• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

RTX 5090 FE Molten 12VHPWR

It will be Karma on here when some people experience this problem. Just amazes the lengths some will go to defend a poor design, but some people also defended the Pinto despite a terrible design flaw.
 
Gideon said:
It will be Karma on here when some people experience this problem. Just amazes the lengths some will go to defend a poor design, but some people also defended the Pinto despite a terrible design flaw.
Click to expand...
It may not be the ideal design, but it works for many millions of people just fine. The melted connections are just being sensationalized heavily by tech media looking got clicks. The actual percentage is lower for defects from what I've heard, than prior cards, including the tiny number of actual melted connectors.
 
GoldenTiger said:
It may not be the ideal design, but it works for many millions of people just fine. The melted connections are just being sensationalized heavily by tech media looking got clicks. The actual percentage is lower for defects from what I've heard, than prior cards, including the tiny number of actual melted connectors.
Click to expand...
and what of the statistics for those using adapters with those connectors?

can you cite your sources above also
 
Sources can't be cited because Nvidia reddit mods keep deleting melted connector posts. I can say that I've seen way more native 16pin cables melted now that they've been out for a few years. The adapters seemed more common in the beginning, but that quickly shifted once people started spouting "adapters are the problem, use native cables instead!"

It's obviously a shitty design that easily fails. We wouldn't have so many temperature, wattage, and current monitoring products for the connector if it weren't so shit.
 
Yea, it's just the connector in general, it's not about adapters.

Because there is no load balancing, modern GPUs that are doing 300w and beyond can easily overload some pins. There is not many reports of AMD GPUs or cards below the 5080/4080 TDP range doing it, yet it has happened. Limiting your wattage on something like a 5090 is helpful, but does not fully protect you.
 
This cable/connector was a misstep. They’ll switch to another design. Is it getting a lot of press of course, welcome to the age of sensationalizing everything. Duh. Like what do you expect? But it’s definitely a significant issue. We’ve all seen lots of different reports and most of them believable. Most of them can’t be attributed to user error- it’s plugging in a cable. Not rocket science, anyone is capable.
I would love to know actual numbers but nobody in the know is going to let that out and get in nVidia’s crosshairs. I’m most curious if we see these incidents continue or if they will trail off or ramp up over time.

I’ll tell you what though, all things being equal I’m probably not going to be inclined towards buying a used card with that connector if I can avoid it.
 
arestavo said:
Sources can't be cited because Nvidia reddit mods keep deleting melted connector posts. I can say that I've seen way more native 16pin cables melted now that they've been out for a few years. The adapters seemed more common in the beginning, but that quickly shifted once people started spouting "adapters are the problem, use native cables instead!"

It's obviously a shitty design that easily fails. We wouldn't have so many temperature, wattage, and current monitoring products for the connector if it weren't so shit.
Click to expand...
fascinating

many people are saying it’s my user error for using the adapter
 
erek said:
fascinating

many people are saying it’s my user error for using the adapter
Click to expand...

Which is probably an official product that came with either the PSU or the card? Or is it entirely 3rd party?
 
GoldenTiger said:
It may not be the ideal design, but it works for many millions of people just fine. The melted connections are just being sensationalized heavily by tech media looking got clicks. The actual percentage is lower for defects from what I've heard, than prior cards, including the tiny number of actual melted connectors.
Click to expand...
I doubt even a million people have 5090s
 
another burned out gpu

https://www.reddit.com/r/Corsair/co...m_name=androidcss&utm_source=share&utm_term=1

1789125959286.png

“Just got a smell of burning plastic from my system while gaming, quickly turned it off to investigate, and it appears that 4 of 6 12V pins for gpu connector have melted, with 2 of them being much worse than the other 2. Been running this exact setup untouched on my 5090 FE for over a year with no issues until today. Purchased a brand new Corsair premium sleeved 12V 2x6 cable when I got this card last year. Surprisingly the GPU side of the cable and connector on the card look brand new with no signs of any damage. Both of the psu side connectors are fused to the unit now as well so it’s hard to see the extent of the damage. What should I do here, I never thought this would happen to me.”
 
erek said:
another burned out gpu

https://www.reddit.com/r/Corsair/co...m_name=androidcss&utm_source=share&utm_term=1

View attachment 825001
“Just got a smell of burning plastic from my system while gaming, quickly turned it off to investigate, and it appears that 4 of 6 12V pins for gpu connector have melted, with 2 of them being much worse than the other 2. Been running this exact setup untouched on my 5090 FE for over a year with no issues until today. Purchased a brand new Corsair premium sleeved 12V 2x6 cable when I got this card last year. Surprisingly the GPU side of the cable and connector on the card look brand new with no signs of any damage. Both of the psu side connectors are fused to the unit now as well so it’s hard to see the extent of the damage. What should I do here, I never thought this would happen to me.”
Click to expand...

