News AMD Vows to Replace Overheating Radeon RX 7900 XTX Boards

[Admin]

There are no problems with Radeon RX 7900 XTX from AMD's partners, but some AMD reference boards have an issue with their cooling system.

AMD Vows to Replace Overheating Radeon RX 7900 XTX Boards : Read more

 

[-Fran-]

I find good that AMD (via mr Scott) ate* a good amount of humble pie and acknowledged what rubbed people wrong after the announcements along the issues with the cards now. So, good on them for that.

Now, it's a bad thing they had to be asked point blank and not issue an official statement before this. That is something they should improve as well.

Anyway, at least the issue is now fully accepted and RMA process is in place.

I wish everyone with an affected card good luck with their RMA.

Regards.

 

[rDigital]

AMD waited as long as they could before addressing this one. Does it really have to get addressed by tier 1 YouTubers before they take action?
It feels like bad luck from making fun of Nvidia.

 

[digitalgriffin]

I would wonder how they would classify "a small percentage of users?". Is that among all 7900XTX boards? Well then yes, only a small percentage of boards made use the vapor chamber because board makers want an excuse for a bigger margin (custom designs). But by the hundreds of reports, that sounds like a high percentage for vapor chambers.

In some ways, other than labor and shipping, the cost of the vapor chamber replacement will be bore by the vapor chamber mfg. (Like tikata had to pay for airbag r placement). It won't cost AMD that much. But the damage to their rep is far more costly. Thet better darn well recall all boards made with that vapor chamber mfg.

 

[edzieba]

If it's a bad batch, then have they issued a serial number range of the affected batch(es) in order for customers to proactively tackle the issue?

 

[digitalgriffin]

If it's a bad batch, then have they issued a serial number range of the affected batch(es) in order for customers to proactively tackle the issue?

Not like car mfg. When they put the car on the assembly line they scan the tag on the part and then scan the windshield code. That way they know the serial number for every assembly on that car and that it is the proper part.

Somehow I'm doubting each vapor chamber has a serial number.

 

[OneMoreUser]

I find good that AMD (via mr Scott) ate* a good amount of humble pie and acknowledged what rubbed people wrong after the announcements along the issues with the cards now. So, good on them for that.

Now, it's a bad thing they had to be asked point blank and not issue an official statement before this. That is something they should improve as well.

Anyway, at least the issue is now fully accepted and RMA process is in place.

I wish everyone with an affected card good luck with their RMA.

Regards.

How did you come to the "asked point blank" part?

I find it funny how some people have made a mountain out of almost nothing, especially a few youtubers that have been crying bloody murder which of course had zero to do with them benefiting financially by doing so.

However on the same time it is rather sad to see that is how the internet work, those that shout the highest and are the most offended gets the attention - facts are simply there to be ignored or at best used as a basis to exaggerate to the point almost anything is a major disaster.

 

[OneMoreUser]

I would wonder how they would classify "a small percentage of users?". Is that among all 7900XTX boards? Well then yes, only a small percentage of boards made use the vapor chamber because board makers want an excuse for a bigger margin (custom designs). But by the hundreds of reports, that sounds like a high percentage for vapor chambers.

In some ways, other than labor and shipping, the cost of the vapor chamber replacement will be bore by the vapor chamber mfg. (Like tikata had to pay for airbag r placement). It won't cost AMD that much. But the damage to their rep is far more costly. Thet better darn well recall all boards made with that vapor chamber mfg.

I bet a lot of those "hundred of reports" are less than genuine, not to mention that lots are reports that have been counted multiple times.

And it would be wasteful if AMD were to recall made with the cooling solution by mfg X, not to mention it would be really annoying for all those with no-issue boards having to deal with it. Suggesting a total recall isn't close to rational.

 

[Coffee Fueled Curmudgeon]

Not as if they had anywhere else to go, but the lottery of not knowing which of thousands upon thousands of cards already in the distribution chain is affected and will need the gross inconvenience of an RMA leaves a very sour taste.

A recall would undo some of the irrevocable breach of trust here, the idea that an RMA is a great experience is laughable, literally no one has ever been happy with having to do an RMA, so it seems bizarre that some are now treating the nuisance and inconvenience as manna from heaven. How fickle the internet fanboy is.

