my 2c, some people are going way overboard in conflating the issue and sensationalising it, pointing at all these other things they personally don't like about what's going on that are surrounding factors and who it's about, when really, the 4 to 1 connector is the issue and the real material issue at hand here. No 3090Ti's seemed to suffer this despite the same power requirements (worse spikes) and similar sizing, no other cables seem to suffer this issue whatsoever.

Should have Nvidia gotten the adapter right first time? of course and I'm sure they know this 1000x by now, someone fucked up and let a poor design/execution through, especially one that couldn't be bent given the cards are quite tall, which again didn't seem to happy with 3090Ti's.

Recall all adapters, give people new ones, fix any cards where the adapter also managed to melt the card-side connector, and done.

Much of the rest I read is sensationalist drivel. For every rational take there seems to be one ready to get torches and pitchforks.

ValantarLooking at those connector teardown pics from Igor's Lab, it's really no wonder these are failing. Those outer wires are soldered to a single thin sheet of copper with no bends, ridges or crimps to strengthen it, meaning even moderate bending stress will shear the metal - like you can see happened during disassembly. It's really no wonder Cablemod tells you not to bend the wire horizontally, as that essentially guarantees that the outer solder tabs will shear off. Nvidia did use a strain relief boot to try and alleviate this (or as @jonnyGURU mentioned in another thread, give people something to grab so they don't tear out the wiring when trying to disconnect the plug), but it was clearly insufficient - and arguably this design is fundamentally flawed and should never have made it to market. I wouldn't call this a major scandal, but for a hyper-expensive, ultra-premium GPU? It's completely and utterly unacceptable. Heck, it would be unacceptable for a $300 GPU - but a $300 GPU likely woldn't need a stupid power adapter in the first place.


... or they want to look at what went wrong so that they can figure out how and improve on their design? How else would you suggest that they do so?

Since they are aware of the problem they can simply cook a few in their own R&D lab and go with those.
I guess they failed to do that earlier and it's not like they coudn't - They can if they want to but it's really for getting the evidence out of the hands of those that may file suit and disposing of it.

Same basic thing that happens if you're in a fast food joint and find a bug or something else nasty in your burger and you ask for the manager to complain about it - They will want you to let them "See" it and that always means they want you hand it to them to look at.
They are actually trained to get it away from the customer, not to give it back no matter what, securely dispose of it ASAP and not to admit fault except to offer an apology and a refund over it.

What that means is no evidence to present - No lawsuit and it's the same thing going on here.

That I can promise you and there have been past precidents set from a legal standpoint like when AMD was sued for saying a BD chip was an 8 core when it had only 4 real cores.
I believe at least a few here got some of that payout - I could have but didn't bother with it.

kapone32I guess you have never had to do an RMA with them. Gigabyte is not bad, even MSI is not that bad but Asus has (as long as I have dealt with them) been less than honourable in the way they administer their Warranty service. This is not some feeling that I have but experience that I am talking about.

I have experience with Asus RMA, took a while, but was okay. Absolutely nothing special.

ValantarNo, but they want to be stingy. This is a company that isn't happy with sub-60% margins, after all.

At least they aren't TI level bad. Basically forcing American schools to only buy their calculators with chips from the 80s or 90s and make students pay 5-6 times more than what their old junk is actually worth. I never understood why would someone put up with their bullshit, but that's the most extreme predatory profiteering case in semiconductor industry that I know.

The red spiritI have experience with Asus RMA, took a while, but was okay. Absolutely nothing special.


At least they aren't TI level bad. Basically forcing American schools to only buy their calculators with chips from the 80s or 90s and make students pay 5-6 times more than what their old junk is actually worth. I never understood why would someone put up with their bullshit, but that's the most extreme predatory profiteering case in semiconductor industry that I know.

Asus RMA shipped to the wrong state for me. EVGA has had the best customer service from all brands that I've personally seen. Just because you think their business croaked or whatever on greed doesn't mean that's what happened. They explained what happened to multiple outlets, and it makes sense.

Can't change your mind, since well, you're pretty dead set on this.

ToothlessAsus RMA shipped to the wrong state for me. EVGA has had the best customer service from all brands that I've personally seen. Just because you think their business croaked or whatever on greed doesn't mean that's what happened. They explained what happened to multiple outlets, and it makes sense.

