News PCI-SIG Tells GPU Makers to Improve Testing in Response to Nvidia 12VHPWR Lawsuit

Page 2 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
The RTX 3090 FE (and other high end RTX 3000 FEs) did use use this 16 pin connector (well, minus the 4 sense pins, that don't deliver power anyway). The 3090 Ti even had the same 450W TDP as the 4090.

Hence why I wrote "many". In any case far as I remember (and I'am to lazy this morning to look at it lol) the 12 pin power connector for the RTX 3090 FE was facing in a different direction, right ?
 
Unless you are an electrical engineer who has run the numbers, your thinking on this matter has no merit.

No one said you had to agree or understand it. That doesn't change it. Just wait for the lawsuit to finish and the forced recall to happen and then hopefully NVidia and PCI-SIG wont be so stupid next time. At least AMD figured it out
 
Fun experiment. One of my colleagues bought in his 4090 and had everyone plug the connector in.

Five people out of ten people failed to plug the connector all the way and safely seat it - three of them build their own PC's.

I'm flabbergasted. It's almost impossible to see that you've not plugged the connector in, it's clear as day, makes me wonder how some of these folk make it through the day.

If people are this sloppy, it's time for an idiot-proof connector as much as a safer one.
 
There was a design defect in the Ford pinto that if you hit the rear of it. The suspension ran the risk of piercing the fuel tank setting it on fire. These days if there's a risk of even the car stalling out on a highway there's a recall.

Same deal. It's a known safety issue. It needs to be fixed.
 
  • Like
Reactions: RodroX
User error is the issue. You should be confirming every connector is fully seated and latched visually, not assuming just because you pressed it in "real hard" that it is properly connected. They don't need to over-engineer the connector to account for idiots. Simply add the warning that this connector needs to be fully connected and that they recommend installation by a trained professional.
 
User error is the issue. You should be confirming every connector is fully seated and latched visually, not assuming just because you pressed it in "real hard" that it is properly connected. They don't need to over-engineer the connector to account for idiots. Simply add the warning that this connector needs to be fully connected and that they recommend installation by a trained professional.
Especially because there is an endless supply of such idiots. If every engineer had to make products for every unique kind of stupid we would have nothing but OEM computers. If you make a mistake, own it and learn from it and move on. Don't sue people and proclaim it wasn't your fault to begin with.

Both 8 pin and 12 pin connectors are susceptible to the same issues, however, because the new cable is bigger and longer with more surface area contact it makes the plug easier to not plug in all the way.
 
User error is the issue. You should be confirming every connector is fully seated and latched visually, not assuming just because you pressed it in "real hard" that it is properly connected. They don't need to over-engineer the connector to account for idiots. Simply add the warning that this connector needs to be fully connected and that they recommend installation by a trained professional.

The tag "install only by a professional" would put the consumer division out of business. There is an instance of an professional OEM not seating a 4090 connector properly, resulting in fire.

If the connector is prone to insertion failure then it's NVIDIAs problem. They should have made one pin shorter to ensure full insertion.
 
  • Like
Reactions: RodroX
Especially because there is an endless supply of such idiots. If every engineer had to make products for every unique kind of stupid we would have nothing but OEM computers. If you make a mistake, own it and learn from it and move on. Don't sue people and proclaim it wasn't your fault to begin with. Both 8 pin and 12 pin connectors are susceptible to the same issues, however, because the new cable is bigger and longer with more surface area contact it makes the plug easier to not plug in all the way.

User error is not the issue. Design is the problem. More specifically, not leaving enough resistance headroom during design phase is the problem. I've seen plenty of situations where an 8-pin was not plugged in all the way and there was no trouble. The 8-pin standard has been around for 10+ years. How often do you see reports of an 8-pin plug causing a fire from not being plugged in all the way? I cant remember the last time I saw one. Its extremely rare. It almost never happens, and not nearly as much as it does to the 12-pin. I wonder how many instances of 12-pin melting there were when the user made sure it was plugged in all the way, but slowly came out unplugged over time due to weight, gravity, and just the naturals of how the cable sits, when they stood up their case after the build was done and the plug came out just a hair. This can happen.