Of course, some, with agendas are complaining that it's a mountain from a molehill created by greedy youtubers, but it's not as if they minded when the roles were reversed a month or so back and Youtubers were blaming Nvidia for operator error. Tubers gonna tube, there's deluded fanboys on both sides all too happy to spam comments sections with faux reports of the same issue and long-winded narrative peddling.

 

[PEnns]

At the very least, AMD is not blaming its customers & users, like you know who....

 

[Eximo]

Not like car mfg. When they put the car on the assembly line they scan the tag on the part and then scan the windshield code. That way they know the serial number for every assembly on that car and that it is the proper part.

Somehow I'm doubting each vapor chamber has a serial number.

But they do have serial numbers on the cards and manufacture dates. Wouldn't take much to track down what days the bad cards were made on and just assume any card between those dates is potentially eligible for replacement.

I would not be surprised if they did have a serial number sticker on each cooler. This would be done for the express purpose of tracking down incidents like this.

 

[atomicWAR]

AMD waited as long as they could before addressing this one. Does it really have to get addressed by tier 1 YouTubers before they take action?
It feels like bad luck from making fun of Nvidia.

Same exact thing happened with nvidia on the 4090s.

I would wonder how they would classify "a small percentage of users?". Is that among all 7900XTX boards? Well then yes, only a small percentage of boards made use the vapor chamber because board makers want an excuse for a bigger margin (custom designs). But by the hundreds of reports, that sounds like a high percentage for vapor chambers.

In some ways, other than labor and shipping, the cost of the vapor chamber replacement will be bore by the vapor chamber mfg. (Like tikata had to pay for airbag r placement). It won't cost AMD that much. But the damage to their rep is far more costly. Thet better darn well recall all boards made with that vapor chamber mfg.

I heard thousands according to UFD tech but their news is a little click baity at times. Guess we will see but from the numbers I've seen online it already looks worse than Nvidia stuffed up launch of the 4090.

 

[edzieba]

Not like car mfg. When they put the car on the assembly line they scan the tag on the part and then scan the windshield code. That way they know the serial number for every assembly on that car and that it is the proper part.

Somehow I'm doubting each vapor chamber has a serial number.

Intake of the bad batch of chambers would be at a known date, time, and quantity. Assembly of (serialised) cards would be logged. With an estimated buff of time-of-stock-intake to time-of-first-assembly (low for modern JIT manufacture) you can get very tight to the range of SNs where those vapour chambers ended up.
Plus, there's a better than even change the software monitoring production already has records of which vapour chambers from which source were installed in which SN cards, even if the vapour chambers are not individually serialised, because the packaging the chambers are carried in for stocktaking and handling (e.g. tray) will be serialised. Parts are tracked through production as a matter of course for catching exactly this sort of QC issue, even on far lower value products and subassemblies (e.g. even on some $10 widget the manufacturer can probably track down a blown cap to the serialised tape of parts that was installed in the pick & place at time of assembly).

Same exact thing happened with nvidia on the 4090s.

Except there's nothing to recall on the 4090 power issue, because the issue is failure to fully insert the connector rather than any actual hardware issue. As far as I am aware, nobody has been refused replacement of an affected card in the case of self-installs, and replacement of cards or other component sin pre-builts would be down to the system integrator. For pre-builts, it should be 100% on the integrator to make good, as the assembly failure was theirs. For self-installs, replacement would fall under goodwill and would likely be continued as unofficial policy (now the reason for the issue is known and a worldwide wave of people making sure the connector is fully seated has occurred, failure rates should be low enough going forward for the cost of a few cards to be worth the reputation value).

 

[digitalgriffin]

I bet a lot of those "hundred of reports" are less than genuine, not to mention that lots are reports that have been counted multiple times.

And it would be wasteful if AMD were to recall made with the cooling solution by mfg X, not to mention it would be really annoying for all those with no-issue boards having to deal with it. Suggesting a total recall isn't close to rational.

It would behoove them not to do a full recall. Admittingly it might be a handful, or it might be every last stinking one. We just don't know. That also means any cards sitting in inventory could be potentially defective. That is where-in the problem lies.

The percentage of users who just pop their card in and assume it works without looking at the temps is high, I'm sure. So mass media coverage is vital unlike what you claimed. If a vehicle has a recall, the dealers are notified for every car sitting on the lot and then the NTSB requires the car manufacturer to send multiple letters informing the end user of the needed repair.

