New 12V-2×6 Device Featuring Load Balancing Technology Could Be the Cure-All for 16-Pin Connector Meltdowns

Wait so this 80 dollar marvel doesn't make the connector better but makes it safer to be bad. And Nvidia can't add this to their designs and OEMs because.... it's too expensive on a 2k card? Every company with a 12vhp port should be licensing this for their cards. The cost to manufacture and include these at scale should be a blip at best. I know if I saw 2 cards. One with this integrated and one without id pay 40 bucks more for the one with it.
 
This is a neat little product.

I may just have to pick one up when I get around to assembling the game machine with the 5090.

If it fits with the Alphacool water block I bought that is. I haven't looked closely at it yet, but the EKWB block I had for my 4090 recessed access to the power connector so my cablemod bends didn't fit. That later turned out to be a blessing in disguise as those Cablemod bends were later recalled for burning up 😅

$80 seems a little steep, but it is a relatively cheap insurance policy compared to burning up a 5090....
 
Last edited:
Zarathustra said:
This is a neat little product.

I may just have to pick one up when I get around to assembling the game machine with the 5090.

If it fits with the Alphacool water block I bought that is. I haven't looked closely at it yet, but the EKWB block I had for my 4090 recessed access to the power connector so my cablemod bends didn't fit. That later turned out to be a blessing in disguise as those Cablemod bends were later recalled for burning up 😅

$80 seems a little steep, but it is a relatively cheap insurance policy compared to burning up a 5090....
Click to expand...
No, don't do it. Don't play by the rules of a-holes. It is not our responsibility to pay extra to make their product safe. If the FTC had any teeth it should've forced nvidia / OEMs to change the design.
 
MadMummy76 said:
No, don't do it. Don't play by the rules of a-holes. It is not our responsibility to pay extra to make their product safe. If the FTC had any teeth it should've forced nvidia / OEMs to change the design.
Click to expand...

Aquacomputer didn't design Nvidia's GPU's. They are just coming up with a solution since Nvidia refuses to do it right.

And Aquacomputer is a pretty good company.
 
Last edited:
Zarathustra said:
Aquacomputer didn't design Nvidia's GPU's. They are just coming up with a solution since Nvidia refuses to do it right.

And Aquacomputer is a pretty good company.
Click to expand...
They deserve per unit licensing to include this tech on every card with the connector.
 
Zarathustra said:
Aquacomputer didn't design Nvidia's GPU's. They are just coming up with a solution since Nvidia refuses to do it right.

And Aquacomputer is a pretty good company.
Click to expand...
No they didn't they are just exploiting nvidia's anti consumer bs. As soon as you buy it you normalize that they can just sell a faulty and potentially dangerous product and its up to the user to fix it if they want to sleep at night.
 
Grimlakin said:
They deserve per unit licensing to include this tech on every card with the connector.
Click to expand...
There is nothing to license here, a load balancer is not magic, nvidia could've added their own but they decided to go with the pinto route.
 
MadMummy76 said:
No they didn't they are just exploiting nvidia's anti consumer bs. As soon as you buy it you normalize that they can just sell a faulty and potentially dangerous product and its up to the user to fix it if they want to sleep at night.
Click to expand...
Look, I understand where you are coming from, but if this cheap little device can prevent a meltdown of a multi-thousand dollar build, then it is silly not to.

Nvidia is unlikely to make any changes to the power connector either way. They have been completely intransigent on this issue.

And honestly, even withholding sales are unlikely do make a difference. They consider themselves to be doing us a favor by even letting us buy anything at all, because they could make much more money by using the consumer silicon allotment for AI chips and just discontinuing consumer GPU's all together,

As long as they are raking in the billions from AI, they don't care about the consumer market in the slightest. Every time they make a consumer GPU, it is a relative cost, because of how much more those square millimeters of silicon chip could sell for if instead they used them for AI specific chips. It's honestly weird that they continue to make consumer GPU's at all. At this point it is little more than leather jacket man's pet project or maybe even some sort of fallback plan for if/when the bottom falls out of the AI market, but that's it.

They simply do not care about the consumer GPU market, and chances of them ever giving the GPU's enough attention to change the power connector are slim to none.

No amount of consumer pressure will change that. They do not care about the consumers and their money. It is insignificant compared to their enterprise AI customers.

So we are on our own, and we have to do the best we can to protect ourselves.

The same is happening to AMD what with OpenAI reportedly taking a 5% stake in the company.

