Monday, January 2nd 2023
AMD Radeon RX 7900 XTX May Feature Faulty Coolers, Causing Overheating
AMD's latest GPUs have been reported to be experiencing overheating issues, with many users claiming that the vapor chamber cooler works better in a vertical rather than a horizontal position. Regardless of orientation, vapor chamber coolers should equal roughly the same heat dissipation performance and move the heat away from the source; however, testing showed that some reference AMD Radeon RX 7900 XTX GPUs feature defect coolers. According to the testing conducted by Roman "der8auer" Hartung, AMD's Radeon RX 7900 XTX RDNA3 GPUs are experiencing problems with overheating caused by a faulty vapor chamber design.
What der8auer found is that these coolers could have a defect in the manufacturing process, where the liquid inside the vapor chamber faces problems in circulation after condensation. It could relate to manufacturing issues of the cooler itself, with an inadequate amount of fluid or insufficient pressure inside the chamber. For more in-depth testing and performance benchmarks, see the video below. It is important to note that we didn't see other reports that replicate this behavior, so always take these reports with a dash of salt.
What der8auer found is that these coolers could have a defect in the manufacturing process, where the liquid inside the vapor chamber faces problems in circulation after condensation. It could relate to manufacturing issues of the cooler itself, with an inadequate amount of fluid or insufficient pressure inside the chamber. For more in-depth testing and performance benchmarks, see the video below. It is important to note that we didn't see other reports that replicate this behavior, so always take these reports with a dash of salt.
https://www.techpowerup.com/
286 Comments on AMD Radeon RX 7900 XTX May Feature Faulty Coolers, Causing Overheating
Wtaf have female dogs got Todo with anything and how.
It's the life, this is a waste.
At least shout poor Volta POOR Rdna3 with a bit of bold and vinegar, maybe red text.
Not that any of them are innocent, cause they are not. Far from it. In the end, it is companies that Try to earn as much money as possible.
The largest online retailer here in Australia has 7900XTX cards sold out… 4080s are much more plentiful, and another A$500 more.
How do folks have A$2,500 laying around to spend on a graphics card? At that point there’s little reason you couldn’t ditch PC gaming and use console(s).
videocardz.com/newz/amd-exec-makes-fun-of-nvidia-rtx-4090-power-cables-melting-issue
Not denying the 4080 hasn't sold like hotcakes, but I wouldn't hang my hat on that evidence alone.
Did they messed up with market share vs shipment numbers ?
I can hardly imagine Intel sold 3.6% worth number of the whole market share of cards in ONE quarter
3.6% of the whole discrete GPU market share is like Millions of units.
Sticking an Arc A370 into an notebook and let it rot inside a warehouse doesn't count as market share, but it surely count as 'one unit shipped' .
I'm calling "Derb8ur"'s or whatever his name is "analysis" typical youtube clickbait until Buildzoid says otherwise. I'm not going to hold my breath though, especially given that the results are unable to be replicated.
By the way, my 2070 has just died for no reason while sitting on a shelf, so Nvidia should recall the whole Turing lineup and offer compensation. "Obviously."
Nvidia vs AMD trolling right out of the gate? Classy stuff.
I 've picked one article to fact check and this is the very first paragraph :Clearly they messed up between shipment and market share
Their report is based on shipment numbers, but kept saying market share in their paragraph and charts.
In their article on 30, Nov
This is how they presented their charts
Which, the chart itself stated 'Market Share' in the picture
But the description below it says 'Shipments and market share percentages'
I think this is highly deceiving.
Since people usually quote the charts only, and failed to realize it isn't the whole picture.
In their 3 Dec article,
Despite the '31,776 -> 31.776' mistake which, should never happen in a 'Professional research group'.
First, assume the 31.776 billion figure is correct for the whole AIB market.
They marked Intel increase from 1% to 4% in the 'AIB share'
Which translate to 31.776*1,000,000,000*0.03 = $ 953280000
That's a lot of dollars
The most expensive AIB graphics card Intel could offer is the Arc A770 = $349 each
$ 953280000 / $349 = 2,731,461 units
Lets say the market price is inflated ~$500 per card
$ 953280000 / $500 = 1,906,560 units
Are they suggesting the AIBs sold 2 million Arc A770 unit (or equivalent $$ of cards ) in 3 months ?
That is simply unrealistic and deceiving information.