A non-recall where a future consumer doesn't have a clue about the issue, tarnishes the reputation of the company as a whole. Imagine a consumer reaction 2, 3 years down the road where their GPU fails due to constant chip baking. Said consumer goes to the forums only to discover there was a manufacturing issue, and he wasn't notified after buying the card. That is a permanently LOST customer. Be honest, wouldn't you be livid?

AMD is being foolish if they are betting on "Most people will not notice that their card died early outside warranty". Heat kills circuits early. It leads to burnt traces, electron migration, and junction pitting among other things.

And to be honest, it's certainly cheaper to fix/replace a cooler than a cooler + board. AMD should suspend all sales and grab samples from all batches (including AIBs) till this is figured out fully. A future driver update MIGHT be able to notify the end user his board might have been affected.

 

[-Fran-]

How did you come to the "asked point blank" part?

I find it funny how some people have made a mountain out of almost nothing, especially a few youtubers that have been crying bloody murder which of course had zero to do with them benefiting financially by doing so.

However on the same time it is rather sad to see that is how the internet work, those that shout the highest and are the most offended gets the attention - facts are simply there to be ignored or at best used as a basis to exaggerate to the point almost anything is a major disaster.

Have you seen an official statement from AMD, before this, acknowledging how they, ahem, over-estimated the performance of the 7900XTX in their own charts during the announcement, the driver issues (e.g: idle power and video playback power) and other random issues they've had with games along the way? (yes, there's a few bugs with nVidia's drivers as well, but good luck with an apology from them xD).

As for the cooler issue. The only thing I've seen from AMD, officially, is this: https://community.amd.com/t5/part-r...l-amd-statement-for-customers/m-p/573646#M543

That was Wed (2 days ago) and several days after der8auer put out his video (5 days ago). They've also said, on record, they've known of the issue for about 2 weeks (over 10 days). That's not a good look.

For all the hate you can throw at nVidia for whatever issues they've had (in recent history; never forget https://web.archive.org/web/2008091...dia_hid_serious_flaw_in_graphics_chips_1.html), they did good this time and just accepted RMA for the melted connectors. AMD didn't initially, no matter the reason why (yes, their RMA process is managed by an external Company, doesn't matter) and until other outlets forced their hand, only then they started the RMA officially.

In short, AMD is no saint in this, even if you can (or want to) argue nVidia is worse overall or whatever. It doesn't excuse AMD and, as I said, at least they promised to do better.

Regards.

 

[Sleepy_Hollowed]

Good, hopefully it's very limited and they caught either the process or materials.

It'd be great if the one thing that is a huge issue both in longevity and performance would have strict requirements, but it is what it is, maybe Intel can make them get their act together, as much as I hate them.

 

[thisisaname]

I find the statement

The senior vice president of AMD said that the company knew how to identify reference Radeon RX 7900 XTX graphics cards with faulty cooling systems and that all of them will be replaced.

To be at best miss leading if at the same time he did not say what that method was.
If they said what that method was people could check their cards and ask for a replacement rather than wondering if their card was affected.

 

[Alvar "Miles" Udell]

WCCFTech (English) reports, via IgorsLab (German) that AMD will not have enough replacement cards to cover RMAs for weeks, and that AMD is not covering shipping costs. I haven't seen an official recall announcement so this would seem to be the truth.

If their stance now is

"We have the fix we are ready to send it to you. Just call our tech support line if you bought it from amd.com or if you bought it from one of our AIB partners call them, they have [replacement] units," said Herkelman. "We know how to make sure and identify that they are good, and we will ship it to you right away because we want you to have a great product."

Then that sounds like they're going to ship a heatsink replacement at their cost and aren't accepting returns for defective cards and shipping a working replacement.

What's the truth?

RMA wave stagnates: Are AMD running out of cards now? The bearings of the Radeon RX 7900 XTX are yawning empty | igor ́sLAB (igorslab.de)

AMD Allegedly Has No More Radeon RX 7900 XTX Reference GPU Stock To Fulfil RMA Requests (wccftech.com)

 

[JayGau]

I don't know how many cards got impacted and we will probably never know but for sure it's not all the cards. I got a Sapphire reference model a week ago (I guess it's made by AMD since it looks exactly like the AMD model) and the temperatures are really good, even better than my old Evga RTX 2080, that is still an awesome card but was starting to show some struggling in new games.

