![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
I thought it was already established that it's a design flaw and not a user flaw? This keeps happening to a lot of users would suggest that Nvidia did a poor job designing the connector. Exactly how many other GPU's in history ever had a problem from melting connectors? Well, besides this one 6950.
They aren't likely to have errors with it. The number of actual cases is infestiminally small compared to the number sold. There's always going to be a handful of people who have an issue with anything. This is below normal numbers even. If anything, it's well-designed by your view.
There is a reason why engineering controls (connector design) is placed higher than administrative controls (teaching people to properly plug things in) in effectiveness. Whenever you have the possibility of taking human error out of the equation, you should do so. And they did so (presumably) with the 12V-2X6 connector.
Infinitesimally small is still not 0. If you can design a connector that has 0 chance of user error short of physical damage (they did if the testing is to be believed), why not use it? It's not like the connector costs more or the R&D was extremely expensive.
Common sense stuff; didn't think those basics needed to be elaborated upon. You will never reach zero, human stupidity knows no bounds. The defect rate on these us is below average, so by your logic it's actually a very robust design.
That is something I would like to see actual stats on, but I don't know that anyone has them or has released them: What is the failure rate, regardless of the cause, in terms of failures per million units and then equally important how does that compare to other high-current connections like 8-pin PCIe power. While there's been lots of tech-tuber coverage, that doesn't mean it is common, just that it is popular to talk about. However likewise, just because it isn't happening to most people doesn't mean it isn't a bigger problem than it should be.
Neither of us said anything about what an average defect rate should be. Stop putting words in our mouths.
Then why did they redesign it?
If your car caught fire 1/100th of a percent of the time because you didn't shut the door over the gas cap properly, nobody'd try to claim it was fine. IIRC there have been NHTSA-mandated recalls in the past for as few as 7 incidents.
This is important. If the original version of the connector was fine, why fix it? nVidia could've done the Apple "you're holding it wrong" thing. But they didn't.
Yeah, my MSI SuprimX 4090 has a hard-set limit of 520W, well below the 600W some of these are allowed to push. This was also a theory of mine as well. Honestly, I get even better OCs and benchmarks than most cards with the 600W limit. It is still pretty much silicon lottery on GPUs for OCing the Core/Memory. Honestly, past benchmarking or a demanding game like CP2077 at 4K, I generally leave it at default power limit with a medium OC.
Nah the design is garbage which is why they already v2’d it and the 12VHPWR connector has been phased out and replaced with the 12v- 2x6 connector. Which is subtly different from the 12VHPWR connector, but backwards compatible with the cable.
This doesn't catch fire, it just stops working because the plastic deforms some.
I'm pretty sure it will be more clearance recommendations rather than requirements. The sense pins should disengage and limit power if the connection wasn't good. The PSU manufacturers will still claim requirements though because they don't know whether you're using a GPU with the 12VHPWR connector or 12v-2x6.
Most PSUs have 2 8-pin GPU connections per single 8-pin PSU connection. The number of 12v wires is 3 per connection. Those 3 wires (and connectors) are already handling 300 watts. 6 wires at 600 watts is the same amount of power per pin and wire. There was arguably too much overkill in the 8-pin design but changing power allowances without changing the connector would lead to people overloading single 8-pin GPU cables.
That's just semantics and you know it. A robust design would have prevented physical damage from reasonable human negligence. It can be hard to spot a connector not seated in properly when working in tight spaces, i.e. ITX cases.
No, it isn't. A fire can be dangerous, while a melted connector is an inconvenient rma for a fix.
Alright, I'll put words in your mouth this time. So basically you're saying it's okay for electrical connectors to be designed such that only in perfect conditions will they not melt and as long as they only melt and don't burn it's perfectly acceptable. And by extension, that goes for everything else that can be damaged by a moment of overlooking things. As long as it doesn't cause injuries, property damage as a result of minor mistakes is absolutely okay.
Well, something you do have to keep in mind is that you can't have something that is 100% safe or failure proof. If your requirement for a power connection is one that can never melt, never fail, etc then you have no connections you can use, including even power cords like NEMA 5-15P. You can't demand "no failures, ever, even if not used properly."
Every electrical connector/plug known to man has failed and melted or burned. They still do to this day. Simple 110v plugs melt or burn every day.
No, you are not wrong. People seem to ignore the countless melted 8 pin pcie connectors in favor of bitching about this connector because Nvidia, plain and simple.
The increase at the amount of idiots is what kills me, it's exponential. I always knew there were a lot of idiots when I was younger, but never in my wildest dreams did I ever imagine that idiots would start to dominate many facets of our everyday lives, let alone a geek niche such as computer hardware.
I remember the same issue, the 8 pin connectors were being worked loose as people were tugging things to make the cases look pretty and as a result the high voltages on those were melting and blah blah blah, hence the need for a new design. Be it the 12VHP, 2x6, or the upcoming ASUS and MSI board powered slot. The new connectors didn’t appear in a vacuum they were all dealing with too many warranty returns and sought out solutions.
It's not just fashion, it is practical, lots of cables can be an issue in all sorts of situations. Servers would be an example. Suppose you have a system that has 4 high power cards in it, which is not unheard of. You wanna run 16 cables? How do you even route all that and not screw up airflow? I mean you could make the same argument going back to OG AMP mate-n-lok/molex connector. Just use 6-8 (or more) of those!
And it seems that nVidia hired those idiots to design the connector. This is coming from a guy who has been building PC's since the early 90's and has always used nVidia cards until I got my son a 6600xt last Christmas. The other 4 computers in my house use a RTX 3070, two RTX 3060ti's and a GTX 1070. It is OK to admit that this connector was a bad design and I am glad that it has been updated.
Something like card edge connectors, instead, would be my first thought.
It occurs to me now that maybe many of the people putting the blame squarely on the connector may not deal with your average consumer on a daily basis. I, for better or for worse, have to deal with hundreds of average people a day. And let me tell you, their idiocy knows no bounds. The thirst for doing the most moronic things is unquenchable.
So think about how dumb the average person is. Imagine HALF of all people are dumber than that.