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

My 7900XTX is running great! It's a bummer some people got paper weights though . I'm hoping we will see some solid third party cooling solutions by this summer like a new NZXT Kraken bracket.

 

So...
A) Melting overpriced nVidia
B) Burning AMD
C) Intel

There is literally never been a better time to "not buy" a GPU.

EDIT: Turing. Nvm.

 

Pretty sure the melting is a tiny issue among 4090 users, the price however is atrocious mainly the 4080 and 4060LE(aka 4070"TI").
7900XTX issue creates a lot of negative buzz to AMD that they don't need, we might see Intel overtaking AMD to counter nVidia.

I hope that AMD will have their Zen moment with Radeon.

 

the melting connector comes from I'm pretty sure the fact that nvidia never tested in a real pc case, technically it's not the connector the problem but the fact that when you close a mid or even large size tower it crushes the connector and there's no way to avoid that I checked a good 10 cases none has the sufficient width, because the rtx 4000 are already wider than rtx 3000 it became a problem, would have been ok with 1cm less
I hate test benches because their use has been completely distorted by reviewers and clearly companies too, I have a Streacom BC1 V2 but they should be kept to quick testing or troubleshooting hw not considered a real pc use test which requires a closed case

@kanenas cool share thanks, it's perfectly normal that each operator (that's what they are called, not specialists) has it's own way of doing it, we're not all right handed we don't have all the same liberty of movement, especially in repetitive work like this everyone has it's own comfort way of doing it

EDIT : the problem with igor's lab article is that this is end product testing, that happens 1-2 years after the actual product design and when the mass manufacturing tools are already in place, the problem should have been found before that in the engineering lab who designed the cooler

 

totally wrong comment
you can connect it properly all you want, this connector is incompatible with 99.99% of the market PC cases that's the problem
- the cards are wider than other models and previous gens
- we never needed 35mm of space for the cables
- we could always put the cables in whatever direction we wanted, pcie8 had no problem with heavy abuse
as soon as you close your side panel it crushes and so MOVES the connector unless you're extra careful..if your side panel isn't transparent you have no way to know it didn't move
I thankfully have a glass window and can remove the front panel on a lian li 011 dynamic xl so I made sure with a strong led light and replacing my connector from the opened front panel that the 12vhpwr it correctly snug to the gpu but if you can't see or have no access it's a gamble

edit : only the official nvidia seems okay too I tried the aftermarket asus two to 1 cable and emergency shutdown my computer after 5 minutes of testing, removed it went back to nvidia 3 to 1
it smelled like a dying rtx 3090 trust me I know it wasn't from the cables but from gpu
I don't know what's wrong and don't care unless brands start paying for my "debugging"
those braided cables on top are the old ones corsair gave with "premium" psus back in the day they are 12 years old (2010) and still the best for me

 

Nvidia has an idle GPU problem when combined with hardware acceleration

Lots over at overclock dot net with it

 

To be fair, although I don't really want to be fair here because the connector isn't good, worrying about things fitting inside the case or having enough clearance for the sake of other component (like a big air cooler being hampered by RAM modules) has always been a thing in PC building. Long graphics cards not fitting inside a PC case, especially if you aren't able to remove the 3.5" HDD cage was a very valid worry back in the day. I'd say the fact my graphics card basically blocks most of the active chipset cooler also falls into this category of worries.

I do admit it can be extremely difficult to prepare for these problems. I suppose gigantic graphics cards haven't been so common that people would have needed to not only consider the size of the card but also the clearance of the power connector/cable. It's almost a thing you need to see to figure out. Nvidia is getting close to the Apple level of arrogance, so I don't think anybody there cares, though.

 

If prices were in the realm of 500 to 700 i'd agree.
For the current prices, those small mistakes are not even remotely acceptable.

Obviously my opinion, everyone will have their own tolerances with these things.

 

AMD should just sell the Radeon division to someone who cares.

 

For me the problem is AMD behavior because anyone can make mistakes - nobody is perfect: silence, didn't admit a thing until it was obvious, like any other big greedy corporation out there without any consideration for the consumer - and they were supposed to be the good guys that would free us from the green empire...

 

I dont think anyone really believed AMD would be any different than Nvidia.

People just like to imagine things are different, and for some it becomes their "alternative facts".

 

If being let down by a cooler builder that normally knows what they are doing is enough to sell off the Radeon division, Nvidia should have been sold of 5 times the last 5 years.

 

No one doubts that AMD is just like NVidia and Intel, trying to maximize profits.

But for a company that had a quarter where it only shipped 6% of all discrete GPUs, it needed to be more aggressive with it's pricing.
Especially considering it is still at a disadvantage regarding RT and upscaling tech.