Daz 3D is part of 
Connect
DAZ Productions, Inc.
7533 S Center View Ct #4664
West Jordan, UT 84084
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2025 Daz Productions Inc. All Rights Reserved.
Comments
I think if I was connecting a cable to a power supply, and intending to pul up to 40 Amps through it I would use a spade terminals well screwed down, but hey I'm sure Nvidia know what they,re doing ..
The whole cable heat thing most likely comes from underpowering. Same principle in audio, better too much power than too little, because too little power will cause arcs in the coils and you'll burn the speaker rather than blow it. CPUs sometimes faced that some time back, the shreiking from the underpower could be audibly heard, and it heats up big time. Older electronics had fuses, which melt when that heat starts to build up. But your PC power sulpply has no such thing. It will let that heat go and potentially blow up everything it's connected to. From what I have read, the 5090 is a thirsty boy compared to, for example, the 4060ti I have in my board currently. Always when upgrading something like this, look at the power-consumption specs and buy a power supply beefier than you think you'll need - AND use the cables that came with it. That's so often overlooked when people are building their own system.
...⬆ this ⬆
Looks like the 4090 and 5090 power delivery design is inadequate. I am skipping it. My 3090 has three 8-pin power connectors which is a lot better. The electrical load is distributed over 24 wires vs 12+4 for the 4090 and 5090. Not sure why Nvidia did this, when you consider how much more power the newer cards draw.
Here's a video that explains the limitations of the 5090 interface:
How Nvidia made the 12VHPWR connector even worse
Per ChatGPT:
Yes, three 8-pin cables would likely be a safer and more reliable choice than Nvidia’s 12VHPWR (12+4 pin) connector, which has been experiencing melting and failure issues, particularly with 4090 and potentially 5090 GPUs. Here’s why:
Why the 12VHPWR Connector is Failing
High Current Density:
Poor Cable Seating:
Thin Wires in Some Adapters:
Why Three 8-Pin Cables Would Be Better
Lower Current Per Cable:
Redundancy & Safety:
Proven Reliability:
Conclusion
Nvidia's 12VHPWR was introduced for cable management and future power scaling, but its real-world reliability issues suggest that multiple 8-pins would have been the better and safer choice. Many PSU manufacturers are now making native 12VHPWR cables to mitigate issues, but for users concerned about safety, sticking with 3-4 separate 8-pin cables is still the safest approach.
As I mentioned above, my RTX 3090 has three 8-pin connectors. Even though the specified power is 150W, the rated power is 252W. That's a safety factor of 1.68. The rated power of three 8-pin connectors is 756W, higher than the 684W rated power of the 12VHPWR (H+) / 12V-2x6 (H++) connectors. The load is also distributed over three cables which allows for less heat buildup and redundancy in case there is a loose connection.
Per Wikipedia:
https://en.wikipedia.org/wiki/16-pin_12VHPWR_connector
Don't forget that ChatGPT is pretty much an average of what others are saying - it is no more reliable than its sources.
Correct. It is a convenient method to summarize a topic though.
Regardless, the 5090 power delivery design is completely flawed, as pointed out by numerous electrical engineers, and the fact that cables are melting similar to the 4090 incidents.
https://en.wikipedia.org/wiki/16-pin_12VHPWR_connector
Yes, good points. A good quality PSU is a must. And good cables. "jayztwocents" looked at a few 12VHPWR cables and some have pins that are recessed too far in. This could explain why cables are melting even if the connector is plugged in all the way, but a connection is not made. Regardless, Nvidia's design should have accounted for this. Perhaps they should have included two 12VHPWR connectors on the 5090 card. Four 8-pin connectors would have a rated power of 1,008 watt which would have been perfect, but reduced compatibility with some PSUs? And all those cables would create messy cable management. I'd take that though over melted wires.
The EVGA Geforce RTX 3090 TI included two 16-pin power connectors:
https://www.evga.com/articles/01571/3090ti-kingpin-hybrid/
https://wccftech.com/nvidia-four-16-pin-connectors-gpu-prototypes-rtx-5090-rtx-4090-up-to-2400w-power/
NVIDIA Had Up To Four 16-Pin Connectors on Its GPU Prototypes For RTX 5090 & RTX 4090, Up To 2400W Power Delivery
Are these all Nvidia cables, or are the 3rd party cables (or are both melting)? Some of the pictures don't look like the ones I've seen of the Nvidia cables from the 5090 FE advertisements. Anyway, doesn't the PCIe slot also supply 75 W to the card, so if the card is running 600 W it is actually drawing 525 W across the cable? I also note that several of the newer 870E motherboards (not sure about Intel boards) are also adding an additional 8 pin power connector to the bottom of the board which is supposed to be able to supply up to 150 W additional power across a PCIe 5.0 slot (probably to address the power concerns of the 5090 cards). I have a new board supposed to arrive soon which has one of these so I am interested in what it actually does. Of course I only have a 3090 right now, so 350 W power draw is a much less of a concern than 600+...
The issue is the compact design of the 12VHPWR standard. When you bundle a bunch of power wires next to each other the wires heat each other up under load. If the load is small it does not matter as much. At higher electrical loads it becomes an issue. The NEC (National Electric Code) for example specify a maximum number of conductors in 1/2", 3/4", 1" etc. conduit. There is a limit to how many conductor wires you can run in conduit. More conductors can be added as the size of the conduit goes up because the larger diameter allows for the dissipation of heat.
The 5090 cables are not in conduit obviously, but the same principle applies.
