Support NeoGAF

[Leonidas]

AMD Admits Faulty Vapor Chamber Causes RX 7900 XTX Throttling

A limited number of reference Radeon RX 7900 XTX GPUs are affected.

www.tomshardware.com

"We are working to determine the root cause of the unexpected throttling experienced by some while using the AMD Radeon RX 7900 XTX graphics cards made by AMD. Based on our observations to-date, we believe the issue relates to the thermal solution used in the AMD reference design and appears to be present in a limited number of the cards sold. We are committed to solving this issue for impacted cards. Customers experiencing this unexpected throttling should contact AMD Support."

Good to see them say something, it's too bad this happened since the reference "made by AMD" RX 7900 XTX were the cheapest 7900 XTX's on the market.

 

[CLW]

This combined with the shit nvidia is trying to pull is making me lean towards sticking with my trusty ps5

 

[jaysius]

Man QC has gone into the fucking dumpster in 2022-2023 now eh?

 

[bentanchorbolt]

Man QC has gone into the fucking dumpster in 2022-2023 now eh?

Product recalls are nothing new, nor are issues like these.

 

[Skifi28]

AMD, you had one job. Leather jacket man already thinks of doubling prices with no competition.

 

[daffyduck]

AMD Admits Faulty Vapor Chamber Causes RX 7900 XTX Throttling

A limited number of reference Radeon RX 7900 XTX GPUs are affected.

www.tomshardware.com

Good to see them say something, it's too bad this happened since the reference "made by AMD" RX 7900 XTX were the cheapest 7900 XTX's on the market.

But are they readily available? If it's like the 6xxx series, then it doesn't make much difference.

 

[sandbood]

AMD, you had one job. Leather jacket man already thinks of doubling prices with no competition.

 

[GreatnessRD]

Yeah, its good that they addressed the issue and are issuing RMA's and refunds.

 

[RoboFu]

Yeah the xtx is a great card. So is the xt . If they would cut $200 off each they could be big competition against nvidia this go round.

 

[mitchman]

The AMD issue is only a problem if you vertical mount the GPU, and most won't do that. It should be fine in 90% of the cases.

 

[01011001]

The AMD issue is only a problem if you vertical mount the GPU, and most won't do that. It should be fine in 90% of the cases.

you are 100% wrong

1: the issue is happening mostly on horizontal mounts
2: 99% of people mount horizontally

 

[FingerBang]

Yeah the xtx is a great card. So is the xt . If they would cut $200 off each they could be big competition against nvidia this go round.

There is no bad graphics card, only bad prices. The XTX and XT are great cards, but not at that price. They would have been loved by everyone if they were the 7800xt and the 7900xt, for 749 and 899, but no, they played the same game Nvidia is playing and they don't offer a much greater value than the competition.

That said, the RT performance seems to be a non issue in some titles and moving forwards Nvidia might lose that advantage.

 

[Drew1440]

This combined with the shit nvidia is trying to pull is making me lean towards sticking with my trusty ps5

Intel Arc will save us

 

[mitchman]

you are 100% wrong

1: the issue is happening mostly on horizontal mounts
2: 99% of people mount horizontally

Yes, I was 100% wrong, I rewatched the der8auer video now and it's horizontal mounting that is indeed the problem.

 

[sendit]

 

[Zathalus]

With the limited power available (2x 8pin) and now this, AIB is the way. AMD reference has always sucked.

 

[Leonidas]

 

[GHG]

Yeh their statement is bogus. They have no way of knowing how many cards are affected until they start receiving cards from customers.

Edit: its also funny how little traction this thread has got here. Apparently high pricing is worse than faulty products?

 

[RoboFu]

Yeh their statement is bogus. They have no way of knowing how many cards are affected until they start receiving cards from customers.

Edit: its also funny how little traction this thread has got here. Apparently high pricing is worse than faulty products?

Yeah they do. If everything points to a certain batch then they know how many are in that batch.

 

[DaGwaphics]

Yeh their statement is bogus. They have no way of knowing how many cards are affected until they start receiving cards from customers.

Edit: its also funny how little traction this thread has got here. Apparently high pricing is worse than faulty products?

AMD barely produces any reference product and typically stops production very early on. If there was an issue with the chip itself that would get a lot of traction. Problems with a specific card that will likely represent less than a few percentage points of the total units sold isn't going to make a lot of headlines. Certainly an unfortunate situation though, seems like something that should have been noticed during testing.

As it is, just grab a Gigabyte Gaming or a XFX Merc310 at the same price and get a better cooling solution.

 

[hlm666]

Yeah they do. If everything points to a certain batch then they know how many are in that batch.

