You'd think SOMEONE would've checked if the cables (during and after finalizing the design) were sturdy...
Well, at the least, that's added to a checklist.
Why would nVidia care about that? They know that there's enough clueless noobs who will buy anything in a green box no matter what happens. The problem is that they're 100% right.
I would not be surprised if the development was rushed to meet a deadline.
And that's where things always go wrong. Remember the
Atari E.T. game?
IIRC the 4090s were built well ahead of time and held for a strategic release date. But development may have been too rushed somewhere along the line. Looking at the connector construction... this is an awful piece of work.
I know, eh? It's like when the USSR rushed the building of Chernobyl. When you care more about the propaganda surrounding the product than the product it self, things can go wrong in spectacular fashion. Hey, it's nVidia though, what do you expect for "only" US$1,600, quality?
As mentioned, a recall should be in order... I would think that Nvidia, or "whoever manufactured the adapter", should take responsibility though a recall of the defective item. I don't see where it is different from an auto manufacturer that puts out a vehicle with defective seat-belts or airbags. If the product has been discovered to be defective, somebody should take ownership and make it right.
Yeah right, like nVidia would ever take responsibility for anything!
Here's a little reminder of how nVidia operates from the original tech investigative journalist, Charlie Demerjian:
JUST WHEN YOU thought things could not get any worse, Nvidia’s Jen-Hsun Huang gets in front of the press and blames TSMC for Fermi’s problems. Yes, you heard it right, Fermi’s del…
semiaccurate.com
To the final conclusion of the article poster: Better I buy no 4090 and no ATX 3.0 PSU, and keep my good ATX 2.0 PSU, guy. The way some people support Nvidia on this tech site is just unacceptable.
Yeah, I understand the sentiment and I do share it to some degree but I think that's Shawn's advice to people who were dumb enough to by a 4090 in the first place and are now stuck with it. Let's face it, for a lot of the jokers who think that US$1,600 is reasonable for a video card, even warning them about this in advance wouldn't stop most of them from buying it anyway.
You'd think a $1600 item would get a cursory pinch of QA, if any!!
Do they test these combustible paperweights at nGreedia before they unleash them upon its adoring buyers or is that too much to ask??
Oh come on now, we both know that nVidia expects
at least US$2,000 before any kind of serious QC is considered!
Has anyone tried buying an ATX 3.0 PSU? Here in the UK at least, I cannot find any in-stock or even listed on most popular tech shops.
Seasonic won't have theirs out until December so I assume we might actually get some ATX 3.0 PSU's early next year?
I honestly don't care since I'm not stupid enough to buy video cards that require them. My EVGA 1000 G2 Supernova will still be in service ten years from now because Radeons don't need special connectors to just work.
Well, there's people out there who use GPUs for their workflows and a fast GPU like the RTX 4090 is an investment, assuming that it doesn't catch fire. Getting an ATX 3.0 PSU, or even simply a 90° 16pin adapter, could save them a lot of time and money.
An RTX 4090 is NOT a workstation GPU. You don't just take a gaming card, pop it into a workstation and expect it to work well. The Titan cards all had a separate driver set for professional use which is why Jim at AdoredTV did a video about how the RTX 3090 is NOT a Titan. The same goes for the RTX 4090. It's not a workstation card, it's a flagship gaming card and without the proper workstation drivers, it falls flat in a workstation. Here's the video:
Just me. I never buy new release bleeding edge stuff. I wait for the guinea pigs to finish testing the new stuff. Besides, I refuse to pay a premium for goods that no one really needs.
Good call on that!
Hear me out... great opportunity for Nvidia to sell Founder's edition power adapters HAHAHAHA
Heh, that sounds EXACTLY like something that Jensen would try!
It is not possible to escape with no responsibility. You can delegate responsibility, but at the end of the day, the adapter is in the box of the product, and thus, you will be accountable.
And what will be the outcome of that? People will smarten up and not pay through the nose for nVidia cards? I think that we have a better chance of putting a live human in orbit of Neptune within the next ten years.
And in this case, the bad soldering may not be just a few or affecting certain batches. For all you know, it may well been the design based on the price that Nvidia and partners are willing to pay, and so may not be a QC fault. Now with this becoming an issue, they will have to rectify/ improve on the quality of the adapter.
Well, with what EVGA said about nVidia making it impossible for AIBs to actually make money, it could be that the AIBs were stretched as thin as can be so they looked for what seemed to be a reasonable way to save on costs.
Well well well, I'm curious to see what NVIDIA is going to do with this problem. Obviously they can't re-design the connector. Probably sending a new "quality" cable with "quality" adaptor to everyone who bought a 4090 ?
It's not as simple as that because the female connectors on the cards themselves have also melted. These cards all have to be replaced.
Looks like the connector was designed in a shed, a disaster waiting to happen. it is not about execution, not a bad solder, not bad qa. You have those double thick wires in order to send a lot of amps but one can't solder those double wires to a thin plate and expect it to not melt while 50A pass through the tiny section of the contact.
You're 100% right about that. A circuit is only as robust as its thinnest and weakest contact point. Maybe they should have used a fuse.