It would take Jensen himself coming out to admit to a problem for some people to believe it’s real. No amount of proof will ever be good enough.
 
https://www.techpowerup.com/352579/...ectors-ditching-nvidias-melting-16-pin-design

Modders at the Brazilian YouTube channel TecLab have an RTX 5090 running on three standard 8-pin PCIe connectors instead of the 16-pin design that keeps melting on high-end NVIDIA cards. The team stripped a Galax GeForce RTX 5090D HOF OC LAB XOC down to the bare PCB, removed both native 16-pin inputs, and soldered three 8-pin connectors straight to the power regulators. Sense pins were rewired so the card would still request full power, with liquid cooling keeping temperatures in check.
 
zehoo said:
It would take Jensen himself coming out to admit to a problem for some people to believe it’s real. No amount of proof will ever be good enough.
Click to expand...
The NVIDIA RTX 4090 / 12VHPWR class action was dismissed.


The case was Genova v. NVIDIA Corporation, No. 5:22-cv-07090, filed in the Northern District of California over the allegedly defective/melting 12VHPWR connector.





• Dismissed: March 10, 2023


• Type: Voluntary dismissal with prejudice


• Meaning: The plaintiff voluntarily ended the case, and because it was with prejudice, that particular lawsuit cannot simply be brought back by the same plaintiff on the sameclaims.


• Reason: No public reason was given.NVIDIA had filed motions seeking dismissal or arbitration, and the parties had reportedly engaged in settlement discussions before the plaintiff dropped the case.

https://www.classaction.org/media/plaintiff-v-nvidia-corporation.pdf
 
GoldenTiger said:
It may not be the ideal design, but it works for many millions of people just fine. The melted connections are just being sensationalized heavily by tech media looking got clicks. The actual percentage is lower for defects from what I've heard, than prior cards, including the tiny number of actual melted connectors.
Click to expand...
tunatime said:
I doubt even a million people have 5090s
Click to expand...
i doubt too about “millions”,

—-
best single-number estimate: ~900,000RTX 5090s sold worldwide.

I'd put a fairly wide confidence interval around that—±250,000 or so—because NVIDIA doesn't disclose 5090-specific shipments, and retailer data is incomplete.

One interesting sanity check: the 5090 is currently only 0.43% of GPUs in Steam's survey, despite being an unusually expensive card. Steam Store That makes a worldwide installed base in the low millions plausible, but 900k-ish sold is more consistent with the observed retail data than, say, 3–5 million.

If you want, I can go one step further and reconstruct the sales month-by-monthfrom January 2025 → September 2026, including an estimate of 5090 revenue ($) and NVIDIA's likely gross profit.

—-

also only heard about a few million additional over the next few years,

“NVIDIA to Deliver 2 Million Additional GPUs through 2028​

...and bring new co-engineered solutions to customers faster. As part of the expanded collaboration, the companies are working to: Deploy 2 million additional NVIDIA GPUs across AWS's global infrastructure in 2027-2028 Bring NVIDIA Vera CPU-based infrastructure to AWS Extend NVIDIA NVLink...
 
GoldenTiger said:
Millions have cards with these connectors. Don't be disingenuous and try a technicality "gotcha".
Click to expand...
Don't be disingenuous- the focus has been on these connectors being run too close to their rated limits with no protection. Obviously cards with much lower power limits are several orders of magnitudes less likely to run into these problems.
 
erek said:
another burned out gpu

https://www.reddit.com/r/Corsair/co...m_name=androidcss&utm_source=share&utm_term=1

View attachment 825001
“Just got a smell of burning plastic from my system while gaming, quickly turned it off to investigate, and it appears that 4 of 6 12V pins for gpu connector have melted, with 2 of them being much worse than the other 2. Been running this exact setup untouched on my 5090 FE for over a year with no issues until today. Purchased a brand new Corsair premium sleeved 12V 2x6 cable when I got this card last year. Surprisingly the GPU side of the cable and connector on the card look brand new with no signs of any damage. Both of the psu side connectors are fused to the unit now as well so it’s hard to see the extent of the damage. What should I do here, I never thought this would happen to me.”
Click to expand...
This is why I stopped using my Corsair HX1500i with a 5090. I don't think the connectors that plug into the PSU are safe enough though I didn't have any issues.
 
TaintedSquirrel said:
8-pin connectors have also melted over the years, but nobody ever talked about it because nobody cared.
People care now because it's a new connector and people are paying attention to it.