Can't change your mind, since well, you're pretty dead set on this.

And I clearly said that eVGA was no good outside of USA.

They should redesign the cards with standard proper ports; extend the cards a bit and add 3 or 4 6/8 pin ports prob solved.

I should get paid for this idea btw wink ;)

BonesSince they are aware of the problem they can simply cook a few in their own R&D lab and go with those.
I guess they failed to do that earlier and it's not like they coudn't - They can if they want to but it's really for getting the evidence out of the hands of those that may file suit and disposing of it.

Reproducing an issue isn't the same as seeing how it occurred in the wild. Any responsible investigation would do both. Non-self-induced errors are vastly more valuable to such an investigation than anything reproduced in labs. Why? Because they occurred in real world use.

BonesSame basic thing that happens if you're in a fast food joint and find a bug or something else nasty in your burger and you ask for the manager to complain about it - They will want you to let them "See" it and that always means they want you hand it to them to look at.
They are actually trained to get it away from the customer, not to give it back no matter what, securely dispose of it ASAP and not to admit fault except to offer an apology and a refund over it.

Uh, so.... you know that photos exist, right? These things will be extensively documented, including every step of the RMA process, which will be logged in writing in multiple steps long before these adapters are sent to Nvidia. And people will be posting about it - wouldn't you if your new $1600-2000 GPU melted? There is exactly zero chance of Nvidia being able to hide evidence of this happening.

BonesWhat that means is no evidence to present - No lawsuit and it's the same thing going on here.

See above.

Seriously, this line of argumentation makes no sense whatsoever, and the fast food analogy isn't really suitable (there won't be a documented RMA for your burger; it won't take days or weeks for it to be returned for inspection, among other major inconsistencies).

BonesThat I can promise you and there have been past precidents set from a legal standpoint like when AMD was sued for saying a BD chip was an 8 core when it had only 4 real cores.

... wait, are you saying that back then, AMD went around and collected and destroyed people's CPUs so that nobody could verify whether these actually had 4 or 8 cores? :confused: 'Cause if that isn't what you're saying, what's the relevance here?

BonesI believe at least a few here got some of that payout - I could have but didn't bother with it.

... and? "Companies can be successfully sued" doesn't prove anything more than that companies exist in the world, period.

ChaitanyaLeather jacket guy following Apple's philosophy of searching problems for which solutions dont exist yet.

Much better now.

JismIf they make compute and / or graphics cards universal in relation of power delivery they are cutting costs but also e-waste.

The connector is'nt bad; it works. It just does'nt work when the thing is being bended. Which increases resistance and things start to get quite hot.

By creating sub par solutions, not only are they increasing e-waste, increasing costs and worse creating something dangerous.

Ot looks like Nvidia could be forced to recall all adapters. This situation breaks fire hazard regulations in many countries, so authorities will be rather quick in telling them what to do.

they are going to mess Huang's kitchen

With respect:

ValantarReproducing an issue isn't the same as seeing how it occurred in the wild. Any responsible investigation would do both. Non-self-induced errors are vastly more valuable to such an investigation than anything reproduced in labs. Why? Because they occurred in real world use.

No it's not but the same basic cause of failure about the hardware can be determined. An example is when the starter in your auto goes bad and burns up, requiring a new one.
It can be caused by several things but in the end it's always one of two things - Wear and tear (age) or it got too hot and cooked, these two things being root causes of failure which can be determined. Wear and tear is a given, if it's worn out it's worn out and actually expected since it has a certain "Life Span" anyway but premature failure is another thing.

Let's run down the list of what can cause it in this example off the top of my head:
Weak/bad battery, battery getting weak because the alternator wasn't charging it, bad battery cables, weak/loose battery wiring connections, the engine's ignition timing being too far advanced making it difficult to turn it over, the vehicle always being driven short distances so the battery never recover's it charge from being started (Slowly dying as a result - Weak battery), improperly maintained battery (Dirty posts or dry cells) or the battery going bad by being too close to a heat source causing it to cook and swell up along with drying out internally....

There are many things that can be a cause of failure - True, but the failure based on the hardware is almost always the same if it wasn't engineered correctly and that can be determined in-house if they really took time to investigate it like Igor did.
A small lab solved what a big corp with all it's resources coudn't during R&D time - There is no denying it. |
To be fair, there is no way to take into account every possible scenario but at least it's entirely possible to cover most things so you won't have failures on such a large scale by the numbers.

