NV 12VHPWR issues revisited

[DrMrLordX]

(looking at you, corsair, cablemod, and MSI).

Cablemod???

 

[Hail The Brain Slug]

Cablemod???

Did you somehow miss that cablemod was starting numerous fires and issued a recall of their 12vhpwr adapters? And then they continued to cause fires even after the recall and their claim it was fixed.

 

[igor_kavinski]

If I suddenly had money to waste, I would go with the ASUS BTF 5090. Shame that they can't (or won't) standardize on this connector.

 

[DrMrLordX]

Did you somehow miss that cablemod was starting numerous fires and issued a recall of their 12vhpwr adapters? And then they continued to cause fires even after the recall and their claim it was fixed.

Umm can't tell if you're joking or not . . . you aren't serious, are you? You think THEY caused those fires?

 

[igor_kavinski]

Umm can't tell if you're joking or not . . . you aren't serious, are you? You think THEY caused those fires?

This is an admission, don't you think? https://cablemod.com/adapterrecall/

 

[DrMrLordX]

This is an admission, don't you think? https://cablemod.com/adapterrecall/

No, it isn't. If it's an admission of anything, it's an admission of defeat: it isn't possible to make a "safe" 12VHPWR cable.

 

[igor_kavinski]

it isn't possible to make a "safe" 12VHPWR cable.

It might be but then it would be Iron Man level of tech

 

[Hail The Brain Slug]

No, it isn't. If it's an admission of anything, it's an admission of defeat: it isn't possible to make a "safe" 12VHPWR cable.

Oh no! I said something bad about a company you like!

 

[coercitiv]

No, it isn't. If it's an admission of anything, it's an admission of defeat: it isn't possible to make a "safe" 12VHPWR cable.

I know what you meant but... it's the weekend so nitpicking is allowed:

This recall [...] does not affect our angled cables.

 

[Hail The Brain Slug]

Umm can't tell if you're joking or not . . . you aren't serious, are you? You think THEY caused those fires?

I don't know if you intentionally didn't read the rest of the post. The 12Vhpwr/12v2x6 spec lacks details about how the pins and sockets are supposed to be designed, as well as lacking tolerance information.

This has led to multiple, fundamentally different pin and socket designs, some of them are much worse than others.

I called out these three brands because they are the most prevalent brands I've seen associated with these incidents. It's not strictly their fault, they didn't design the pins and sockets they are using.

You seem to have conflated my calling out that the data seems to overrepresent these brands as somehow they are solely responsible for the issue.

It all goes back to a bad spec, obviously, but to pretend like the issue just exists at the exact same rate for everyone and every brand is to ignore the nature of the problem.

 

[Dave3000]

Suppose my RTX 4090 Founder's Edition connector is to burn out any day now and the card is still under warranty, and I end up RMA'ing the card, and the manufacturer did nothave any more of the 4090's in stock, what would I most likely get back; a RTX 5080, RTX 5080 and some money, or a RTX 5090 (I doubt that because it's much more expensive than the 4090 even though it's in the same class)? I just worry that if I ever need to RMA the card and it can't be repaired, that I will get a lower class card in return due to the retail price of the 5090. Anyone here RMA a 4090 then got a 5080 in return?

 

[gdansk]

Suppose my RTX 4090 Founder's Edition connector is to burn out any day now and the card is still under warranty, and I end up RMA'ing the card, and the manufacturer did nothave any more of the 4090's in stock, what would I most likely get back; a RTX 5080, RTX 5080 and some money, or a RTX 5090 (I doubt that because it's much more expensive than the 4090 even though it's in the same class)? I just worry that if I ever need to RMA the card and it can't be repaired, that I will get a lower class card in return due to the retail price of the 5090. Anyone here RMA a 4090 then got a 5080 in return?

That would be a substantial downgrade and should not be accepted. It doesn't have the Error Checking memory option in the Nvidia Control Panel and is thus not even an in the same product category by Nvidia's standards. I would demand a 4090 or a refund.

 

[DrMrLordX]

I don't know if you intentionally didn't read the rest of the post. The 12Vhpwr/12v2x6 spec lacks details about how the pins and sockets are supposed to be designed, as well as lacking tolerance information.

This has led to multiple, fundamentally different pin and socket designs, some of them are much worse than others.