The PCI-SIG documentation says that the cable can't be safely bent before 35MM to the plug. I dont see any 8-pins with that stipulation. Do you? No? What other electrical cable anywhere on the face of gods green earth do you know of that has a bending stipulation? Anything? No? Oh, me neither. Obviously the designers knew what was up after testing. They knew it was a safety hazard. So, rather than fix the product or scrap it like an intelligent human being would (this costs companies profit), its easier just to fudge the cable a bit, put a bend stipulation on it that no one can adhere to because there is no case that will support that 35MM offset, nevermind the fact that the 35MM straight offset wont stay there, because gravity exists, then, when people's 1600$+ GPUs start melting, and the lawsuits start coming: "Did you leave 35MM away from the plug before bending it?" can be blurted out in the courtroom to pump the user error argument. This is common business practice, equivalent to "fine print" or "the agreement" that they know no one reads. But its not going to fly in court. Its going to backfire

"Install only by a professional."

At the end of the day, all we are doing is plugging an electrical cable into a socket, something mankind has been doing for hundreds of years. What could possibly be so skilled about plugging a cable into a socket, that we need a professional to do it? So now we are done with DIY PC's without a safety compromise? Again, in the court room: "Did you have a professional plug it in for you? Are you a qualified professional? No? USER ERROR!!! LAWSUIT DISMISSED!!!!" Lol. Do you see how this works? These stipulations are designed to protect NVidia in the courtroom. Again, common business practice. Who knew plugging a cord into a socket could be such a skill intensive job?

Ugggh, I could go on forever.....

Gawd dammit. Quit blaming the user and design your products properly. AMD didn't use it on the new Radeon 7000 series because they knew exactly what was up. AMD was thinking. Holy crap were they thinking. Look at that 12 pin connector. Look at those tiny little pins on that stupid thing. You aren't SAFELY jamming 600W through all those tiny little pins on one connector, not even close. There is a reason we had 8 pin connectors. The 8 pins leave much more headroom and are far safer. "We made the new 12 pin to save PCB space" goes NVidia. Lol. There is still plenty of PCB real estate to include 4 8-pins. This 12-pin garbage is just not happening. How did that connector even make it past the design phase?

If someone offered me a 4090 for free, I wouldnt use it. I would off it for 80% MSRP and tell the buyer: You wanna burn your house down, you do you. It's just not worth the risk. Stop delaying the inevitable and quit being so stubborn, and quit blaming the user for your mistakes. Just get the recall over with already so we can move on. Let it be a lesson learned. Stick with the 8-pins. Many a wise man have said: Don't fix what ain't broken. I hope NVidia gets pummeled in this lawsuit. They deserve it for senselessly allowing the safety of their customers to be compromised, and then they have the audacity to blame the user. Lol. Get a grip!!!!
 
User error is not the issue. Design is the problem. More specifically, not leaving enough resistance headroom during design phase is the problem. I've seen plenty of situations where an 8-pin was not plugged in all the way and there was no trouble. The 8-pin standard has been around for 10+ years. How often do you see reports of an 8-pin plug causing a fire from not being plugged in all the way? I cant remember the last time I saw one. Its extremely rare. It almost never happens, and not nearly as much as it does to the 12-pin. I wonder how many instances of 12-pin melting there were when the user made sure it was plugged in all the way, but slowly came out unplugged over time due to weight, gravity, and just the naturals of how the cable sits, when they stood up their case after the build was done and the plug came out just a hair. This can happen.

The PCI-SIG documentation says that the cable can't be safely bent before 35MM to the plug. I dont see any 8-pins with that stipulation. Do you? No? What other electrical cable anywhere on the face of gods green earth do you know of that has a bending stipulation? Anything? No? Oh, me neither. Obviously the designers knew what was up after testing. They knew it was a safety hazard. So, rather than fix the product or scrap it like an intelligent human being would (this costs companies profit), its easier just to fudge the cable a bit, put a bend stipulation on it that no one can adhere to because there is no case that will support that 35MM offset, nevermind the fact that the 35MM straight offset wont stay there, because gravity exists, then, when people's 1600$+ GPUs start melting, and the lawsuits start coming: "Did you leave 35MM away from the plug before bending it?" can be blurted out in the courtroom to pump the user error argument. This is common business practice, equivalent to "fine print" or "the agreement" that they know no one reads. But its not going to fly in court. Its going to backfire

"Install only by a professional."