ValantarUh, so.... you know that photos exist, right? These things will be extensively documented, including every step of the RMA process, which will be logged in writing in multiple steps long before these adapters are sent to Nvidia. And people will be posting about it - wouldn't you if your new $1600-2000 GPU melted? There is exactly zero chance of Nvidia being able to hide evidence of this happening.

Depends on the photo and camera taking the shot.
Not to mention for the sake of a lawsuit photos can be "Doctored" to try and prove it as real if the one filing suit was trying to sue for the sake of getting paid - It's been done before and will happen again at some point.
Of course a pic can go a long ways to that end but still, there is nothing like having the actual article physically onhand to present as evidence.
If there is no fried card to present, there is nothing to cover up in the first place in the legal sense of it.

ValantarSee above.

Seriously, this line of argumentation makes no sense whatsoever, and the fast food analogy isn't really suitable (there won't be a documented RMA for your burger; it won't take days or weeks for it to be returned for inspection, among other major inconsistencies).

You're not well versed in legal stuff or so it seems to me - I don't get why you don't understand the simple fact about evidence to support a claim.
Hearing about it is one thing, physically seeing it up close and personal is way better and stronger too than word alone.

As for the burger, yes it's still relevant in how or what the company could try to cover their asses to avoid a payout.
I don't get why this concept eludes you, it's not about THE burger or THE card, it's about a claim being factually proven and justification of a ruling made - The end result (Ruling) they are trying to dodge is a judgement resulting in payout to the one(s) filing suit.

Valantar... wait, are you saying that back then, AMD went around and collected and destroyed people's CPUs so that nobody could verify whether these actually had 4 or 8 cores? :confused: 'Cause if that isn't what you're saying, what's the relevance here?

Seriously - Twisting the meaning of what I posted will get you exactly nowhere.
You're trying to make it all based on the physical rather than the concept, the concept which is my point about it. Concept always being someone has the opinion they were slighted, screwed over, hustled, taken from unfairly which is why suits are filed but at the same time, if you know you're been screwed over then it's more than just simple opinion - You still have to have the evidence to prove it as fact to more folks than yourself alone.

Valantar... and? "Companies can be successfully sued" doesn't prove anything more than that companies exist in the world, period.

No - It proves they can be sued exactly as said - The reason(s) why, real or imagined, faked or truthful can vary to about any degree you can name.

Crackong

Jay just found out the original 3090Ti adaptors are fine ( pin to pin construction ) (08:35)

Maybe Nvidia thought themselves confident ? then comes the cheapout 4090 adaptors ( the soldering mess )

looking Jayz videos for serious content is like looking for water in the desert

BonesNo it's not but the same basic cause of failure about the hardware can be determined. An example is when the starter in your auto goes bad and burns up, requiring a new one.
It can be caused by several things but in the end it's always one of two things - Wear and tear (age) or it got too hot and cooked, these two things being root causes of failure which can be determined. Wear and tear is a given, if it's worn out it's worn out and actually expected since it has a certain "Life Span" anyway but premature failure is another thing.

Let's run down the list of what can cause it in this example off the top of my head:
Weak/bad battery, battery getting weak because the alternator wasn't charging it, bad battery cables, weak/loose battery wiring connections, the engine's ignition timing being too far advanced making it difficult to turn it over, the vehicle always being driven short distances so the battery never recover's it charge from being started (Slowly dying as a result - Weak battery), improperly maintained battery (Dirty posts or dry cells) or the battery going bad by being too close to a heat source causing it to cook and swell up along with drying out internally....

There are many things that can be a cause of failure - True, but the failure based on the hardware is almost always the same if it wasn't engineered correctly and that can be determined in-house if they really took time to investigate it like Igor did.
A small lab solved what a big corp with all it's resources coudn't during R&D time - There is no denying it. |
To be fair, there is no way to take into account every possible scenario but at least it's entirely possible to cover most things so you won't have failures on such a large scale by the numbers.