The video GHG was commenting on was saying if AMD new what batch it was they have serial numbers and could do a proper recall on the effected cards and not leave it to the users to find they have a defective card and them chase AMD support.

On a seperate note, In the pcworld video he says we have the fix and will ship it to you. Are they planning on just shipping the vapor chamber and letting the user do the work?

 

[GHG]

if AMD new what batch it was they have serial numbers and could do a proper recall on the effected cards and not leave it to the users to find they have a defective card and them chase AMD support.

Yep that would be the proper way to do it but the fact they are not doing it this way makes me think they still have no idea and are only putting out the statement they have to save face/damage control.

 

[winjer]

AMD RX Radeon 7900 XTX (reference) Failure rate would be 11% opposed to 1%

We cannot vouch for the numbers ourselves, but there are issues with the latest reference AMD graphics cards (RX 7900 XTX), mostly related towards failing vapor chamber cooling. ...

www.guru3d.com

 

[Unknown Soldier]

Nobody fucks with the leather jacket and lives

 

[M1chl]

Yeh their statement is bogus. They have no way of knowing how many cards are affected until they start receiving cards from customers.

Edit: its also funny how little traction this thread has got here. Apparently high pricing is worse than faulty products?

Went here to say that "no bad GPU, just bad prices". I guess buying cheaper GPU which breaks/does not work as intended is good.
Not sure how everyone else, but I am always seething when something breaks, I pay for it, so it has to be serve me reliably.
Even if cost is higher it is worth it

 

[Kenpachii]

The face of the guy, always makes me wonder if he's really serious or just laughing deep inside at us clowns.

 

[Ozriel]

Man’s holding it up like a communion wafer

 

[nemiroff]

This combined with the shit nvidia is trying to pull is making me lean towards sticking with my trusty ps5

This is not an attack on you nor Sony, and your opinion is 100% valid and fine (of course). I just wanted to insert a point about hardware and design flaws (and factory process issues etc). If you read the stats on hardware ALL kinds of hardware breaks (hardware failures are sometimes "scary" high in general), and design flaws can sometimes slip through and hit any type of "trusted brand". Another "just an example", I'm the kind of guy that buys every console, and in MY personal experience there's no hardware I've owned that has broken on me so much as my Playstations in total (3 PS OGs, 3 PS2s 1 PS3 (and of course 2x Xbox 360..) broken) and Sony gear in general (internal batteries especially, cameras, phones, multiple Sony headsets) - Except my Apple Macbooks and other hardware (it broke me as Apple fanboy, lol). And for illustration, not evidence, I've honestly never had any hardware break on me when it comes to my decades of PC building, even mechanical hard drives. These days I do not fear Sony's hardware, including Apple, because things change, it's all kinda dynamic and often by chance. When it comes to Nvidia's, cold calculated shitty business tactics however, it's a different beast indeed.. So that's why I really rooted for AMD this round..

 

[RoboFu]

Went here to say that "no bad GPU, just bad prices". I guess buying cheaper GPU which breaks/does not work as intended is good.
Not sure how everyone else, but I am always seething when something breaks, I pay for it, so it has to be serve me reliably.
Even if cost is higher it is worth it

Faulty products are always bad, but in this case it’s only a small amount which are being replaced. Lower price will matter more in the long run.

But anway I already see prices dropping. The 7900 xt I bought is already $50 cheaper on bestbuy.

 

[Buggy Loop]

http://[URL]https://www.guru3d.com/...tx-failure-rate-would-be-11-opposed-to-1.html

[/URL]

So I guess the board partners that used the same reference cooler? Haven’t heard anything about non reference boards having the thermal problem. So 10% of those reference coolers they mean? Surely not 10% of all 7900xtx cards. Clear as mud that article..

 

[winjer]

So I guess the board partners that used the same reference cooler? Haven’t heard anything about non reference boards having the thermal problem. So 10% of those reference coolers they mean? Surely not 10% of all 7900xtx cards. Clear as mud that article..

No, the article is very clear. It's 11% of reference cards.

 

[Buggy Loop]

No, the article is very clear. It's 11% of reference cards.

So 11% just on AMD side based on that Scott employee comment, with some sprinkle of AIB but no numbers on that, got it.

But 11% of RMAs, that’s insane, because for sure there’s peoples who don’t even look at junction temps or even aware of this.

That’s why DerBauer said that a full recall or at least a notice for a series number should be the right move. Surely they know which cards are potentially affected.

 

[k_trout]

I feel bad for those with issues, that sucks, way to take the shine of your new thing...bummer

 

[Buggy Loop]

I feel bad for those with issues, that sucks, way to take the shine of your new thing...bummer

Shit happened in manufacturing. Sucks but nobody is immune to that.