Or maybe they oversimplified their data and not disclosing some key elements.
He had a video on that subject before the one we are talking about, and in this first video he also didn't find any problem. That`s why he got so many people reaching to him with possiable problem.
So no one need to be surprised that reviewers didn't found it themselves. They might have got a different set of GPU`s that are OK, maybe 'cherry picked' if you want to go into conspiracy's that AMD know about the problem and decided to ship anyway (must get it to holiday, don`t they?).
No reviewer was on the bad 12vhpwr adaptor in the beginning as well. In fact, after it was known many big shots try replicate the fire but didn`t succeed and the one who did invest a lot of time and effort to get there. It just show that reviewer not catching it say nothing.
There is a problem, do denying that. See the flip test.
How widespread it is is the question.
To identify the problem you must have a bad card so you want more than one to be sure.
That helps you to isolate what is the cause- not to determent how widespread it is so don`t mix it.
And by the conclusion- a bad vapor chamber design- He conclude that the issue can be a big problem out of his experience and knowledge about vapor chamber mechanics and manufacturing process.
You can argue against his conclusion - vapor chamber design problem - and suggest other cause for the irrational card behavior.
You can argue that a vapor chamber design problem is not that much of a problem and no reason for it to be a widespread issue.
You can argue that his experience and knowledge about vapor chamber is lacking.
You can argue against his test methods.
But going against the whole thing just because he knowingly buying and testing a bad GPU from users in the first place is is just wrong.
And it's impossible to get overall market share from the data of shipped cards - what's the "best before" date of graphics cards? Some people retire their cards in one generation, and don't even sell them on, some people buy every two generations, and sell their old card, some people use very old cards - either bought, or are upgrading only once every few generations...
The only more or less useful market share numbers I know of is Steam Hardware Survey, and people have problems with that chart since the introduction! I wasn't polled, so it's inaccurate! It's clearly biased and it's numbers are fake! And so on... So people more or less stopped using it as a data source to create their charts on forums (due to attacks), and Valve doesn't provide very helpful visualizations - they bundle together everything - integrated graphics, laptop cards and discrete desktop cards...
And you are more overreaching with your accusation of him then him to call for mass recall- I didn`t see any demand from his side. Please show. I think you are falling into your own accusation.
See what his conclusion was and react to that, not to the mambo jumbo around it.
I will help:
"I`m not sure what will be the right move will be"
"when there is a possible design failure"
"and I guess they will have to recall"
"not all cards affected obviously"
"It seems to be a high amount of (affected) cards"
"I have a felling they will have to recall the entire MBA"
I`m sure we can all agree that if this is a widespread issue than a recall is a must and waiting for people to RMA is not the right way to go from AMD side if they are fair.
If a small % of the cards are affected than RMA will be OK I think and he did a good service to all consumer bringing it to attention.
And if it will happen to be just a very small % of affected cards than his credibility will suffer, despite him not conclusively determinate anything regarding recall.
I think the conclusion is well balanced according to the findings.
I guess people have problem with the somewhat bold headline of "AMD is in BIG trouble" but if he really wanted to entice/clickbait it he could have written something like "7900XTX cooler is a design flow - total recall of is a must". Others would have done so with those findings.
Regardless of his "feelings" and "guesses", the solution is clear as day: AMD has to do a proper investigation on the matter, and then issue a statement on what happens next. There's a proper way of handling a mass fault / recall. There always has been. One person's "feelings" don't change that, especially someone's who has zero clue on how widespread the problem is.
That's why sometimes it is worth spending 2 more minutes to dig a little bit more since most of them liars are quite bad in lying.
There have been recent examples of companies shipping products with high RMA rate - even with possibility of fire, and companies only reacted when journalists begun to add together "totally anecdotal stories".
Edit: I quickly skimmed through the video (20 minutes is too long to watch so early in the morning). He based his entire findings on the delta between hotspot and junction temp. He indicated that at some point, the delta drops, and the cards start to throttle. Earlier in the video, he shows that even though the cards reach max hotspot quite early, the junction temp starts to rise later, and that leads to the delta dropping... So... if the junction temp has room to rise, then where is the throttle? What are the clock speeds and power consumption? Has he run any stress test that give you an actual indication of the cards throttling? We don't know.
Edit: He claims a voltage and power consumption decrease. Does this manifest in decrease in performance too? Again, we don't know.