rtx5090 Rtx 5090

[2000joe]

Hello everyone,

 

I have an RTX 3090 GPU and was hoping to upgrade to an RTX 5090 for 4K gaming. However, I was about to order one (stock is slowly appearing), and I came across an article about the melting connector. After researching online, I'm quite concerned. I upgraded my power supply in February to a Corsair HX1200 ATX 3.1 in anticipation of the upgrade (the card isn't cheap; some saving time was needed). I'd like to know everyone's opinions on the issue. Is it likely I might be affected, or has this been blown out of proportion, with factors like third-party cables not being considered? I would be using either the Corsair connector or the 4x 8-pin adapter included with the graphics card. I was looking at the Zotac 5090 Solid as it seems more reasonably priced and is in stock. I'm seeking something that can play games comfortably and hassle-free at 4K with the highest possible settings. 

[Millios]

Its a problem with the connector itself and how much "juice" it draws

If you wanna calc the heat it goes Q= I^2 R Δt or the heat is equal to the square of the amps times resistance times time being on

 

To make it easier we will use P=I^2R so its Q=PΔt so Q or heat is 600 x However long you got the card running

And the problem is cause Nvidia instead of using seperate connectors uses its own stupid singular one increasing the heat it makes so it melts and or burns

 

To not overcomplicate it the TL DR is that the card draws too much power over 1 connector so it gets the cable too hot

[jaslion]

54 minutes ago, 2000joe said:

likely I might be affected, or has this been blown out of proportion, with factors like third-party cables not being considered?

Yes the issue is VERY real, it is most definitely not blown out of proportions and no cables are truly safe.

 

Pretty much comes down to a connector that can technically handle the 600w the 5090 asks is being used improperly to the point where it doesn't know if all the pins are properly connected or not so it will just send the same amount of power over less wires which then causes melting, damage and potential burning/fires. This also happens on the psu side because well no safety features for this are built in.

 

Even a properly connected cable can have this happen the reference 5090 is one of the worst ones as it bundles ALL connections into 3 groups and as long as a single pin can send power to the group it will keep working. It's how most 5090's are made. there are one or 2 with extra safety features but none solve the real issue which is too much power over too small cables being allowed because there is NO safety check.

 

57 minutes ago, 2000joe said:

and hassle-free at 4K with the highest possible settings. 

Even with a 5090 that won't happen. High refresh high settings native 4k isn't possible across the board. You need upsacling, frame gen,... and whatnot all which impact image clarity and often causes visual bugs that are indeed noticable during gameplay.

 

 

The only thing for you to do if you are DEAD set on a 5090 is to REALLY make sure the cable is 1000% plugged in, it has the LIGHTEST bend as possible and to NOT MOVE the system without ALWAYS checking the cable over and over. This is for both the gpu AND PSU SIDE!!!

[WereCat]

The issue is real and it does not matter what connector you use as it's a bad design with current balancing on the graphics cards side. How likely are you to actually have this issue manifest? The chances are quite low but the concerns will always haunt you. I personally would not trust to leave my system running in load when leaving home when knowing about these issues.

[2000joe]

I see i thought that might be the reality just wanted to check thanks for all your comments based on what you've all said do you think nvidia will try and come up with some "fix" or do you think they will just wait for the next generation?

[WereCat]

2 minutes ago, 2000joe said:

I see i thought that might be the reality just wanted to check thanks for all your comments based on what you've all said do you think nvidia will try and come up with some "fix" or do you think they will just wait for the next generation?

considering they made it worse than 4000 series I wouldnt bet on it

[podkall]

2 hours ago, 2000joe said:

I'd like to know everyone's opinions on the issue.

Mostly resolved and if it melts they should replace/refund you everything, because it is factory defect.

 

Especially if you use high-end proper PSU that is ATX 3.1.

 

 

 

The resolved part is mostly, properly connecting the GPU, and getting branded GPUs.

 

Because the melting was more often happening on Founder's reference cards, that have angled connector:

 

 

 

 

Meanwhile most if not all brand 5000 RTX cards have normal straight pointed connector:

 

 

 

1 hour ago, jaslion said:

Even a properly connected cable can have this happen the reference 5090 is one of the worst ones as it bundles ALL connections into 3 groups and as long as a single pin can send power to the group it will keep working. It's how most 5090's are made. there are one or 2 with extra safety features but none solve the real issue which is too much power over too small cables being allowed because there is NO safety check.

Isn't this only 50% true? If cable sends too much power through one pin, it's not properly connected cable.

 

The other 50% is the calbe/pins themself that are improperly connected even if the user makes sure they are properly connected due to some kind of defect? (which sounds like something that can be quite rare, but not rare enough to keep all high-end GPU users safe?)

[xg32]

6 hours ago, 2000joe said:

.

gonna try and keep this short. I'm not an engineer but i don't think the card is designed for 600w, the connector is likely not going away even with next gen. these are options i'm considering.

 

undervolt to 0.9v 400-450w and "lose" 3-5% performance, so "only" 20% faster than a 4090, and/or buy a temp gun 125-150usd (gonna count it as part of the 5090 and cheaper than buying an astral), and check the temps of the psu side, gpu side, and individual wires under full load. Once the gpu/plug is in place and temps are checked, it'd alot less likely to melt, i still wouldn't run the card past 450w.

 

The connector isn't the only issue, the pcb, memory etc all run warmer than i'm comfortable with.