At the end of the day, all we are doing is plugging an electrical cable into a socket, something mankind has been doing for hundreds of years. What could possibly be so skilled about plugging a cable into a socket, that we need a professional to do it? So now we are done with DIY PC's without a safety compromise? Again, in the court room: "Did you have a professional plug it in for you? Are you a qualified professional? No? USER ERROR!!! LAWSUIT DISMISSED!!!!" Lol. Do you see how this works? These stipulations are designed to protect NVidia in the courtroom. Again, common business practice. Who knew plugging a cord into a socket could be such a skill intensive job?

Ugggh, I could go on forever.....

Gawd dammit. Quit blaming the user and design your products properly. AMD didn't use it on the new Radeon 7000 series because they knew exactly what was up. AMD was thinking. Holy crap were they thinking. Look at that 12 pin connector. Look at those tiny little pins on that stupid thing. You aren't SAFELY jamming 600W through all those tiny little pins on one connector, not even close. There is a reason we had 8 pin connectors. The 8 pins leave much more headroom and are far safer. "We made the new 12 pin to save PCB space" goes NVidia. Lol. There is still plenty of PCB real estate to include 4 8-pins. This 12-pin garbage is just not happening. How did that connector even make it past the design phase?

If someone offered me a 4090 for free, I wouldnt use it. I would off it for 80% MSRP and tell the buyer: You wanna burn your house down, you do you. It's just not worth the risk. Stop delaying the inevitable and quit being so stubborn, and quit blaming the user for your mistakes. Just get the recall over with already so we can move on. Let it be a lesson learned. Stick with the 8-pins. Many a wise man have said: Don't fix what ain't broken. I hope NVidia gets pummeled in this lawsuit. They deserve it for senselessly allowing the safety of their customers to be compromised, and then they have the audacity to blame the user. Lol. Get a grip!!!!
Its very clear to me that you have no idea what you are talking about on a physics, engineering, or statistical evaluation level. I do not mean this as a slight. You have strong opinions based in consumer rights and advocacy. I will leave it there, as there is no sense on trying to move an immovable object or perpetuate toxicity on this forum. I hope you have a great day and even better holidays.
 
  • Like
Reactions: TJ Hooker
Do you have a link you can share?

According to article it was a cyber power pc
Two more reports followed, including one where a YouTube user claims to have run into the same problem with a CyberPowerPC pre-built desktop equipped with an RTX 4090.



 
Its very clear to me that you have no idea what you are talking about on a physics, engineering, or statistical evaluation level. I do not mean this as a slight. You have strong opinions based in consumer rights and advocacy. I will leave it there, as there is no sense on trying to move an immovable object or perpetuate toxicity on this forum. I hope you have a great day and even better holidays.

Companies have professional mathematicians roll the numbers and the potential cost of human lives plus lawsuits vs additional engineering safety.

Ford had a minor issue with engines suddenly stalling when people stepped on gas from a stop. The risk was incredibly small and there was only a handful of incidents. Was there a full recall? You bet ya. That's a good way to get t boned.
 
Its very clear to me that you have no idea what you are talking about on a physics, engineering, or statistical evaluation level. I do not mean this as a slight. You have strong opinions based in consumer rights and advocacy. I will leave it there, as there is no sense on trying to move an immovable object or perpetuate toxicity on this forum. I hope you have a great day and even better holidays.

It's also very clear to the flat earth society that NASA doesn't have a sweet clue what they're talking about. Yes, I do have strong opinions..... based on the holy trinity of logic, reason, and common sense. No one says you have to agree. Disagreeing doesn't change it
 
Last edited:
Companies have professional mathematicians roll the numbers and the potential cost of human lives plus lawsuits vs additional engineering safety.

Ford had a minor issue with engines suddenly stalling when people stepped on gas from a stop. The risk was incredibly small and there was only a handful of incidents. Was there a full recall? You bet ya. That's a good way to get t boned.
Since "Companies" have professional "mathematicians" roll the numbers that means you know for a fact that Nvidia, PCI-SIG, and both cable manufacturers have them for a fact? This also implies that if they did have them they all concluded together or separately that the cable they designed was flawed in some way but "good enough" on a cost analysis perspective to release anyways? These assertions you are making rely on a lot of either inside information of these companies or assumptions based on some other companies and how they operate. Just because one company operates one way that does not mean they all operate that way.