The only thing of interest is the response to the problem.

If they can’t offer a replacement though, I would say that sucks especially with how niche their market share is, you refund a customer that spend $1k on a new architecture.. who’s to say he won’t just buy an 4080 now? If reference cards are out of stock and AIBs are within spitting distance to 4080 prices.. AMD should have a ticket system to promise a replacement in a timely manner, if they want to assure the customer is staying.

 

[winjer]

So 11% just on AMD side based on that Scott employee comment, with some sprinkle of AIB but no numbers on that, got it.

But 11% of RMAs, that’s insane, because for sure there’s peoples who don’t even look at junction temps or even aware of this.

That’s why DerBauer said that a full recall or at least a notice for a series number should be the right move. Surely they know which cards are potentially affected.

Source fo Guru3d says that Cooler Master made the cooler, but the vapor chamber was done by a company named "Asia Vital Components" (AVC). Apparently, it was AVC's quality control that failed.
Something tells me AVC is going to have to pay for this mess.

 

[diffusionx]

How many people are even going to know that their card is throttling? They say “small number” but how sure are they? It’s quite possible that people will just think the performance is what it is. Without a foolproof way to check, AMD is just pissing on their customers right now.

 

[KyoZz]

A few 4090 had problem because some users don't know how to plug a cable: internet meltdown.
AMD having problems not only with their vapor chamber but also on the latest drivers for the 6000 series causing cards dying? It's ok, it happens.

Anyway, it sucks because I wish Nvidia had real competitors, it's always for the best but AMD just keep failing.

 

[diffusionx]

A few 4090 had problem because some users don't know how to plug a cable: internet meltdown.
AMD having problems not only with their vapor chamber but also on the latest drivers for the 6000 series causing cards dying? It's ok, it happens.

Anyway, it sucks because I wish Nvidia had real competitors, it's always for the best but AMD just keep failing.

Actually no, the 4090 thing was because of the risk of fire, and they said they were going to redesign the cable. It was a flawed design that led to a very real problem.

Throttling because of shit vapor chamber is not the same thing.

 

[KyoZz]

Actually no, the 4090 thing was because of the risk of fire, and they said they were going to redesign the cable. It was a flawed design that led to a very real problem.

Throttling because of shit vapor chamber is not the same thing.

I'm not comparing the problems, but the reactions. 11% of the cards could be affected. That is MASSIVE.

For the 4090 sure the design of the cable could've been better I 100% agree. Still, because some users don't know how to plug it properly we got tons of clickbait videos and trolls jumping into shitty talks. Seriously that video by LinusTech (I think? I may be wrong and it could be another Youtuber) on how to plug a cable is hilarious:

"So here you uh, uuuuh, take the cable and plug it in and uuuh... by exerting a little force you should be fine"

Thank you for that, next time I may need a tutorial on how to drink water!

And I know what I'm talking about, I got a 4090. The cable is nothing too hard honestly.
Not to mention that anyone who don't make sure to properly clips a cable in a 2k card is just calling for problems.

 

[Buggy Loop]

Actually no, the 4090 thing was because of the risk of fire, and they said they were going to redesign the cable. It was a flawed design that led to a very real problem.

Throttling because of shit vapor chamber is not the same thing.

The revision of the cable is actually to make the 4pin sensor higher so that if the cable is not all the way in, it won’t power at all. That’s actually a legit improvement for future cables and something good out of this mess. It’s actually more idiot proof than all legacy cables with this revision. Gamers Nexus’ investigation pretty much shut down any connector flaw theories. A standard PCI-Sig

Original is not any more flawed than any cables.You can find 8-pin PCIe, Molex, USB, 4-pin CPU, 24-pins motherboard, etc melted online. There’s a ton of newcomers in PC space that don’t have experience. There’s a reason we have « warning hot coffee » on cups, you can’t idiot proof everything, especially anything with electricity.

Proof that it’s not a cable flaw? Is that you would still see all the YouTube tech clickbaiters talk about the issue if it wasn’t the case.



When you have peoples like Jonny Guru, one of the best known reference for PSUs and involved in many cable specs telling you it’s not a cable problem, you listen closely. And this was pre meltings from user error.

 

[kiphalfton]

Man QC has gone into the fucking dumpster in 2022-2023 now eh?

Gotta cut costs somehow.

Especially at the prices they're selling them. They're practically giving them away.

 

[Insane Metal]

AMD Radeon RX 7900 XTX Failure Rates Reportedly At 11%, RMA's Piling Up But Users Not Receiving Cards

AMD still feels thinks that the number of affected Radeon RX 7900 XTX is "small," but thousands of consumers are feeling the brunt of it.

wccftech.com

Damn. Get your shit together AMD.