Are the rates worse than the previous connector? Who can say.
Click to expand...
Mathematically yes. See that's the beautiful thing about electrical engineering it's very easy to calculate the failure point and safety factor of wiring and connections. Compared to an 8 pin it has significantly less overhead to deal with problems in the power delivery...... Nvidia knew this and didn't care
GoldenTiger said:
Other cards have the exact same melting issue, so it can't be just that, obviously.
Click to expand...

Thank you for proving my point cards are melting. I would think most of the non 400w+ cards melting could have been prevented as I'm fairly confident the cable can handle 400 to 450 Watts loads all day no problem. When I first installed my 5090 I ran an experiment using my handheld Flir. The cable and connector stayed basically ambient temperature till about 400 to 450 Watts. By full load it was a good 10 to 15 Celsius higher than my case even with a fan on it.
 
Gideon said:
It will be Karma on here when some people experience this problem. Just amazes the lengths some will go to defend a poor design, but some people also defended the Pinto despite a terrible design flaw.
Click to expand...
This.

I don't have one of these graphics cards, but as an old electronics engineer it is clear these power connectors and the power scheme are insufficient. I can't believe there is a market for connector monitors for these things. I suspect the other part of this problem is the video cards not load balancing the power inputs. You cannot just parallel the connections, the connection with the least resistance will take most of the load until it fails - you have to actively draw certain amounts from each power feed.

This is clearly screwed up, but on the bright side, leather jacket man gets more leather jackets. Erek's experience is quite convincing.
 
TXE36 said:
...I suspect the other part of this problem is the video cards not load balancing the power inputs. You cannot just parallel the connections, the connection with the least resistance will take most of the load until it fails - you have to actively draw certain amounts from each power feed...
Click to expand...
Nvidia used to balance power draw with the 30 series and older. 40 and 50 series doesn't have that at all - just one power bus across all of the 12V pins.

I've linked it before, but Buildzoid goes over this and it may interest you:

View: https://www.youtube.com/watch?v=kb5YzMoVQyw
 
arestavo said:
Nvidia used to balance power draw with the 30 series and older. 40 and 50 series doesn't have that at all - just one power bus across all of the 12V pins.

I've linked it before, but Buildzoid goes over this and it may interest you:

View: https://www.youtube.com/watch?v=kb5YzMoVQyw
Click to expand...

I wish his video did more to explain. Why do we want to current balance and what does he mean by that? Force equal current through each pin? What is the shutdown mechanism when wires are disconnected in his example (why dont the other VRMs simply get driven harder)? Do VRMs have a low voltage cutoff? Why not just have a common +12V topology like the 5090 does, and have a lower safer low voltage cutoff?
 

"Cooler Master GPU Shield Adapter Officially Launches in the US​

by Cpt.Jank Today, 20:37 Discuss (10 Comments)
Cooler Master previously showed off the GPU Shield adapter cable at CES 2026, demonstrating the small 12v-2×6 adapter that is able to not only alert the user with a visual and auditory warning when there are anomalies in the current in the GPU power cable, but also actively intervene and reduce current when higher-than-normal current is detected. GPU Shield has officially launched in the US after showing up in China in early August 2026, and it is available on the Cooler Master online shop for $49.99.

In addition to monitoring high current, which is often the cause of the melted 12v-2×6 GPU connectors, GPU Shield also attempts to mitigate the meltdowns in the first place with dual-color connector tips that are an easy indicator to ensure proper cable insertion in the GPU and PSU. The adapter itself comes with 200 mm of total cable length, which should be enough for most DIY PC cases with conventional PSU placement, and it improves on Cooler Master's GPU Shield tech that is built into the PSU itself, since that is often not visible in many modern PC cases."
 
Holy crap what is next mobile alerts, with a PSU that will throttle the 12v-2×6 connection with thermal sensors?!?

This is going to be funny is a PSU company comes up with a solution before Nvidia will even admit there is a problem.
 
TXE36 said:
This.

I don't have one of these graphics cards, but as an old electronics engineer it is clear these power connectors and the power scheme are insufficient. I can't believe there is a market for connector monitors for these things. I suspect the other part of this problem is the video cards not load balancing the power inputs. You cannot just parallel the connections, the connection with the least resistance will take most of the load until it fails - you have to actively draw certain amounts from each power feed.

This is clearly screwed up, but on the bright side, leather jacket man gets more leather jackets. Erek's experience is quite convincing.
Click to expand...
Yep that is why I worry about the longevity of 5090s as each heat cycle the imbalance will get slightly worse.... changing the cables so often only helps till the tiny little pins on the GPU connector start wearing out.

