If you're running a 7900XT(X)...

[EightyDuce]

Check your hotspot temperatures. There may be a potential issue with the vapor chamber on some cards which causes high hotspot Temps to the point where the card throttles.

 

Might turn into something not far off from Nvidias power adapter issue...

[AngleOff66]

Wow what a quarter for AMD.  MB prices hurting CPU sales, VR drivers and now this(if that is what it is).  If I was in charge at AMD, I would be looking at my management teams very carefully.  Fish rots from the head first. 

Re the card flip,  within these chambers and pipes does the cooling fluid in them re liquefy after shutting down and letting the card cool? If so, I'd remount my card vertically if that was an option.

As for AMD waiting for RMA, every person that doesnt because they didnt hear about it/didnt register is one less card replace and $$ saved for AMD.  Der8auer must know that.

Anyone want to bet the 4080 price may go up a bit more now?

Just my two cents.

Edited January 1, 2023 by AngleOff66

[SkateZilla]

It looks like a mounting issue, 
Vapor Chamber not working in a flat orientation vs a vertical orientation doesn't hold weight, 
But if the card is oriented vertically (bench builds), or GPU Up (upsidedown), there's no issue

The hotspot issue happens when the card is mounted cooler down, like long screws + weak retention springs + weight allow the heatsink to separate from the GPU.

There's also some images from someone's teardown, that the retention bracket around the Infinity fabric was a hair higher than the GPU surface.

I planned to replace the cooler with an aftermarket one anyway :-).

Edit: Der Bauer actually tested all that in the video as well, seems it could be a VC Mfg Defect or design defect (Pipe Design etc)

Edited January 1, 2023 by SkateZilla

[Pilotasso]

Another train wreck of a launch by AMD. The saving grace is that the reference models are out of stock everywhere still. The partner cards are the way to go and they don't have this problem.

[SkateZilla]

the issue is the Problem : Sales ratio show it's likely a bad batch of Heatsinks.,

[EightyDuce]

1 hour ago, SkateZilla said:

It looks like a mounting issue, 
Vapor Chamber not working in a flat orientation vs a vertical orientation doesn't hold weight, 
But if the card is oriented vertically (bench builds), or GPU Up (upsidedown), there's no issue

The hotspot issue happens when the card is mounted cooler down, like long screws + weak retention springs + weight allow the heatsink to separate from the GPU.

 

 

Not sure if you watched the entirety of the video, but it doesn't seem to be a straight up mounting issue. One of the tests, he tested (14:35 video time stamp) in vertical orientation, it was "ok",  then flipped to horizontal and temperature spiked. He then returned it to the vertical position (this is all with card powered up and live without remounting) and the hotspot remained. Tested with all of this cards on hand and all exhibited the same behavior.

Granted, I only have a cursory (and that's probably being generous) understanding of the cooling solution, but this guy  has been around for years and is very knowable on the subject and seems to have done more testing on this issue so far than anyone else and he seems to be fairly convinced that its the vapor chamber issue.

 

EDIT: EDIT for the EDIT  . 

 

 

Edited January 1, 2023 by EightyDuce
saw SkateZilla's edit :)

[SkateZilla]

See my edit

[EightyDuce]

Going to be interesting how this pans out. If its an initial batch issue or a defect in manufacturing process of the VC.

[SkateZilla]

I asked DB to try and mount facing up, see if it's a heat pipe or wicking material problem.

[BitMaster]

It clearly shows, despite all doubts, that QC is a thing of the past.

 

How can such things happen in masses. Der 8auer alone had 48 guys writing him... I doubt it's a bad batch, I fear a design flaw.

Edited January 6, 2023 by BitMaster

[SkateZilla]

it's funny how Social Media is blowing it up, when the issue is about the same % as the 12VHPWR Problem.

Seems AMD may have gotten a bad batch of Heatsinks.

[Thinder]

Right, another Youtube Nvidia fanboy in need of more suscribers, had to mention Nvidia connectors and ther4090 catching fire mocked by (everyone really) AMD...

Two things, he never mentions which cards (manufacturers) he covered and assume that they all use the same design = WRONG, then we can oppose tons of tests including very precise thermal data and in any case, I doubt very much that apart for the sensationalism of the title, there is much to say about it...

In short, the usual one vs the other thing still goes on with the mediocre Youtube videos to increase noise level...

Cooler = Saphire Vapor X. NOT the "standard" ref card design. 5: 04.

Edited January 3, 2023 by Thinder

[EightyDuce]

9 hours ago, Thinder said:

Right, another Youtube Nvidia fanboy in need of more suscribers, had to mention Nvidia connectors and ther4090 catching fire mocked by (everyone really) AMD...

Two things, he never mentions which cards (manufacturers) he covered and assume that they all use the same design = WRONG, then we can oppose tons of tests including very precise thermal data and in any case, I doubt very much that apart for the sensationalism of the title, there is much to say about it...

In short, the usual one vs the other thing still goes on with the mediocre Youtube videos to increase noise level...

Cooler = Saphire Vapor X. NOT the "standard" ref card design. 5: 04.