All the NV designs we've seen can (and sometimes do) push over 40a over a single pin! The only thing saving AMD cards so far is that they don't pull as much power and there aren't as many of them. You can't design a compliant cable that can operate under those conditions.

I called out these three brands because they are the most prevalent brands I've seen associated with these incidents. It's not strictly their fault, they didn't design the pins and sockets they are using.

I don't even "like" Cablemods, I only know they got dumped on by NV shills when the reality is they were trying to fix problems that were initially blamed on awkward cable bends leading to improper installations. It turns out that even properly-seated cables can and will burn out, and that Cablemods' efforts were in vain.

You seem to have conflated my calling out that the data seems to overrepresent these brands as somehow they are solely responsible for the issue.

They aren't even remotely responsible for this issue!

It all goes back to a bad spec, obviously, but to pretend like the issue just exists at the exact same rate for everyone and every brand is to ignore the nature of the problem.

The rate at which certain cables experience these problems is ultimately irrelevant. You can not make a 12VHPWR or 12V2x6 cable that can handle that much current over one pin, AND there's nothing the cable manufacturers can do to prevent that happening. The cables most likely to experience this problem early on in the 4090's run were the cheap adapters, so the companies that sold supposedly-safe replacements for those adapters were the ones getting more failures down the road, simple as.

All 12VHPWR and 12V2x6 cables risk catastrophic failures due to current imbalance. All known 12VHPWR/12V2x6 cards - including AMD cards! - are suspect, with the higher-power draw cards being more likely to induce failures.

 

[igor_kavinski]

ASRock's $40 16-pin power cable has overheating protection designed to prevent meltdowns — company claims a 90-degree design ensures worry-free installation

Only compatible with ASRock Taichi and Phantom Gaming power supplies, though

www.tomshardware.com

ASROCK's solution but it requires their own PSUs, or those compatible with their protection mechanism.

 

[DrMrLordX]

ASROCK's solution but it requires their own PSUs, or those compatible with their protection mechanism.

It's sad that they have to resort to a temp sensor instead of being able to throttle current per pin. "It is what it is".

 

[igor_kavinski]

It's sad that they have to resort to a temp sensor instead of being able to throttle current per pin. "It is what it is".

It's just Nvidia giving a middle finger to their loyal zombies. Hey, if it melts, just get it repaired! You had the money to afford it in the first place. Don't be cheap!

 

[DrMrLordX]

It's just Nvidia giving a middle finger to their loyal zombies. Hey, if it melts, just get it repaired! You had the money to afford it in the first place. Don't be cheap!

Let's be fair: AMD doesn't seem to have any better implementations of 12v2x6 or 12VHPWR.

 

[Fallen Kell]

Let's be fair: AMD doesn't seem to have any better implementations of 12v2x6 or 12VHPWR.

Exactly. The problem is the 12v2x6 itself. It needs the power supplies to have over-current limit protection on each of the 6 lines to match the rating of the physical wires and connector...

 

[DrMrLordX]

Exactly. The problem is the 12v2x6 itself. It needs the power supplies to have over-current limit protection on each of the 6 lines to match the rating of the physical wires and connector...

Right, but if I recall correctly (based on one of der8auer's videos), there's no practical way for the PSU to do that alone. There needs to be changes on the cards.

 

[Ranulf]

This is on the card makers who wanted a new standard and then approved a bad design, bad cable quality and Nvidia drops their old better circuit design on their newer rtx cards (as Buildzoid pointed out months ago).

 

[igor_kavinski]

This is on the card makers who wanted a new standard and then approved a bad design, bad cable quality and Nvidia drops their old better circuit design on their newer rtx cards (as Buildzoid pointed out months ago).

Conspiracy theory: Nvidia did that to make their FE cards look better than their AIB partner cards.

 

[DrMrLordX]

Conspiracy theory: Nvidia did that to make their FE cards look better than their AIB partner cards.

According to der8auer, the FE cards are more prone to burnout that some AIB cards (especially on the 5090).

 

[regen1]

aqua computer AMPINEL Exclusive Review – The end of all problems and fear with the 12V-2X6 or 12VHPWR connector | igor´sLAB

There are products that are not so much a response to a gap in the market, but rather the result of a collective failure that created the need in the first place. This is where the aqua computer…

www.igorslab.de