• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Can 12VHPWR issues be fixed?

Do you think dual connectors will solve the 12VHPWR connector problems with high power GPU's?

  • Yes - That would be fine

    Votes: 4 6.6%
  • Yes - When in doubt just double everything to solve problems

    Votes: 1 1.6%
  • No - Just give me more 8-pin connections

    Votes: 23 37.7%
  • No - That's just double trouble

    Votes: 8 13.1%
  • Don't care, I'm done with high power GPU's (less is more)

    Votes: 6 9.8%
  • Don't care, I'm done with Nvidia

    Votes: 4 6.6%
  • What is electricity?

    Votes: 7 11.5%
  • Other (leave a comment)

    Votes: 8 13.1%

  • Total voters
    61
  • Poll closed .
Every time you add a component into the mix you are adding resistance, right? So maybe it’s not a good idea to use cable extensions and 3rd party adapters?
 
I still dont understand why the 4070 has 8-pin but the 4070 Super has 12VHPWR when there are holes for 2x8pin on the PCB on Gainward/Palit cards.
 
I still dont understand why the 4070 has 8-pin but the 4070 Super has 12VHPWR when there are holes for 2x8pin on the PCB on Gainward/Palit cards.
I think both 4070/4070S FE cards have the 12VHPWR
 
I still dont understand why the 4070 has 8-pin but the 4070 Super has 12VHPWR when there are holes for 2x8pin on the PCB on Gainward/Palit cards.
Nvidia have mandated it to try and force 12VHPWR connector on everyone.
Vote with your wallets, people - it's the only language they understand.
 
Nvidia have mandated it to try and force 12VHPWR connector on everyone.
Vote with your wallets, people - it's the only language they understand.

It is indeed REALLY lame...

And fact is that the connector is just alot less safe than 8pin pcie...
 

Maybe a bit late, but I just read this. Interesting update on the CableMod situation. About 1% failure rate. Storm in a teacup I reckon.

According to the CPSC notice, CableMod is recalling 25,300 adapters, which were sold between February, 2023, and December, 2023. Of those, at least 272 adapters failed, as per reports and repair claims made to CableMod. That puts the failure rate for the angled adapters at 1.07% – if not a bit higher due to the underreporting that can happen with self-reported statistics. All told, the manufacturer has received at least $74,500 in property damage claims in the United States, accounting for the failed adapters themselves, as well as the video card and anything else damaged in the process.
 
I still dont understand why the 4070 has 8-pin but the 4070 Super has 12VHPWR when there are holes for 2x8pin on the PCB on Gainward/Palit cards.
Not dual 8-pins, if I'm not mistaken one is all ground including the 4 signaling pins other is 6x12v and 2 ground. a 16 pin internally.
 
I like only having to plug in one connection, so I guess I am the only one who likes it :D
 
Me myself I don’t think I would buy another card with this type of connector. I have not had an issue with mine, but I live in fear. I run mine with a custom sleeved cable from the PSU to the card, but yeah, just give me some 8 pin connectors, I’m okay with that.

I like only having to plug in one connection, so I guess I am the only one who likes it :D
I do like that aspect of it, cleaner,
 
The future is MicroFit. Not only for GPU but mobo and CPU also They will refine it after some growing pains. Yes plugging only in one connection is leaner whatever that means. Lol
 
Maybe a bit late, but I just read this. Interesting update on the CableMod situation. About 1% failure rate. Storm in a teacup I reckon.

The problem is the POTENTIAL consequences of those fails: you can LITERALLY put the house on fire.


The 8-pins are, worst-case, rated to 288W based on Molex MiniFit Jr specifications. The fact PCIe CEM standard limits them to 150W is safety margin to account for damaged connectors and cheap Chinese manufacture.

That's a 92% over-protection.

From my understanding, that 600W cable has a 660W worst case, rate: that's a mere 10% over-protection.

I've seen this being explained in a video but don't recall the name ... so i can't search for it :(
 
Last edited:
I do like that aspect of it, cleaner,
Yeah cleaner and then they put the connector in the bloody center top of the card, sticking out too far for its width to fit in most cases proper. So clean!

It all screams lazy and clumsy design to me
 
Yeah cleaner and then they put the connector in the bloody center top of the card, sticking out too far for its width to fit in most cases proper. So clean!
They put it at the edge of the PCB dude, the rest is cooler. Fits my case just fine. But then again, I do not have a monster cooler taking extra space.
It all screams lazy and clumsy design to me
Maybe, but I do not have a high powered GPU compared to some others. So I guess it works ok? Regardless.. my insurance is paid.
 