I never worried about GPUs 5 years down the line before..........But with how crazy the hardware is right now I don't even know if we are getting a new generation of cards and if we do i hate to see what they go for.... honestly if I knew 5090s would end up going up to 5k+ and the whole computer industry feeling like it's imploding a year later I would have spent the extra on a card with two plugs.
 
serpretetsky said:
I wish his video did more to explain. Why do we want to current balance and what does he mean by that? Force equal current through each pin? What is the shutdown mechanism when wires are disconnected in his example (why dont the other VRMs simply get driven harder)? Do VRMs have a low voltage cutoff? Why not just have a common +12V topology like the 5090 does, and have a lower safer low voltage cutoff?
Click to expand...
To answer your questions:

Each pin has a maximum current they can handle. If there is any difference in resistance between the pins, the current carried by each pin will become unbalanced (some pins will carry more than others). In the case of the PCI-E connectors, each pin has so much headroom that they can easily handle an imbalance or even complete disconnect of one of the pins. In the case of 12v-2x6, you only need to lose half of one pin before you're already starting to overload the rest. Once you start overloading, it becomes a positive-feedback loop where the heat creates more imbalance and thus more heat until the connector melts. There is no way to detect that this is happening unless each pin's current is directly monitored.

The shutdown mechanism in the other examples is simply the GPU being unable to remain stable, or, in the case of multi-rail GPUs, a shutdown of the computer due to triggering the OCP on the connector. The VRMs all have to work together in unison to provide stable power delivery and when they don't, instability and crashing will occur. VRMs do not have a low voltage cutoff- they will simply demand more current until the supply voltage drops below the voltage needed for the GPU.

With regards to voltage cutoff, you will want a higher voltage rather than a lower one. Higher voltage = less current at the same power level. Current is what generates heat, not voltage. The problem becomes you'll need to monitor voltage before and after the connector in order to have an accurate assessment of how much heat is being generated in the connector. All PSUs vary in their voltage setpoints so a generic value won't work. This is also where having all pins tied together masks the problem- 6 pins dissipating 1 watt of heat each is very different from 2 pins dissipating 3 watts each.
 
Tsumi said:
To answer your questions:

Each pin has a maximum current they can handle. If there is any difference in resistance between the pins, the current carried by each pin will become unbalanced (some pins will carry more than others).
Click to expand...
if you have somewhat matched resistance on the pins, you will already have relatively matched current. If you have mismatched resistance on the pins, balancing current will cause a very large increase in heat on the pin with poor resistance. Again, why do we want this?
In other words: mismatched resistance is the actual problem, mismatched current is the side effect of this. Balancing current when you have mismatched resistance does not sound like a good idea
The shutdown mechanism in the other examples is simply the GPU being unable to remain stable,
Click to expand...
Is this a guess? Im not looking for guesses. It's conceivable that a system could still run with -1 VRM, if there are multiple VRMs powering a single rail and it was overdesigned.
VRMs do not have a low voltage cutoff- they will simply demand more current until the supply voltage drops below the voltage needed for the GPU.
Click to expand...
VRMs have undervoltage protection, im not sure what you mean.
They typically also have OCP.
https://www.techpowerup.com/review/asus-rog-maximus-xiii-hero/4.html
https://www.renesas.com/en/document/dst/isl6269-datasheet
With regards to voltage cutoff, you will want a higher voltage rather than a lower one.
Click to expand...
Yes, typing too quick and wasn't thinking.
 
serpretetsky said:
if you have somewhat matched resistance on the pins, you will already have relatively matched current. If you have mismatched resistance on the pins, balancing current will cause a very large increase in heat on the pin with poor resistance. Again, why do we want this?
In other words: mismatched resistance is the actual problem, mismatched current is the side effect of this. Balancing current when you have mismatched resistance does not sound like a good idea
Click to expand...
We want this because it immediately indicates a problem via a significant voltage drop after the pin that can be easily detected. It also limits the amount of current going through the pin, thus limiting how hot it can actually get. Tying all the pins together masks the problem until the connector melts.
serpretetsky said:
Is this a guess? Im not looking for guesses. It's conceivable that a system could still run with -1 VRM, if there are multiple VRMs powering a single rail and it was overdesigned.
Click to expand...
It's not a guess. Perhaps you need to spend some time reading some articles like this one:

https://buildzoid.blogspot.com/2015/04/what-makes-good-vrm.html

VRMs are not constantly on. They're constantly being cycled through to supply power and a weak link in any of them will cause instability.
serpretetsky said:
VRMs have undervoltage protection, im not sure what you mean.
They typically also have OCP.
https://www.techpowerup.com/review/asus-rog-maximus-xiii-hero/4.html
https://www.renesas.com/en/document/dst/isl6269-datasheet

Yes, typing too quick and wasn't thinking.
Click to expand...
You're looking at the spec sheet for the PWM controller, which has nothing to do with the VRMs themselves. Yes, the PWM controller does have OCP and undervoltage protections, but that's only for the controller itself. The VRMs don't really care, up until their maximum current output.

