Threadripper Overclocking Blows a Hidden Fuse, AMD confirms: Warranty not Voided

[ir_cow]

What do Threadripper and Ryzen have in common? AMD. Think about it for a second.

 

[Count von Schwalbe]

What do Threadripper and Ryzen have in common? AMD. Think about it for a second.

On that note, is there any information/confirmation that those are NOT in Ryzen products?

 

[FoulOnWhite]

Don't overclock a TR CPU simples

 

[demu]

Hi,
Likely just a quick sign so they don't have to dig deeper
Replace the fuse case over
Only question is will they charge for replacing the fuse or not.

Yeah, replacing the fuse is quite an easy task.

(Beware the freaks, though)

 

[Frick]

Surely you can just pencilmod it back to working order.

 

[Count von Schwalbe]

On that note, is there any information/confirmation that those are NOT in Ryzen products?

Further to this, where is the fuse located? CCD? IOD? SMD on the package? That could make a difference, for replaceability and whether it is in other CPUs that use the same CCD.

 

[LabRat 891]

1st take:
This is 'research' for whether the bad press from denying RMAs for using advertised features* outweighs the savings to the bottom line.


*admittedly, this is more an Intel thing @TM.

 

[FoulOnWhite]

I doubt very much replacing the fuse is in the realm of a normal person, unless you have access to very special equipment, and even then, is it actually replaceable @W1zzard

 

[R-T-B]

Hi,
You mean they can't just pop the hood damn lol

Welcome to the world of efuses.

 

[thesmokingman]

Having actually built TR stations, no one is really overclocking these. The reality of trying to cool 32 to 64 cores is a limiter. And really trying to run an all core OC is the height of folly.

 

[FeelinFroggy]

Is overclocking a Threadripper CPU really necessary? I'm curious how many owners of these CPUs overclock them, especially given their price. For starters, overclocking these days brings very little benefit. The chips overclock themselves now and they can milk just about as much as they can out of the chip. And on top of all that, getting it stable considering all of the cores on a TR is really just a waste of time for little gain. But I'm not a prosumer and maybe it will save them many hours by squeegeeing out a couple more MHZ.

 

[thesmokingman]

Is overclocking a Threadripper CPU really necessary? I'm curious how many owners of these CPUs overclock them, especially given their price. For starters, overclocking these days brings very little benefit. The chips overclock themselves now and they can milk just about as much as they can out of the chip. And on top of all that, getting it stable considering all of the cores on a TR is really just a waste of time for little gain. But I'm not a prosumer and maybe it will save them many hours by squeegeeing out a couple more MHZ.

No it's necessary. Overclocking these chips is ridiculously hard and expensive and is kind of dumb given these are usually money producing machines. Overclocking these is counter to the high uptime and profits.

 

[95Viper]

Discuss the topic and not each other.
Stop the insults, too.

 

[tabascosauz]

Further to this, where is the fuse located? CCD? IOD? SMD on the package? That could make a difference, for replaceability and whether it is in other CPUs that use the same CCD.

I'm pretty sure these fuses are not nearly the size or form factor you're thinking of this isn't Slot 1

It's been rumored a long time that AMD has had these built in hardware checks since they moved to chiplets, if not before. It just never mattered because:

• AMD's RMA policies are pretty lax and inconsistent as hell
• The check in AGESA has never implied any permanent changes and still comes up every time even if you accept

The statement from the AMD rep is largely in line with the stance they already take to RMAs. AMD's well aware that overclocking is a key universal selling point of AM4/AM5 - they can get away with a lot of things on something much less popular and receiving less attention like TR platforms.

If anything, their RMA dept is more focused on feeding you the same canned response in hopes that you'll go away instead of RMAing for specific issues (e.g. reboots); if you come to them with an outright dead or defective chip, to which their canned response doesn't apply, they don't usually put up much resistance in my experience.