What does Ford have to do with these companies? They are completely unrelated. As an example of what Ford does this may apply to how Ford would act or respond to a similar potential issue, not any of the companies we are talking about. Just because big company X responds in Y manner does not mean big company Z will also react or respond in the same Y manner.
 
According to article it was a cyber power pc



No where in the entire article does it say that the connector or graphics card were on fire or caught fire at any point. The two instances that its referring to was just that OEM computers with connectors that melted presumably from the assemblers inability to fully connect the connector or possibly that the connector became loose or otherwise in shipping and the purchaser did not look over the computer before using it. Do you have another reference that says specifically that they were on "fire?"
 
Last edited:
  • Like
Reactions: TJ Hooker
The PCI-SIG documentation says that the cable can't be safely bent before 35MM to the plug. I dont see any 8-pins with that stipulation. Do you? No? What other electrical cable anywhere on the face of gods green earth do you know of that has a bending stipulation? Anything? No? Oh, me neither.
It looks like the old 6/8pin connectors may have a similar stipulation. Page 22 here: https://www.molex.com/pdm_docs/ps/PS-5556-001-001.pdf
The “T” dimension defines a “free” length of wire, or a length of wire that is not subject to significant bias
by external factors such as a wire tie, wire twisting, or other means of bending or deforming of the wires
that repositions them from their natural relaxed state or location where they enter the housing. Wires are
to be dressed in such a manner to allow the terminals to float freely in the pocket.
Tminimum for an 8 pin connector is listed as 19.1 mm, 12.7 mm for 6 pin.
 
  • Like
Reactions: helper800
Companies have professional mathematicians roll the numbers and the potential cost of human lives plus lawsuits vs additional engineering safety.

Ford had a minor issue with engines suddenly stalling when people stepped on gas from a stop. The risk was incredibly small and there was only a handful of incidents. Was there a full recall? You bet ya. That's a good way to get t boned.
I'd say there a fundamental difference between issues that stem from design and/or manufacturing defect, such that they exist regardless of user action, and issues that stem from user (mis)use.

If I get a car fresh off the production line and the wheel nuts aren't properly tightened, is the manufacturer liable if my wheel comes off while driving? I'd imagine so. If I change a tire on my car and fail to properly re-tighten the nuts, is the car manufacturer (and/or wheel manufacturer) still liable? I highly doubt it.
 
Last edited:
It looks like the old 6/8pin connectors may have a similar stipulation. Page 22 here: https://www.molex.com/pdm_docs/ps/PS-5556-001-001.pdf

Tminimum for an 8 pin connector is listed as 19.1 mm, 12.7 mm for 6 pin.
Like I said, "Both 8 pin and 12 pin connectors are susceptible to the same issues." Thanks for finding the exact reference to my claim, I could not find it anywhere for reference.
 
  • Like
Reactions: TJ Hooker
Companies have professional mathematicians roll the numbers and the potential cost of human lives plus lawsuits vs additional engineering safety.

Ford had a minor issue with engines suddenly stalling when people stepped on gas from a stop. The risk was incredibly small and there was only a handful of incidents. Was there a full recall? You bet ya. That's a good way to get t boned.
No where in the entire article does it say that the connector or graphics card were on fire or caught fire at any point. The two instances that its referring to was just that OEM computers with connectors that melted presumably from the assemblers inability to fully connect the connector or possibly that the connector became loose or otherwise in shipping and the purchaser did not look over the computer before using it. Do you have another reference that says specifically that they were on "fire?"

Someone made the assertion that Nvidia should slap a label that says "professional assembly only". I said even the "professionals" screwed it up. I didn't claim it was a faultless installation. I was citing the fact the connector was poorly designed without a safety feature which allowed it to be incorrectly installed.

Can Nvidia be held liable? Possible.
 
I'd say there a fundamental difference between issues that stem from design and/or manufacturing defect, such that they exist regardless of user action, and issues that stem from user (mis)use.

If I get a car fresh off the production line and the wheel nuts aren't properly tightened, is the manufacturer liable if my wheel comes off while driving? I'd imagine so. If I change a tire on my car and fail to properly re-tighten the nuts, is the car manufacturer (and/or wheel manufacturer) still liable? I highly doubt it.

You know they used to put tents over baby cribs. They were designed to prevent kids from crawling out in the middle of the night and getting hurt.

Problem is there was an outside risk of a child choking on them. Almost overnight they all disappeared.

Products defects that can cost lives need to be recalled.