In that sense, you do have undervoltage protection in that the PWM controller won't work and thus won't command the VRMs to turn on if the voltage is too low. This is not an inherent feature of the VRMs.
 
serpretetsky said:
if you have somewhat matched resistance on the pins, you will already have relatively matched current. If you have mismatched resistance on the pins, balancing current will cause a very large increase in heat on the pin with poor resistance. Again, why do we want this?
In other words: mismatched resistance is the actual problem, mismatched current is the side effect of this. Balancing current when you have mismatched resistance does not sound like a good idea
Click to expand...
I ran a couple of simulations to illustrate:

With all connectors having a resistance of 0.01 ohms, about 0.69 watts is being generated in each pin at 600 watts with a voltage drop of 0.083 volts. Total heat being generated is about 4.14 watts. Let's use that as the baseline and run through some scenarios (numbers might not add exactly as I'm rounding here but pulling numbers straight from the simulator):

# of pins with increased resistance and valueSeparate pin wattage (Total heat generation)Combined pin wattage (Total heat generation)Separate pin voltage dropCombined pin voltage dropHighest current (combined pins)
6x 0.01 ohms6x 0.69 = 4.146x 0.69 = 4.140.0830.0838.33
1x 0.02 ohms1x 1.38 + 5x 0.69 = 4.831x 0.41 + 5x 0.83 = 4.550.170.0919.01
1x 0.04 ohms1x 2.78 + 5x 0.69 = 6.231x 0.23 + 5x 0.91 = 4.760.330.0959.5
2x 0.02 ohms2x 1.38 + 4x 0.69 = 5.522x 0.5 + 4x 1.0 = 5.000.170.10010
2x 0.04 ohms2x 2.78 + 4x 0.69 = 8.322x 0.31 + 4x 1.24 = 5.560.330.11111.11
3x 0.02 ohms3x 1.38 + 3x 0.69 = 6.213x 0.62 + 3x 1.24 = 5.560.170.11111.11
3x 0.04 ohms3x 2.78 + 3x 0.69 = 10.413x 0.44 + 3x 1.78 = 6.670.330.13313.33

So yes, you are absolutely correct in that combining the pins will have a significantly lower overall heat production than having the pins be separated. However, the important part to look at is the voltage drop and the highest current value. Even with 3 marginal pins at 4 times the resistance, the voltage drop doesn't even come near the voltage drop of a single separated pin having double the resistance. Yet we can see that we're at the 9.5 amp rating of the pins with one pin at 4x resistance and 2 pins at 2x resistance are exceeding the amp rating. This is the problem with combining pins- it masks problems until the damage is done.

With separate pins, the GPU can monitor the voltage of each pin and if there's a significant difference between them (0.1 volts is enough to detect a problem as shown by the table), it can alert the user to a potential problem. Combined pins would require a voltage detector before and after the connector- not feasible, and the voltage differences aren't that high so more sensitive detectors might be needed. With separated pins, the monitoring is self-contained on the card.

Of course, the best solution would be to reduce the number of pins in use. You can increase the size of the connector such that all the current can be handled by one pin, or you can increase the voltage so the current is lower and thus run fewer pins that aren't so close to the limit. These have their own issues- a single thick cable isn't very bendable, while higher voltage has the whole changing of standards to contend with.

arestavo said:
Kinda like this one? https://rog.asus.com/us/power-supply-units/rog-equalizer/rog-equalizer/

Although I'm no Asus fan, so I've not really looked into it.
Click to expand...
This cable is completely useless for the issue at hand. The problem exists in the connectors, not the wires. In fact, you can say this cable would actually make things worse- by having a giant block bridging all the wires right next to the connector, any differences in connector resistance will cause more current imbalance. The wires, being the same length, should have identical resistance. Since resistance is additive, this adds to the resistance at the connectors. With not-bridged wires, you might have resistances of 0.012 and 0.014 from connector to connector, for example. With bridged wires, you might have a resistance from connector to bridge of 0.06 and 0.08 instead. The difference is the same, but the percentage difference and resulting current imbalance is significantly higher for the Asus bridged wires.
 
You must log in or register to reply here.
Back
Top