RTX 5090 cards may have problem with uneven power distribution on 12VHWPR cables Roman “Der8auer” Hartung quickly reached out to the person who revealed the first melted RTX 5090 Founders Edition card this weekend. As it turns out, this person lives close to Roman, allowing them to have a long conversation about what happened. The […]
Article is a summary of der8auer’s video. Measurements show 22A/260 watts – nearly half of the card’s power draw – going through a single wire heating it up to 150 °C in an open-air test bench.
I don’t understand how the ATX standard hasn’t adopted higher voltage rails for GPUs. Higher volts mean fewer amps which means less heat. 44A through an 8 pin connector is going to make a lot of heat. 11A at 48v would be much better.
The cards aren’t affected* it’s the custom cables people are using.
I wonder why he didn’t do a temp test without the extensions and just Nvidias adapter? Those parts aren’t hot and haven’t caused any issues from any reports so far. It’s ALWAYS had a custom extra cable. Increasing the length of cables increases power draw, it’s the basics of electricity. So yeah… increaseing cable length, and adding extra connections will make it waste more power and draw more…… it’s not surprising it’s only happening with systems with the extra cable length and connections that cause loss.
Also, the headline is misleading, he’s blaming the cards, when the only piece that’s ever melted is custom non approved parts.
There wasn’t even anything wrong with the minifit jr 8 pin pcie connectors, the official spec of 150W massively derated. If Nvidia really wanted to cap the number of plugs and not change any specs then the 8 pin EPS connectors are rated at 288W.
If they did that then they likely would end up with melted 8 pin connectors as that derating is a protection against one pin being overused due to small expected differences in resistance on each pin
Using custom cables even in that scenario would still lead to cables melting.
It’s from increasing the cable length and therefore the required power draw to draw the required amount to power the device. It already happens with 8pin systems as well, this isn’t unique to the delivery system either.
Using too many extension cords has the exact same affect… it’s like blaming the heater because the cords melted. But no, let’s not blame physics, let’s blame the GPU(heater) -.-
Now you’re really showing how uniformed you are. It’s not an extension cable, it’s 20cm in length total which is significantly shorter than what comes with standard 12Vhpwr native PSUs at 60-70cm.
It’s mostly just one loud person who very apparently has not actually interacted with the case presented here. Either through ignorance or to purposely deflect the blame from Nvidia with the evidence weighing against them.
You are not logged in. However you can subscribe from another Fediverse account, for example Lemmy or Mastodon. To do this, paste the following into the search field of your instance: [email protected]
No game suggestions, friend requests, surveys, or begging.
No Let’s Plays, streams, highlight reels/montages, random videos or shorts.
No off-topic posts/comments, within reason.
Use the original source, no clickbait titles, no duplicates.
(Submissions should be from the original source if possible, unless from paywalled or non-english sources.
If the title is clickbait or lacks context you may lightly edit the title.)
Anyone wanna take bets on how long it takes before there’s a lawsuite?
I bet a few months.
It’s just a bad design and a bad standard.
I don’t understand how the ATX standard hasn’t adopted higher voltage rails for GPUs. Higher volts mean fewer amps which means less heat. 44A through an 8 pin connector is going to make a lot of heat. 11A at 48v would be much better.
Apple did something similar, then they put 48V pin next to 3V pin 🤦
That’s…sooooooo Apple…FFS 🤦
one guy had a bad device
Isn’t that like a third of available stock then?
If so, who cares?
Math cares? With a smaller sample size, even a handful of issues makes the percentage of problems fairly high.
5 bad apples in a dozen is a problem 5 bad apples in a truckload isnt.
There are already two 5090 FE affected by this just in the video.
It’s a result set of two. Please use your brain. You’re wasting it.
The cards aren’t affected* it’s the custom cables people are using.
I wonder why he didn’t do a temp test without the extensions and just Nvidias adapter? Those parts aren’t hot and haven’t caused any issues from any reports so far. It’s ALWAYS had a custom extra cable. Increasing the length of cables increases power draw, it’s the basics of electricity. So yeah… increaseing cable length, and adding extra connections will make it waste more power and draw more…… it’s not surprising it’s only happening with systems with the extra cable length and connections that cause loss.
Also, the headline is misleading, he’s blaming the cards, when the only piece that’s ever melted is custom non approved parts.
He isn’t using a custom cable though. He is using the Corsiar cable that came with his PSU.
Time for a large busbar connecting the power supply, motherboard, and graphics card?
A 48V rail would make more sense. Just like USB-C did to get 240W power delivery over a small cable.
There wasn’t even anything wrong with the minifit jr 8 pin pcie connectors, the official spec of 150W massively derated. If Nvidia really wanted to cap the number of plugs and not change any specs then the 8 pin EPS connectors are rated at 288W.
If they did that then they likely would end up with melted 8 pin connectors as that derating is a protection against one pin being overused due to small expected differences in resistance on each pin
I’m not saying to ignore the derating, I’m saying use EPS connectors at their derated 288W.
Using custom cables even in that scenario would still lead to cables melting.
It’s from increasing the cable length and therefore the required power draw to draw the required amount to power the device. It already happens with 8pin systems as well, this isn’t unique to the delivery system either.
Using too many extension cords has the exact same affect… it’s like blaming the heater because the cords melted. But no, let’s not blame physics, let’s blame the GPU(heater) -.-
Now you’re really showing how uniformed you are. It’s not an extension cable, it’s 20cm in length total which is significantly shorter than what comes with standard 12Vhpwr native PSUs at 60-70cm.
Did you watch the video?
Sure they did…had their eyes crossed, fingers in ears yelling:
This comment section looks exactly like the one he called out in the video.
It’s mostly just one loud person who very apparently has not actually interacted with the case presented here. Either through ignorance or to purposely deflect the blame from Nvidia with the evidence weighing against them.