They put it at the edge of the PCB dude, the rest is cooler. Fits my case just fine. But then again, I do not have a monster cooler taking extra space.

Maybe, but I do not have a high powered GPU compared to some others. So I guess it works ok? Regardless.. my insurance is paid.
The 4090. Its the only card actually 'needing' it, if you discard the idea of adding another 8 pin. The rest of them is just pure cost reduction so they can move to one connector
 
The rest of them is just pure cost reduction so they can move to one connector
There is literally no room on the tiny PCB.
 
I like only having to plug in one connection, so I guess I am the only one who likes it :D
I like having only one cable to route, I'm not really fussed about what's on the end of that cable as long as it works.

Having to run two modular cables between a PSU and GPU is a bit of a faff, and in that respect, 12VHPWR got something right. It's just mind-boggling to me that they decided to shrink the pins and connector as they were doubling the current draw. Some safety margins would have been given up even if they hadn't shrunk the connector and just made a 12-pin(+4) MiniFit Jr connector, but that probably would have been okay because there's so much safety margin built into the PCIe 6+2 pin standard that you would still have 30% headroom even at 450W.

There is literally no room on the tiny PCB.
That's an Nvidia FE-only problem though, right?

All of the partner cards that were using 2x8pin, 3x8pin prior to the 40-series super launch clearly had room, and I doubt any of the partners now forced to use 12VHPWR bothered to completely redesign the PCB just for this refresh.
 
It's just mind-boggling to me that they decided to shrink the pins and connector as they were doubling the current draw.
That I can understand. Any word on material changes? That may or may not help.. but 600w is a lot and wtf indeed.

That's an Nvidia FE-only problem though, right?

Here is my Zotac 4070 Ti Trinity OC.. don’t mind the rainbows :D
 

Attachments

  • IMG_0245.jpeg
    IMG_0245.jpeg
    2.4 MB · Views: 78
So buy AMD or Intel? Got it.
or 40-series cards with 2x 8-pins. It's only this set of Super refresh cards that are all forced to use the 12VHPWR, and only one of them (the 4070S) is remotely interesting anyway.

Here is my Zotac 4070 Ti Trinity OC.. don’t mind the rainbows :D
Because Zotac used the Nvidia reference board with 12VHPWR only.

Plenty of cards using older 8-pin connectors and hybrid boards ready for both (possibly same PCB for Quadro/workstation cards with PCIe connectors on the short end)

1708364439721.png

In fairness to all the AIB's who used 12VHPWR the first time around, they weren't aware what a bad connector it was at the time, because there was no real data.
 
Last edited:
Because Zotac used the Nvidia reference board with 12VHPWR only.
So I basically have what Nvidia would sell but with a different cooler?

That would explain why it only has a single bios.. first card in eons that only had 1 bios on it lol.

Edit:

It looks disgraceful in the no0d.
 
Yeah cleaner and then they put the connector in the bloody center top of the card, sticking out too far for its width to fit in most cases proper. So clean!

It all screams lazy and clumsy design to me
It would add about $0.80 to the cost of the card, but I wish GPUs had connectors on both the top edge and the far end. If the PCB is too short to reach the end of the card, just run extension wires to the end like Nvidia used to do it their 10-series and 20-series FE cards. That was much, much neater and more convenient and it solves 100% of compatibility problems by providing flexibility for cents on products that are selling for several hundred dollars.

1708365351738.png

It looks silly when it's naked like this, but it works for GPUs of any length and you can see even Nvidia's own FE PCB has the pinholes and silkscreening for both options.
 
Plenty of cards using older 8-pin connectors and hybrid boards ready for both
What if you're wrong. Look again very closely. Tell me what you see.
 

Attachments

  • whattif.gif
    whattif.gif
    39.1 KB · Views: 71
There is literally no room on the tiny PCB.
Ehhh... so you make the PCB 1cm longer... or hey... perhaps you shouldn't have downsized it to begin with, seeing as its STILL a custom format PCB that changes with every SKU placed on it.

Remind me, why did Nvidia pick a tiny PCB on ever larger cards?
Sounds like an Apple 3.5mm jack situation. Creating problems to tackle with overcomplicated solutions to push new 'standards'.

I'm sorry but I still don't see a single compelling argument to change connectors here. Its a want, not a need for Nvidia, and its built on a strategy none of us benefits from. This isn't for us. Its cost reduction. On a product line with the most abysmal perf/$ no less. Explain the madness

What if you're wrong. Look again very closely. Tell me what you see.
Hehe interesting observation
 
Back
Top