 

[Count von Schwalbe]

I'm pretty sure these fuses are not nearly the size or form factor you're thinking of this isn't Slot 1

I have no idea what size they are; I know there are some SMD components on the package (hence the odd shape of the Ryzen 7000 IHS) but I do not know if they are resistors, capacitors, fuses, magic smoke containers, or nuclear reactors. If they are part of the die, then which die is it? The CCD is universal - used in Ryzen as well, and AFAIK the same BIOS warnings pop up about voiding warranty on standard consumer boards.

 

[tabascosauz]

I have no idea what size they are; I know there are some SMD components on the package (hence the odd shape of the Ryzen 7000 IHS) but I do not know if they are resistors, capacitors, fuses, magic smoke containers, or nuclear reactors. If they are part of the die, then which die is it? The CCD is universal - used in Ryzen as well, and AFAIK the same BIOS warnings pop up about voiding warranty on standard consumer boards.

You said "replaceability". I wouldn't count on any of it being accessible to the end user.

AM4/AM5 do not get the same kind of message. If you read the actual source article, Threadripper also gets the standard PBO message from AGESA that you are seeing, but this short one is a new addition and doesn't appear to be coming from the AGESA. If it's just inserted by the board vendor into their BIOS, then this whole bit of news is just a nothingburger, which is why the AMD rep responded in the way they did.

As to the standard PBO message: if it is simply a message and doesn't make hardware changes, then there's no issue. If it does make hardware changes, it still has always been ignored by AMD RMA and has never been an issue.

 

[FoulOnWhite]

I reckon the actual fuses are on one of the dies and so tiny you'd need a microscope to see them.

 

[xorbe]

Probably nothing to do with RMAs. It's probably for the vendor to know whether to just toss the CPU in the trashcan, or have a look at why a non-oc'd CPU died.

 

[bug]

I reckon the actual fuses are on one of the dies and so tiny you'd need a microscope to see them.

Well, yeah... Where would you put a user-replaceable fuse on a CPU?

Probably nothing to do with RMAs. It's probably for the vendor to know whether to just toss the CPU in the trashcan, or have a look at why a non-oc'd CPU died.

Not meant for RMA, that's for sure. Hopefully it won't be abused, but human beings are know for doing that every chance they get.

 

[Aquinus]

How the hell does AMD know if parameters were out of whack during overclocking? Wouldn't you need more than just a fuse to store that kind of information, like extended overheats or overheat shutdown situations? This is AMD saying, "overclocking alone doesn't void warranty, but if it's the cause for breaking something, it does." Well, how does AMD confirm that overclocking was the culprit if that's the case? Sounds like a PR stunt to save face without rolling back the verbiage. I can't say that I like this tomfoolery.

 

[Count von Schwalbe]

You said "replaceability". I wouldn't count on any of it being accessible to the end user.

Well, I was more referring to well-equipped repair shops, which could replace an SMD component. Or a particularly skilled end user. But if it is part of the die, that would be entirely out of the question.

AM4/AM5 do not get the same kind of message. If you read the actual source article, Threadripper also gets the standard PBO message from AGESA that you are seeing, but this short one is a new addition and doesn't appear to be coming from the AGESA. If it's just inserted by the board vendor into their BIOS, then this whole bit of news is just a nothingburger, which is why the AMD rep responded in the way they did.

Oh, I see. I never really paid attention to those messages, I just assumed it was the same one.

 

[bug]

How the hell does AMD know if parameters were out of whack during overclocking? Wouldn't you need more than just a fuse to store that kind of information, like extended overheats or overheat shutdown situations? This is AMD saying, "overclocking alone doesn't void warranty, but if it's the cause for breaking something, it does." Well, how does AMD confirm that overclocking was the culprit if that's the case? Sounds like a PR stunt to save face without rolling back the verbiage.

They aren't even trying to know whether parameters were out of whack. They just need to know whether overclocking was enabled or not. Probably mostly for statistics, initially.