With my EVGA 3090 the power is spread over 24 wires via three 8-pin connectors, 8 wires per connector. 8-pin connectors have a specified power of 150W per PCI SIG - see the link below.
https://en.wikipedia.org/wiki/16-pin_12VHPWR_connector
This is why Nvidia could not use three 8-pin connectors for the 5090, because per the PCI SIG spec, the total specified power would only be 450W, less than the power draw of the 5090. But the 8-pin connectors actually have a rated power of 252W, so three 8-pin connectors can support 756W. The rated power of the single 12VHPWR connector in the 5090 is 684W but the specified power is 600W, which is higher than the specified power of three 8-pin connectors. So the 5090 would have been better off with three 8-pin connectors instead of the single 16-pin connector but could not be used as mentioned.
The EVGA 3090 TI Kingpin included two 12VHPWR connectors and a VRM (Voltage Regulator Module). That would have been perfect for the 5090 cards.
https://www.evga.com/articles/01571/3090ti-kingpin-hybrid/
In de8auer's video that I posted his 5090 power wires are spaced apart more which helps dissipate heat. He used a thermal sensor that recorded 150 degrees celsius on some of the wires near the PSU connector. He also checks the amps running through each wire with a tester. That would answer your question of how many watts is drawn across the cables, besides the power draw of the PCIe slot that you mentioned. The results he recorded were after a few minutes of high load on his 5090 FE. Now imagine rendering an animation with many frames for a few hours. I should add though that the power draw on the GPU when rendering an animation in Daz 3D is not constant. I've been monitoring my 3090's power usage and it drops down significantly between frames in preparation for whatever Daz needs to do for the next frame. I also notice that my CPU usage would go up to 100% just before handing off to the GPU for rendering the next frame. The 3090 will then use anywhere from 260-300 watts of power with no limit on the power in Afterburner. I have not checked the power usage with very complex scenes.
The only real solution IMO is to revise the power delivery design of the 5090. Or limit the power via MSI Afterburner - but then you might as well not get a 5090.
Here is de8auer's video again. There are many other videos that explain the melting and heat issue with the 5090s.
Kingpin is apparently now working with PNY because EVGA exited the GPU business after the 3090 TI. There are many rumors on why they quit. It's too bad, I really like their products. I have about 30 EVGA GPUs and they have been reliable in my experience. I also have many EVGA PSUs which worked fine under heavy load. The build quality was really good. It will be interesting to see if the PNY Kingpin 5090 TI is ever released, and if they will modify the power delivery.
When I look at my 3090 FE in MSI Afterburner while rendering in Daz I see up to 350 W power on the GPU (this is rendering scenes with up to 16-17 G8 figures and a scene background/props with various lighting and HDRI). I posted in another thread that I tested limiting the 3090 power using Afterburner to around 70% (which in test resulted in a displayed power draw of around 260 W) and the effect on render times was very small (about 5% slower). Of course this is currently a moot point with regards to the 5090 since Daz does not support it for rendering, but if it becomes supported my thought would be to limit the power draw to around those seen with the 4090 (400-450 W) and I expect I would still see a significant improvement in render times over my 3090 without having to worry about excessive heating and current draw on the cable. I agree that 3 8-pin connectors probably gives a lot more safety margin for overcurrent than the current 12 pin connector (which I believe connects to 2 8-pin connectors from the PSU), but I suspect NVidia didn't go that route because not everyone has modular PSU units with 3-4 available VGA 8-pin outputs on it. I am still curious to see some testing by someone of the additional 8 pin connector on the X870E motherboards to see if it helps pull some of the load away from the GPU cable (my board arrived today but I haven't installed it yet).
And a LOT of PC builders and reviewers on Youtube. A lot of them are advising to stay away from them until the issue gets fixed.
I like how simply put this is. Most of my explanations of AI are much longer and involve statistics and probability.
Also, AI is prone to making up stuff if it cannot find the answer...
Anyone tried with the 5090 / 5080 to see if it's working?
It has already been announced that it will not work in any DS 4 version..
Hi, I'm late to the party - I'd not have known about this if I didn't see a Reddit post about it. Has Daz said anything official about when we can expect DS 5? I'm meant to be ordering my new PC this weekend with a 5080 (a loooong overdue upgrade from my 2070!!), so like everyone else I need to know if I've got a paperweight for the next 12 months (and how long to tell my customers I can't offer Daz-based renders for...) :D
As of July 2021 it was almost done. So I presume it's coming any day now.
For a less facetious but unfortunately no more helpful answer:
No, they have not, and the closest thing to any official statement about it in a long time is that there will not be an official ETA: https://www.daz3d.com/forums/discussion/comment/9071906/#Comment_9071906
My own best guess from what I've heard and my general gut instincts... I'd guess the likely timeline for the first public beta is sometime in Q2 2025. But that is a wild-as-heck guess, based on splitting the difference between "Unless it was basically already ready, they're going to have to tie a lot of loose ends" and "They can't leave it any longer than that without it being a serious problem".
and we found a bot
maybe
Ah, darn. Cheers Matt - this is disheartening. Ah well, I was already coming to terms with the fact Daz would be off the menu for a bit, so it's a bit whatever! Fingers crossed it comes sooner rather than later.
Back before I retired I was in the software support field. I'm a bit more pessimistic.
IF 5.0 becomes a true priority I expect the early outside test group will see a preliminary version in the May/June timeframe. The public beta, which might still have some functionality missing, to be out around September and a possible full release in the December timeframe. It all depends on how far along version 5.0 is at this point.
Oh I don't know...
I like, let's keep 'em
Now is a really good time to invest the time to learn how to get DAZ content into Blender and not have to deal with any of this kind of lack of transparency, especially when there are important, high value purchasing decisions to be made. Blender just published their roadmap for 2025, in which they are completely transparent about the coming projects and their schedules. Blender 5, you ask? November.
...yeah received an email on it this morning.