Companies go where they money is, and unfortunately, even at $3k for a high end GPU, that's not us. Those days are over, Probably for good. From here on out we are likely to only get the breadcrumbs that are left over from the AI market. Our business may have built these companies, but they don't give a **** about that. They care about where the biggest profits are coming from today.
 
MadMummy76 said:
No, don't do it. Don't play by the rules of a-holes. It is not our responsibility to pay extra to make their product safe. If the FTC had any teeth it should've forced nvidia / OEMs to change the design.
Click to expand...
The answer there isn’t to boycott the Aquacool device. If anything we should all be supporting them for finding a need and providing that with a well designed product

To achieve what you are after - you need to boycott the root of the problem: nVidia. And no one is talking about doing that because no one takes the competition seriously.
 
Brian_B said:
To achieve what you are after - you need to boycott the root of the problem: nVidia. And no one is talking about doing that because no one takes the competition seriously.
Click to expand...
Been doing that for years. It's getting hard to get a decent video card that isn't from NV. And despite the focus on the AI market, NV is still up to its old tricks when it comes to making life difficult for non-NV card owners.

Don't get me wrong, a 9070XT is a fine card, and AMD apparently sold quite a few of them. But it's not like it was in the old days.
 
Grimlakin said:
And continuing to be the champion/distributor of that connection in spite of it's faults through multiple generations would make you the accomplice. And in turn part of the problem rather than a victim.
Click to expand...

I'd say it's a "both/and" not an "either/or"

1.) PCI-SIG should not have pushed the 12VHPWR standard. It was obviously rushed and poorly tested. And once the problem became apparent in the field, they should have taken action. They will - of course - say they did, and point to the updated 12V-2x6 update, but it still hasn't resolved the problem. The fact that they seem to be washing their hands of it makes them very culpable.

2.) Nvidia - designing a product with this standard, also has some testing burden, and they have failed in the same ways PCI-SIG has, by failing to take action when the problems have become apparent.

3.) Even board partners are partially to blame here. I would love to see an enthusiast board partner 5090 board with multiple traditional 8pin PCIe power connectors, though you'd need four of those to hit the same 600w which might be a bit much, and would take a pretty special PSU (though, my trusty Seasonic Prime 1200w Platinum PSU's can handle it, if I buy some extra modular power cables) I don't understand why they didn't just move to the same power connectors as the 8pin 12v motherboard connectors. As single 8-pin EPS connector can handle 300-384w. Two of those would easily have been enough for a 3090Ti, 4090 or 5090. And most modular PSU's have been designed with the same ports on the PSU side for both the 150w 8-pin PCIe power connectors and the 300w 8-pin 12V EPS connectors, so it would just have been a matter of getting more modular power cables.

If I were cross-shopping between otherwise equivalent 5090 cards, one with a 12VHPWR connector and one with two 8poin 12VEPS connectors, I would go with the latter every single time. And there is no reason board partners couldn't have released a card like this. I bet it would have been a highly desirable model too given how much enthusiasts care about this issue.

That said, I do understand Nvidia has been pressuring board partners to coalesce around the ir reference designs in recent years (to a certain extent, they certainly aren't asking them to make their wacky Founders editions) so maybe they fear reprisals if they try to introduce different power connectors.

Either way, I see it as a three-way problem. PCI-SIG, Nvidia and Board Partners. Any one of them could take steps towards solving it if they wanted to, but they prefer to just ignore the problem, stick their fingers in their ears and watch the world (literally) burn.
 
Let's not forget that AMD did include the same connector on select models and at least one person has reported an issue about a month ago. NV is by far and wide the biggest deserving target but they are not the only ones to use this.
 
I would not be surprised they were working on a different solution to this, but they need to get everyone at pci-sig to agree on the new standard.

and of course changing this would probably royaly piss off some PSU makers as they would be stuck with a ton of matrrials they can't use anymore.
 
Peter_Brosdahl said:
Let's not forget that AMD did include the same connector on select models and at least one person has reported an issue about a month ago. NV is by far and wide the biggest deserving target but they are not the only ones to use this.
Click to expand...
Was that an AMD design decision, or a board partner design decision though?

I'm asking because I don't know the answer, not because it is some sort of "gotcha" question.

I don't think AMD makes any of their own video cards, at least not in the consumer space.

So the question is, did they spec the 12VHPWR connector on those, or did the board partner do it on their own with a custom board design?
 
You must log in or register to reply here.
Become a Patron!
Back
Top