 

[Aquinus]

They aren't even trying to know whether parameters were out of whack. They just need to know whether overclocking was enabled or not. Probably mostly for statistics, initially.

Ehhhh, I think you're being too kind. I seriously doubt that they did this just to gather statistics. At best, it tells them to not even bother doing testing on an RMA'ed CPU where that bit has flipped. At worst it could deny you an RMA. If AMD was serious about it not impacting RMA, they'd remove all the verbiage regarding this, but they didn't. This feels like smoke and mirrors.

 

[LabRat 891]

No it's necessary. Overclocking these chips is ridiculously hard and expensive and is kind of dumb given these are usually money producing machines. [snip]

Correct.
I'm familiar with at least one (semi-local to me) that specializes in (what I'd call) Halo-Tier +1 HEDTs and High-Performance Workstations:


IDK exactly who their clients are, but I imagine
Cinema-CGI and "Sci./Sim." clientele.

You said "replaceability". I wouldn't count on any of it being accessible to the end user.

AFAIK, there's 0 chance these are 'replace-/repair-able'
These are in-silicon 'fuses'

IIRC, This has come up w/ OEM-locking AM4 Ryzens, prior.

AM4/AM5 do not get the same kind of message.
If you read the actual source article, Threadripper also gets the standard PBO message from AGESA that you are seeing, but this short one is a new addition and doesn't appear to be coming from the AGESA.

As to the standard PBO message: if it is simply a message and doesn't make hardware changes, then there's no issue.

It's become concerningly-common to
simultaneously advertise OverClocking
while the fine print says that using the advertised feature voids warranty.
However-

If it does make hardware changes, it still has always been ignored by AMD RMA and has never been an issue.

-^this^ has generally been my experience(s) with every company I've ever RMA'd with; most-recently, AMD for my early R5 5600 that lost cores (PBO'd).

So, to be fair (at least, for the time being)
most companies seem to merely reserve the right to deny an RMA
based on OCing.

IMO,
this knowledge (about die-level fuses) is about as-impactful as the knowledge of Pentium IIIs being 'serialized' (A once-controversial action from Intel; a long time ago)
It may have implications but, for most-of-us
(enthusiasts, niche-markets, etc. inclu.)
it's (mostly) inconsequential.

If it's just inserted by the board vendor into their BIOS, then this whole bit of news is just a nothingburger, which is why the AMD rep responded in the way they did.

Just... 'stay vigilant'
in this day-and-age of anti-consumer, anti-ownership, anti-RTR, everything

 

[trsttte]

Maybe so, but how can the user prove damage was or wasn't caused by overclocking/overvolting? You send in a defective CPU, AMD sees the blown fuse and claims it broke because of overclocking/overvolting. There's no way for the end user to argue.

And to be crystal clear, I'm not saying AMD plans to abuse that. I'm just saying that fuse opens the way for at least some distributors to go that route. At the same time, it's probably a useful tool in diagnosing, because users sending in defective parts are very unlikely to admit they overclocked before the damage happened.

I think this statement is meaningless and kind of a joke in practice, it's not even a matter of abusing it or not, there's not much to do with a broken cpu and they won't put every rma claim under a microscope to debug and find out for sure what happened. So if you overclocked/overvolted the thing you're shit out of luck, simplest cause to explain the problem until there's a large enough sample of defects to imply there's a wider problem with product that envolves a recall or class action or something.

Basically, don't overclock threadrippers cpus if you want to keep your warranty lol

Further to this, where is the fuse located? CCD? IOD? SMD on the package? That could make a difference, for replaceability and whether it is in other CPUs that use the same CCD.

I doubt very much replacing the fuse is in the realm of a normal person, unless you have access to very special equipment, and even then, is it actually replaceable @W1zzard

It's simply not possible, they're in the silicon itself, it's not a fuse like you see on a random appliance, it's like a transistor engraved in the silicon but instead of fullfilling some architecture logic it's designed to work like a fuse.