reading of 50% indicates a doubling of expected potential power disparity to the CPU (not necessarily reflective of how much power IS going to the CPU, but as seen by GN it does indeed mean that some excess is seen, which increases heat).

Numbers above 100% indicate that you haven't achieved full load during testing, or potentially you have some situation where you are experiencing a little less than half of the expected normal value of power delivered to the processor, or possibly an error in the program or sensor readings. If you have a kill-a-watt or similar external system power consumption tool, you can cross-reference results from people with similar configurations to get a greater understanding of just what is going on there.

If your rig is operating well and not producing excessive heat, I'd say you're probably in the clear. It looks like most of these issues have been cleared up with bios updates along the way, but if you want to give it another go, perhaps try a series of aggressive all core loads to see.

 

[Hayabusa Rider]

My understanding is that only a reading during the all core load such as CB20 run will offer useful results. A reading of 50% indicates a doubling of expected potential power disparity to the CPU (not necessarily reflective of how much power IS going to the CPU, but as seen by GN it does indeed mean that some excess is seen, which increases heat).

Numbers above 100% indicate that you haven't achieved full load during testing, or potentially you have some situation where you are experiencing a little less than half of the expected normal value of power delivered to the processor, or possibly an error in the program or sensor readings. If you have a kill-a-watt or similar external system power consumption tool, you can cross-reference results from people with similar configurations to get a greater understanding of just what is going on there.

If your rig is operating well and not producing excessive heat, I'd say you're probably in the clear. It looks like most of these issues have been cleared up with bios updates along the way, but if you want to give it another go, perhaps try a series of aggressive all core loads to see.

Let me look for that.

 

[Iron Woode]

So my tuf gaming x570 with a 3700x shows a minimum of 218%. What does that mean?

before the run of cinebench r20, make sure to reset the values in HWinfo. I paid most attention to the average.