I don't see how any of this is in any way in conflict with what I've been saying. At all. Lab testing to provoke failures allow you to identify many possible failure modes, as well as (hopefully) identify which of these are the most likely. This is a general type of troubleshooting, identifying potential risks of various severity. Investigating actual, real-world failures allows you to identify specific failure modes that appear in the real world. These two modes of investigation complement each other, and neither is sufficient on their own. All I've been arguing here is that there is absolutely nothing suspicious about Nvidia asking for these adapters to be sent to them, because specific investigation of real-world failures is a crucial step of fixing this.

BonesDepends on the photo and camera taking the shot.

No. A photo of a failed part is a photo of a failed part - its quality and informational value is variable and can be debated, but at the very least it serves as documentation of an actual failure of some sort occurring. It might not be sufficient evidence for Nvidia being at fault, but in a civil suit, such a photo + no compelling evidence for user error + a mysteriously disappeared burnt adapter in Nvidia's custody? That'd be an easy win for the customer. Not in a criminal case, but in a civil case? Either that, or it'd see the case handed over to the relevant authorities for a criminal investigation.

BonesNot to mention for the sake of a lawsuit photos can be "Doctored" to try and prove it as real if the one filing suit was trying to sue for the sake of getting paid - It's been done before and will happen again at some point.

Which would put the burden of proof for this on Nvidia to show that the photos were doctored if this was the case.

BonesOf course a pic can go a long ways to that end but still, there is nothing like having the actual article physically onhand to present as evidence.

And if this was necessary, any lawsuit would require Nvidia to provide ample documentation of any investigation of the returned adapter, if not access to the adapter itself, as a part of the discovery process. Alongside all RMA and shipping logs. Witholding relevant, requested evidence from discovery is illegal, and will be sanctioned by the court, and might even result in a default judgement against Nvidia.

BonesIf there is no fried card to present, there is nothing to cover up in the first place in the legal sense of it.

See above. If a photo of a fried adapter exists, as well as a document chain confirming its submission for RMA and subsequent return to Nvidia, no court in any sensible country would accept Nvidia saying "whoops, you know, we might have misplaced that" and not follow up on that part of the suit.

BonesYou're not well versed in legal stuff or so it seems to me - I don't get why you don't understand the simple fact about evidence to support a claim.
Hearing about it is one thing, physically seeing it up close and personal is way better and stronger too than word alone.

Photos are routinely used as evidence in both civil and criminal cases thousands of times every single day, and if sued, Nvidia would need to prove beyond a reasonable doubt that any such photos were doctored - which they wouldn't be able to. Faking a low quality phone photo convincingly is incredibly difficult. They could no doubt find some "expert" to testifythat it "might" be doctored, but then the person/group suing them would find their own expert to counter this, rendering the argument moot, word-v-word. And, of course, the lack of a photo but existence of RMA logs and shipping logs of a burnt adapter wouldn't help Nvidia anyhow.

BonesAs for the burger, yes it's still relevant in how or what the company could try to cover their asses to avoid a payout.

... except for the fact that a burger would spoil within days, and the evidence would disappear by itself unless work is done to preserve it - unlike this case, where there'd need to be actual effort put into destroying it, and a document chain showing what has happened to it. You're conflating two different ends of a vast spectrum of malfeasance here, pretending that they're one and the same. Whisking away a burger with whatever crap in it a minute or two after a customer has a disgusted outburst after finding that thing in their food is vastly different from Nvidia asking their AIB partners to return power adapters submitted for RMA to them. If you don't see the difference between these two, you really need to take a look in the mirror regarding that "not well versed in legal stuff" claim.

BonesI don't get why this concept eludes you, it's not about THE burger or THE card, it's about a claim being factually proven and justification of a ruling made - The end result (Ruling) they are trying to dodge is a judgement resulting in payout to the one(s) filing suit.

... that doesn't elude me whatsoever. I'm arguing that there's a vastly different basis for making evidence-backed claims in these different cases, making them essentially uncomparable. The nature of the basis of the claims, and the things involved in causing the claims, are vastly different and operate differently in material, temporal and causal ways. A GPU power adapter does not spoil or go bad if left unattended. A burger does. A melted GPU adapter literally cannot be taken from the customer by an employee before they have a chance to document the incident - which is what gives companies the opportunity to hide and destroy evidence. And, crucially, failure to uphold hygiene standards and failure to design a safe power adapter are quite different causes of failure, attributable to entirely different people and/or entities.

BonesSeriously - Twisting the meaning of what I posted will get you exactly nowhere.

