Because they still produce and specify most of the connector specs, so you can reference to them as the standard.
There were heaps of things wrong with the old four-pin connector. Not least of which being that nothing positively maintained pin alignment, often making it difficult if not seemingly impossible to plug stuff in. I have had a few incidents where the barrel wall wedged itself into split pins due to how loose tolerances are on these things.
Absolutely 100% right ... an a $1 Trillion company called nVidia did absolutely NOTHING about it and try to absolve themselves of any wrong doing under "ignorance is bliss" approach to engineering.
Already in progress, their legal team may not be aware yet as evidence gathering continues. This IS going to cost nVidia A LOT and they can't hide behind "it's someone else's specification not ours" ... not gonna work in court.
I have a Seasonic 1600W and it's been extremely reliable, but I still do verify that the pins don't push beyond their locking arms whenever I plug any connector into anything GPU, Motherboard, etc. Some of the connector quality used by other brand power supplies is shockingly bad.
i dont recall any hdd fires due to the power connector ( 4 pin molex )
My "absolutely horrible" comment was about the old 4-pin connector. HPWR is a godsend by comparison.
Manufacturers have demonstrated multiple times that their cables are able to deliver over 1000W without issues. Good luck proving melted connectors aren't caused primarily by mishandling damage and normal wear.
The consumer space doesn't want to pay for that and you also need to keep in mind that when the PCI-SIG adopted the HPWR connector, it was intended to eventually replace 6/8-pin cables. That would affect $150 GPUs and $40 PSUs too, not only the high-end.
Is there a new lawsuit? There was a class action started in 2022 over the RTX 4090/12VHPWR melting, but the plaintiff dropped it in 2023.
The spec states the connectors should be UL94V0, so I have no idea what you're talking about saying it's not UL certified. I have not seen a single case of a 12VHPWR cable actually causing a fire, it's just sensationalist reporting treating 'melting' and 'bursting into flame' interchangeably.
And you know this how? The consumer is spending $2000-$6000 on these GPUs ... you seriously believe mil spec is going to make a difference in what they want to pay?
In the works, not filed.
"likely" ... or none at all ... they had similar problem in other high AMP GPUs not just the 5090 ... the incidents are stacking up and it's NOT looking good for nVidia. Sweeping this under the rug as "user error" isn't going to work this time around ... nor blaming it on PCI-SIG.
Which of those points is 12VHPWR/12V2x6 failing? And what are you looking at for evidence?
Fail!
How do you know about it if it hasn't been filed?
Yes, and noone would internally design an out of spec card for commercial sale. Thus a 5090 would require 4x 8pin.
PC connector specs have to consider the whole PC market starting from entry-level, not only $2000 GPUs.
When Nvidia is doing its silicon qualification runs, it has hundreds of pre-production cards running in test centers. If melting connectors were a major issues, they would have seen it there first.
Without knowing the specifics of their testing hardware and if they test using "controlled" conditions that ARE NOT representative of end user conditions? There is an interesting article on VideoCardZ from Intel engineer's perspective is nVidia are being irresponsible and not doing the bare minimum to ensure safety of 4090/5090 and now 5080 GPUs having same issue.
In some cases yes, in many cases no, first run, all stock cables provided by manufacturer. Doesn't matter how you slice it, running a cable at it's threshold of tolerance on "consumer" devices is irresponsible at best and criminal negligence at worse given it was a known issue going back to 4090. When it comes to public safety, I can guarantee you a judge is going to rule in favor of consumers.
And even at 200W still would have huge safety headroom. 3*8 pin + pci-e could reliably supply 675W and still have 250W headroom or so.
I have yet to hear about a single story of a HPWR cable starting an actual fire.
Their use of molex connectors are incorrect and need definately a better connector.
Yes son, it’s me!
You talk about safety in an impudent and cynical way. Please show us all your crazy ideas by your own example. We will be happy to watch this performance.
People are trying to write about this as if it was a super-urgent fire problem. It isn't. None of them have actually caught fire.
Yes, and the damage should not extend past the computer. It would be nice if customers could somehow be sure that the power in their setup is evenly distributed in the cable wires. It hasn't been established if it changes with every connect/disconnect or if it is something inherent in the GPU, PSU or cable. If I had the card, I'd be worried that one of my wires was headed for a meltdown. What could I do right now to make sure that is not the case? I don't think touching it is enough.
Facebook
Twitter
Instagram