Hi, this is Wayne again with a topic “Is AMD STILL lying?”.
Foreign 7900 XTX problems after its release in December, the AMD 7900 XTX has had some issues. There have been a number of users reporting that it was hitting hotspot temps above 110 degrees and slowing the GPU down AMD started by saying this was normal, but soon after they admitted that some reference cards suffered, a manufacturing defect in the vapor chamber AMD was then Interviewed by Gordon at PC World and according to Der Bauer, some questionable statements were made in the interview Scott herkelman from AMD said. Is there a performance issue? What we found is, if you throttle at 110 degrees in certain workloads, you can see a small performance Delta. Oh okay, this did not line up with devourer’s testing that showed that three out of four cards he tested could only dissipate 250 to 280 watts of power, consistently 80 Watts lower than it should be, which can drop performance by 10 to 20 percent whoops. You got ta protect my voice whoops. The problem appears to be stemming from not enough liquid in the vapor Chambers. What’S strange, is that AMD says they traced the problem to a bad batch of coolers and that customers can get in touch with customer support to figure things out, but that doesn’t really make sense, because if it only affected a single batch, it should be relatively simple To use serial numbers to figure out where those cards are and preemptively recall them. Um I mean yeah. Is it it’s yeah like? Why are they relying on customers to test a thousand dollar GPU to figure out if it’s defective like what are they? What are they? Your RMA Department um this either means that the defect isn’t confined to a single batch or that AMD just doesn’t really care, and as long as these gpus don’t die within the warranty period, they’re just ultimately not going to be their problem.
The larger problem is that AMD is super low on 7900 XTX stock, so only about one out of three of customers affected will be able to get a replacement card in the first two weeks in Germany. Devour recommends just taking a refund from AMD and buying an aib 7900 XTX. Instead of waiting for AMD to replace your reference card, or if this is just kind of like um stupid, you could just not buy an AMD card, but then you’re supporting Nvidia who has their own problems. I know right.
There’S been some speculation that gpus have died as a result of this hot spot. So der Bauer received a broken card from a viewer to tear down, and you absolutely should watch the moment that he turns it on, for the first time not safe for nerds um. With the disassembly done, it looks like there was a faulty vrm, so the hotspot temp had nothing to do with the GPU dying uh with a better cooler. It still would have let out the Magic Smoke, though uh some bonus not safe for nerds content.
6900 XTS are also appearing to be dying in large numbers, with on with one German shot receiving 48 Dead gpus uh. The picture of one of the gpus is truly horrific. Let’S see yikes wow did that die just crack. What makes a die crowd? Oh, it came off on the cooler.
Why would it come off on the cooler what the hell wow, okay huh with that said, 48 is not necessarily an enormous number. It just depends on how many this particular uh, this particular store, sold yeah. Some people are speculating that a driver issue caused this. Although much like the faulty vrm GPU that der Bauer showed um, it is possible for a driver to kill a GPU like for real, like a driver, could potentially tell it to to draw way too much power.
Um or it could it could disable a thermal safety. You know yeah it’s possible, but I think that’s fairly unlikely. Our discussion question is: if you were AMD, how would you try and solve the bad Vapor chamber problems? I mean we had some problems with screwdrivers. We contacted everyone who bought one.
We had them keep their existing driver. We don’t serialize the screwdrivers, because how like realistically there’s there’s no there’s no onboard programming like we wouldn’t have a way of actually serializing them that couldn’t be uh that couldn’t be forged right. So what are we going to do? Have a different plastic mold for every stupid driver? What are we going to put a sticker on it like whatever we don’t serialize them? So what we had people do was just Mark the ones that we knew were bad. Just scratch an X send a picture to us of the scratched one, so that we’d know that.
No, that one does not have any existing or does not have any remaining warranty, and then we sent them a brand new one that has a brand new warranty. Um, that’s how we would do it AMD has, you know, can serialize cards though, so maybe they could, you know, have people return the card at least, and then they wouldn’t be out the entire cost of the card. I don’t know, would they would they Salvage memory chips, probably not even they’re? Probably just you know, I would like to know the answer to that.
What does happen to a legit like I’m, not talking a customer return, because I know where that’s going to end up that ends up in the open box, pile right. What happens to a legit dead, GPU memory chip goes bad. Is someone remanufacturing that I don’t know? I I’d be really I’d, be I’d love to follow the Journey of I I I legitimately dead GPU. Does it just go into an E-Waste pile? I don’t know probably, but then some of them are extremely valuable, like you’ve got gpus where the memory chips on them alone are worth a couple hundred dollars right and the IT industry is notoriously stingy. I have I have no idea yeah I mean you could recycle it, but that still costs you two hundred dollars right unless someone’s going to use those chips.
But then, if you use those chips, you can’t sell, you can’t sell it as new unless you just do. I don’t know I mean it’s under a cooler Who’s, Gon na Know, .