By setting a slightly more aggressive fan speed curve than the default one in Adrenalin it basically never goes above 65C and the junction barely hits 80C. And my frame rate is pretty much always above 130 fps at 1440p and ultra settings and often saturates at 144, my monitor max frequency (especially with SAM enabled but no RT of course). And I read several other comments from people saying that their temp is good too.

It's really a great card and I feel really sad for people who got a defective one. Hopefully AMD will fix the problem for everyone and, in the meanwhile, haters will keep enjoying the feast on forums and social medias.

 

[hannibal]

But they do have serial numbers on the cards and manufacture dates. Wouldn't take much to track down what days the bad cards were made on and just assume any card between those dates is potentially eligible for replacement.

I would not be surprised if they did have a serial number sticker on each cooler. This would be done for the express purpose of tracking down incidents like this.

To a certain accuracy. They don´t know exactly what is the first and what is the last GPU that does have faulty vapor chamber. They have generic idea based on found cards when those faulty chambers were made. That is why they can not just send message to owners of the GPUs because someone could be left out and someone would be send even if the chamber is ok. Most like not even all in the certain patch of the chamber are not faulty, but that there are more faulty ones among certain chambers.

 

[Markis82]

I have intense doubts about AMD's chief's claims. There is leaked emails and info recently stating they couldn't even handle further RMAs for fixes and you could only request a refund but then 24 hours later a PR statement about how they will definitely make it right and fix all defective 7900 XTXs. We've also not really heard any information from any sources, including individual users, of their RMAs resulting in a resolved product so its really hard to accept this statement until evidence supporting this is being accomplished and on the scale of every user as promised, or at least hinting at such a thing. For now I can only see this as a PR stunt to fool people and minimize damage done to their brand plus legal consequences with AMD hoping people buy into it and nothing further pops up to dispute their good PR promise.

 

[Friesiansam]

AMD waited as long as they could before addressing this one. Does it really have to get addressed by tier 1 YouTubers before they take action?
It feels like bad luck from making fun of Nvidia.

Firstly, AMD had to check exactly what the problem was and decide how best to address it. Secondly, they didn't just blame users, like Intel did. Thirdly, most YouTubers are dickheads.

 

[drivinfast247]

I don't know how many cards got impacted and we will probably never know but for sure it's not all the cards. I got a Sapphire reference model a week ago (I guess it's made by AMD since it looks exactly like the AMD model) and the temperatures are really good, even better than my old Evga RTX 2080, that is still an awesome card but was starting to show some struggling in new games.

By setting a slightly more aggressive fan speed curve than the default one in Adrenalin it basically never goes above 65C and the junction barely hits 80C. And my frame rate is pretty much always above 130 fps at 1440p and ultra settings and often saturates at 144, my monitor max frequency (especially with SAM enabled but no RT of course). And I read several other comments from people saying that their temp is good too.

It's really a great card and I feel really sad for people who got a defective one. Hopefully AMD will fix the problem for everyone and, in the meanwhile, haters will keep enjoying the feast on forums and social medias.

I too bought a Sapphire model that is affected by this issue. And just like reported if I lay my PC on its side so the GPU is vertical the J temp drops to acceptable levels. I was planning on putting it underwater and have a EK block pre-ordered so it really wasn't that much of an issue for me. I changed my mind and decided to go SFF or close to and would like to keep the card air cooled. I reached out to Sapphire last night via email seeing if I can get a cooler shipped to me.

 

[rDigital]

problem was and decide how best to address it. Secondly, they didn't just bla

Firstly, AMD had to check exactly what the problem was and decide how best to address it. Secondly, they didn't just blame users, like Intel did. Thirdly, most YouTubers are dickheads.

Fair enough. Youtubers can be very self serving when it comes to getting the clicks.

 

[JayGau]

I too bought a Sapphire model that is affected by this issue. And just like reported if I lay my PC on its side so the GPU is vertical the J temp drops to acceptable levels. I was planning on putting it underwater and have a EK block pre-ordered so it really wasn't that much of an issue for me. I changed my mind and decided to go SFF or close to and would like to keep the card air cooled. I reached out to Sapphire last night via email seeing if I can get a cooler shipped to me.

I am very sorry for you. You got unlucky. The day before I received the card I read that the problem was more serious than we initially though and I got really worried for 24 hours. It was so a big relief when I ran the first benchmark and saw that the junction temp was good (and as I wrote, it got even better with a slight fan speed adjustment). I hope Sapphire will help you out.