RTX 4090 has a meltdown after proper installation and only one year of use

Jensen's crowning glory. It certainly gives me confidence that people chose the right person for the top of the most liked company leaders list. And he won't even admit his company's mistake. Yep, he's the right guy for the top of the CEO list. 🤣
 
"The ongoing problem has even sucked Nvidia into a class-action lawsuit."

Waste of time. Even if there is a judgement against Nvidia, the sum will be too small to matter and only the lawyers are getting paid anything of significance.

No it isn’t, it’s about credibility, if they loose it’ll only give bullets to AMD advertising their cards as the ones that don’t melt. This potentially could end up costing nVidia way more in lost revenue from selling current gen not to mention the next gen cards than anything they could be told to pay due to the ruling. And that’s probably the only reason they could ever care about (even if they say it’s not card’s fault) and make sure the next gen doesn’t repeat the same mistakes.. I mean, don’t you remember Apple and the antenna issue when they ended up saying “you’re holding the phone wrong” yet still quietly introduced changes to the next model to rule this out once and for all? Same principle, that’s why this is not a waste of time, not by a long shot!
 
Connectors are meant to eliminate user error. It's why you don't wire your lamp directly into an electrical socket.
And here I thought those plug and socket connections were for portability. After all, you can't call an electrician every time you want to move a lamp from one end of the couch to the other..

"A plug in every socket", that was my motto in my late teens and early twenties. But I digress....

I'll grant you that the "new", phased plugs and sockets prevent reverse polarity issues.

Anyway, electronic's conventions designate "black" as ground. However electrical conventions designate black as "hot" (positive). You can imagine the mayhem that ensued when my daddy, (an electronics technician), decided to do some home wiring.. :eek:
 
OK KIdz, this link/chart shows wire gauge, supply voltage, and length of run to determine amperage capacity. It says that 10 gauge (solid copper) wire, will carry 30 amps @ 240 volts, which amounts to 7200 watts. Cut that in half for safety reasons, and you get 3600 watts on that line.


None of which serves to explain why a device having 16 connections and pulling a maximum of say 800 watts in spikes should be going up in smoke.
 
That will be using the updated ATX3.1 connector. It solves the problem the ATX3.0 connector has.
Yes, but it voids your warranty and the irony is they claimed the same thing with previous cables. There was a reported incident where a cable mod burned and the vendor refused to replace the 4090 because they claimed it voided the warranty. Remember because this time is different we promise!
 
No it isn’t, it’s about credibility, if they loose it’ll only give bullets to AMD advertising their cards as the ones that don’t melt. This potentially could end up costing nVidia way more in lost revenue from selling current gen not to mention the next gen cards than anything they could be told to pay due to the ruling. And that’s probably the only reason they could ever care about (even if they say it’s not card’s fault) and make sure the next gen doesn’t repeat the same mistakes.. I mean, don’t you remember Apple and the antenna issue when they ended up saying “you’re holding the phone wrong” yet still quietly introduced changes to the next model to rule this out once and for all? Same principle, that’s why this is not a waste of time, not by a long shot!

If you think this will cost Nvidia anything worth mentioning, you haven't paid attention to other class action lawsuits. They achieve nothing beyond enriching lawyers. The claimants all get a handful of dollars and the money hit to the losing company is not large enough for them to think twice about doing whatever they did again. If you think they take anything that even moves the needle in terms of marketing/advertising. It takes screw ups on the scale of Takata airbags situation to put a dent in the company.

In regards to Nvidia losing anything in terms of market share, hah. This is an issue that's affecting less than 1% of 4090 cards. It doesn't happen with enough frequency for anyone other than hand wringers to worry about. Their market share will be just fine, especially since they aren't focusing on gamers anymore.

In regards to Apple, you must have missed all the various other engineering "marvels" they've caught flack over that they took years to address and still deny claims on to this day. Apple is another company that hasn't been hit hard enough to change anything of significance.

Connectors are meant to eliminate user error. It's why you don't wire your lamp directly into an electrical socket.

If user error is so common then why isn't the 8pin connector having as many incidents of "user error"?
Because they're using 2-3 connectors per card directly in previous generations? They're not pumping the same wattage per wire.
 
Who would have thought that pumping more power through connectors with fewer cables and less surface area on the contacts would lead to an increase in heat.

All you need is a tiny bit of thermal expansion to create a feedback loop of increased resistance and heat.
Such heretical thoughts! :)
 
4090 is a real liability, spend all that money and if it dies outside of warranty due to a well known issue...
If it dies outside of warranty then buy another RTX 4090. After all, "the more you buy, the more you save!"
 
Any power supply equipped with the ATX3.0 12VHPWR connectors is flawed and needs withdrawn.
It has NOTHING to do with ATX3.0 standard. You are talking about the 12VHPWR cable.

ATX3.0 has new standards for dealing with transients and new OCP threshold. Those PSU are ENTIRELY fine. The problem is stupid Nvidia trying to save PCB space by using a 12VHPWR connector on their GPUs.

You are propagating FUD.
 
It has NOTHING to do with ATX3.0 standard. You are talking about the 12VHPWR cable.
I'm talking about the plug on the end of the cable. And yes, the plug selected for 12VHPWR is part of ATX3.0 standard. ATX3.1 has replaced it with a newer plug, called 12V-2×6.

They're plug compatible I believe. The new name is just to create separation from this controversy.
 
Last edited:
Assuming the new ATX3.1 connectors have been used, then they are different to ATX3.0.

The important thing is to get rid of all the ATX3.0 12VHPWR connectors. They don't work.
They do work at less than 1% reported burn risk. Also the new atx 3.1 cables aren't even ready for prime time. Unless you know of one I can purchase currently? A few things you can do for now is monitor your 12 volt 16 pins connector using techpowerup's gpuz to monitor any deviation, also measuring the temperature of the cable as some suggested and an atx 3.0 psu. Lastly all this has to be done without warranty void and minimal consumer cost as a recall at least on the cables imo. Last thing we need is to unknowingly void our warranties because of mass hysteria.
 
Last edited:
For me, since I haven't yet purchased an ATX3.0 power supply, it's a simple case of not buying any until the ATX3.1 power supplies arrive.

For those that have the defective cables, they should fully expect a free replacement. I'm guessing that all ATX3.0 spec'd power supplies, currently for sale, are also modular. That makes life easier for everyone involved since just the one cable needs replaced.
 
Last edited:
For those that have now got burnt sockets on the graphics card, they'll need to discuss RMA conditions with the retailer to get the card damage remedied.

Ignoring the problem is not a good idea. If used, the 12VHPWR plug is likely to continue to cause problems into the future.
 
Back