 

It's apparent that you don't know who der8auer is. He's been in the PC building, modding, overclocking,  and custom part manufacturing for over a decade and is well-respected for his contributions. If he's making "noise" it's because there's a reason and testing behind it (already more GPUs effected than nVidia connectors), unlike many YouTubers, he's not just doing it for clicks.

As for enumerating manufacturers, he clearly only showed AMD reference-style coolers and one AsRock manufacturered (uneffected). He closes by stating that any manufacturer besides AMD, that is using the AMD reference design can "potentially" be affected, not ALL cards will be effected. So either you didn't watch the video before you decided to take to the keyboard or you completely ignored everything he said and just latched on to certain aspects.

Either way, right now, the only one making "noise" just to make noise is you...

 

VC internals video for those interested. Around 09:30 mark, no apparent design flaw; not enough fluid in VC is a likely possible cause (QC issue?).

 

Edited January 3, 2023 by EightyDuce

[Thinder]

2 hours ago, EightyDuce said:

It's apparent that you don't know who der8auer is. He's been in the PC building, modding, overclocking,  and custom part manufacturing for over a decade and is well-respected for his contributions. If he's making "noise"

 

 

 

Like I give a damned who he is, he makes a living selling his  stuff on Youtube for sure, for your info I build more PCs than he  did since Pentium II, the only difference is that I don't con people with half truth and biased infos.

"We opened the 7900XTX Vapor Chamber". Right, is that NOT a reference card design? How many companies have been using it? And his title "AMD are in trouble" not noise?

You guys need to change channel and try cartoons at least they don't have the pretention of knowing what they're talking about...

So again when  I want a proper information I know where to find it and it's not with noise makers who needs contributors for their living, XMass is expensive for those guys...

Here. NO HOT SPOT.

[EightyDuce]

1 hour ago, Thinder said:

Like I give a damned who he is, he makes a living selling his  stuff on Youtube for sure, for your info I build more PCs than he  did since Pentium II, the only difference is that I don't con people with half truth and biased infos.

"We opened the 7900XTX Vapor Chamber". Right, is that NOT a reference card design? How many companies have been using it? And his title "AMD are in trouble" not noise?

You guys need to change channel and try cartoons at least they don't have the pretention of knowing what they're talking about...

So again when  I want a proper information I know where to find it and it's not with noise makers who needs contributors for their living, XMass is expensive for those guys...

Here. NO HOT SPOT.

Still waiting for more out of you than just carbon dioxide release. 

Hoping you don't take too long to  present us with your technical analysis of the issue based on published testing methodology and your vast experience of "building PC"... Which has absolutely nothing to with the issue discussed. My 15 year old has built multiple computers... Not exactly rocket science. 

[EightyDuce]

From the horses mouth, so to speak:

AMD has been aware of the issue, seems to be a batch thing, will be doing RMAs for affected GPUs. 

Hopefully a done deal. 

 

[ssg]

I ran my XTX 10 minutes on 3D Mark Speedway test. Hotspot never reached above 86c.

I still decided to sell it. Subpar VR performance and recent announcement of DLSS support made me decide to get rid of it now rather than wait for fine wine and/or FSR support from DCS. 

[Voyager]

On 1/2/2023 at 9:21 PM, SkateZilla said:

it's funny how Social Media is blowing it up, when the issue is about the same % as the 12VHPWR Problem.

Seems AMD may have gotten a bad batch of Heatsinks.

It looks like it's a larger percentage of cards than the 12VHPWR issue (as I recall debaur had 48+ people offering to send him a card, when Gamer's Nexus was only able to get 4 melted cables), but at least from what @EightyDuce posted (thank you!), it sounds like it's not a safety issue.

Overall, if AMD is able to get the word out quickly enough, this should calm down pretty quickly. Knowing that there was a specific batch of cards that had a manufacturing defect, and that an RMA process is already standing up to replace them really restores my comfort level with them. 

Now if they can sort their driver situation out in a few months, it will become a real decision whether I go for a 4090 or a 7900XTX to replace my 3080 Ti. 

Won't have time to do any upgrades until February, but definitely going to be watching. 

[SkateZilla]

19 hours ago, Voyager said:

It looks like it's a larger percentage of cards than the 12VHPWR issue (as I recall debaur had 48+ people offering to send him a card, when Gamer's Nexus was only able to get 4 melted cables), but at least from what @EightyDuce posted (thank you!), it sounds like it's not a safety issue.

Overall, if AMD is able to get the word out quickly enough, this should calm down pretty quickly. Knowing that there was a specific batch of cards that had a manufacturing defect, and that an RMA process is already standing up to replace them really restores my comfort level with them. 

Now if they can sort their driver situation out in a few months, it will become a real decision whether I go for a 4090 or a 7900XTX to replace my 3080 Ti. 

Won't have time to do any upgrades until February, but definitely going to be watching. 

The snag right now is AMD Support replies back to users stating they dont have any replacement cards yet.

[lobo]

I waited in line day one for the reference 7900xtx. By the time I got to the counter with cc in hand they were sold out. I went home empty handed and disappointed, now I'm thinking that was not a bad thing.