• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

NVIDIA RTX 50 SERIES - Technical/General Discussion

They need to fix the fact the card is pulling so much more power down just one of the cables instead of evenly distributing the power.
Maybe it can be fixed in a bios update? Unless that power is physically going somewhere it is needed instead of being controlled by programming...
Maybe it's an ASUS PSU issue. Others are reporting even distribution with a Seasonic PSU and the 5090 FE:

Edit: Debauer was using a Corsair PSU not ASUS. Sorry for the confusion.
 
Last edited:
Seeing a multi trillion $ company fail at something that should be elementary will always be fascinating to me :cry:
If everyone keeps buying, did they fail?

Presumably some (if not all) AIBs must be aware of this being a problem - they're all smart companies with a bunch of electrical engineers working on PCB design. One would assume the reason they didn't just fix this on the 4000 series and 5000 series and make a killing on selling the "card with the connector that never melts" is that something about the architecture design itself from nvidia prohibits that from happening. Just speculation on my part though.
It's not technically a problem though is it? Assuming everything is working correctly you shouldn't run into a problem but there doesn't look to be much safety margin left whatsoever. I'm no electrical engineeer, I barely know the difference between a Capacitor and a MOSFET, so how many other areas are these kind of compromises being made? Most companies over the decades push for lowering costs and standards, push it till it breaks (or melts).

I started getting black screens recently, eventually figured out the weight of the card was either pulling it out of the socket or warping the board and sorted it by fitting a brace, it's not even an especially heavy card so I didn't think it'd need a support but it's the same kind of problem really, cards weren't meant to be this heavy. Does make me wonder how quickly a 5090 would fail without support.
 
This issue is getting ridiculous. Thankfully I've had no issues with my Zotac 4090 extreme airo that I've had since launch. The card has been powered up 24/7 with a few thousand hours of gaming, connected to an AX1600i PSU using a 3x8 pin to 16 pin cable, using a cablemod extension type adapter cable. Still though, this connector has to go and something far better needs to be put in its place as it's not fit for purpose.
Using the same cable with my 4090 also and had zero issues, i wonder if the 4090 cable heats up quickly like that video above ?
 
I'm puzzled why this is only just rearing its head weeks after all the big reviewers published their reviews for the 5090/5080, especially considering how the melting cable/connector thing was present on the 4090. Not one reviewer thought to check the temps on the cables for something that pulls nearly 600w? It seems like it'd be a pretty easy thing to replicate, just put the thing under heavy load.
 
Is it bad to admit that I am finding these connector videos far more interesting than the actual 50 series reviews? Oops.
It really is interesting! Pretty shocking though they actually designed the new cards that way, like zero failsafe for worst case where worst case now results in fire lol

I'm puzzled why this is only just rearing its head weeks after all the big reviewers published their reviews for the 5090/5080, especially considering how the melting cable/connector thing was present on the 4090. Not one reviewer thought to check the temps on the cables for something that pulls nearly 600w? It seems like it'd be a pretty easy thing to replicate, just put the thing under heavy load.
Watch the buildzoid vid explaining the wiring, because it's a very niche scenario where all but one wire needs to fail where 600w is now being drawn through a single wire that is nowhere near rated for that power draw, it's not likely going to be a common thing unless there's an issue with the wires as well causing all but one to fail so it's going to be an ucommon failure but due to the design, it could potentially happen to 100% of cards if the condition of 5 out of 6 wires failing happens

Basically if your wires are fine, your card will be fine
If all but 1 wire fails, you're going to smell burning plastic and not be very happy
 
Last edited:
I do recall reading (here) that the pins on the PSU side are always ‘non-standard’ in respect of their layout, hence PSU manufacturers saying never use cables meant for another brand.

Might be something in that…?
 
Maybe it's an ASUS PSU issue.
Not AFAIK. Appliances draw power so if something wants to use 575W & 12v DC (RTX 5090) whatever is supplying the power will supply it through the path of least resistances or until it trips one of the supplies protections meant to protect itself.

Like i said earlier it's a design problem, going on Buildzoid video it's something they worked around with previous cards by keeping the 12v lives separate and staying within specs on each of those cables (iirc 12-14ga). Now they're running them in series you could cut all but one of the lives and the card/PSU would have no way of knowing so it would try to draw 12v DC / 575W (50 amps) though a cable designed to only handle 15 amps.
 
Not AFAIK. Appliances draw power so if something wants to use 575W & 12v DC (RTX 5090) whatever is supplying the power will supply it through the path of least resistances or until it trips one of the supplies protections meant to protect itself.

Like i said earlier it's a design problem, going on Buildzoid video it's something they worked around with previous cards by keeping the 12v lives separate and staying within specs on each of those cables (iirc 12-14ga). Now they're running them in series you could cut all but one of the lives and the card/PSU would have no way of knowing so it would try to draw 12v DC / 575W (50 amps) though a cable designed to only handle 15 amps.

So in some PSU's something could be causing more resistance on some pins more than others, leading to less resistance & more power in just one or two of the wires..?
 
As far as I can tell it’s really not that bad on the price / performance front:

msTmf3B.jpeg
Changes up a bit when you use real world AIB pricing.
Screenshot-988.png
 
So, what can be done to mitigate this issue by those of us with RTX 4090 cards?
You can't, well you could but it would require you putting sensors on each wire to see how much power they are drawing, this problem requires specific conditions of multiple power wires failing either by not being properly connected into the GPU/PSU or being physically broken which ends up in the full load being drawn through a single wire
 
Last edited:
You can't, well you could but it would require you putting sensors on each wire to see how much power they are drawing, this problem requires specific conditions of multiple power wires failing either by not being properly connected into the GPU/PSU or being physically broken which ends up in the full load being drawn through a single wire
You can. just apply a power limit on msi afterburn
 
So in some PSU's something could be causing more resistance on some pins more than others, leading to less resistance & more power in just one or two of the wires..?
AFAIK yes, but like i said I'm not an electronics engineer so if there's any out there I'd more than welcome their input.
 
Back
Top Bottom