I'm not twisting it whatsoever - I'm demonstrating the logical leaps you're making in your reasoning. They're there for all to see - if you don't, then that's on you. That's why I'm pointing them out.

BonesYou're trying to make it all based on the physical rather than the concept, the concept which is my point about it.

... except that such a lawsuit, if it were to happen, would be based on actual events, not conceptual ones. I don't care whatsoever about any abstracted conceptual argument you're making - I'm arguing that in the real world, anyone interested in fixing a flaw with a product will collect as many samples as feasible of the failed product in order to identify and fix the flaw for future revisions.

BonesConcept always being someone has the opinion they were slighted, screwed over, hustled, taken from unfairly which is why suits are filed but at the same time, if you know you're been screwed over then it's more than just simple opinion - You still have to have the evidence to prove it as fact to more folks than yourself alone.

Uh ... so ... what? Sorry, I really don't follow here. You're arguing for Nvidia collecting these being suspect because conceptually they might do so to avoid lawsuits? Is that what you're saying? 'Cause if that's the case, then conceptually it's more likely that they're doing this to better identify the specific modes of failure found in the real world. These are not different ontological levels.

And, as I've said above: "disappearing" the adapters wouldn't help. There would be photographs. There would be RMA logs - most likely with extensive photos attached from AIB partner RMA techs. There would be shipping logs from board partners of adapters being sent to Nvidia, which would be traceable back to specific RMA numbers. There would be tracking numbers and receipt signatures. All of which would be relevant evidence for either the fault at hand, or Nvidia actively working to destroy evidence of a fault. The latter of which would be rather interesting to the relevant authorities, and would land Nvidia in a lot more hot water than any civil suit. You generally don't risk criminal liability to avoid civil liability - that's pretty damn stupid.

BonesNo - It proves they can be sued exactly as said - The reason(s) why, real or imagined, faked or truthful can vary to about any degree you can name.

And how is this relevant? Is "frivolous or unfounded lawsuits exist" an argument for or against anything at all in specific? No. The entire gist of your argument seems to be that if Nvidia can make the physical adapters go away, this would all go away, which just shows a massive blindness to how obvious this would be to any court.

Seriously, the logic underpinning this whole "Nvidia is doing this to hide evidence" spiel is woefully naive and out of touch with the actual workings of both the corporate world and legal systems. Unless they'd be able to also delete all relevant RMA logs, including emails sent to the customer about this, it wouldn't help them at all - it would in fact harm them, as it would show clear signs of trying to hide evidence. Which is, in fact, illegal! Whoops!

Chrispy_I wonder if it's shit like this connector that was the final straw that caused EVGA to nope out. It does appear that Nvidia is forcing AIBs to use the connector, and that Nvidia make the only, flawed adapter. Coincidentally, I made a whole bunch (48, I guess) custom PCIe power cables for my mining rigs; I cannibalised about 15 bags of Corsair Type 4 cables and individually re-pinned the wires into new Mini-Fit Jr. plugs:



Realistically, I wanted peace of mind that I couldn't get from AliExpress/Amazon pre-made cables because I couldn't trust the crimp quality or the wire gauge used. On Corsair cables that come with their PSUs, I trust both - so I spent the time reconfiguring them to my needs. Having dismantled or reassembled multiple pins in almost 100 Mini-Fit Jr. plugs, I think it's fair to say that they are adequate but not more than that. The pins, the crimp mechanism, the receptacle, they're all fine for the ~4A per pair 12V pin they're rated for. I don't know if I'd trust them with more than 8A, and I sure as hell wouldn't be happy doing what Nvidia's done with their 600W adapter which, after Igor's lab teardown of the damn thing, seems to push up to 16.7A down two of the pairs.

I'm not sure about 8A per pin on the standard Mini Fit Jr connector. Nvidia are using over 8A on every pin of the even smaller HPWR pins. That's bad enough IMO but the fact that they've doubled-up the last two because 4 connectors doesn't integer-divide into 6 pairs neatly means that almost 17A is being asked of a pin that might handle 5, on a good day.

I think the saving grace for Nvidia is that of all the 4090s sold, the number that are regularly drawing the full 600W is small, and of that, the number drawing 600W for long periods is even smaller.

I believe 12VHPWR is based on Micro-Fit+, which is rated for 13A on 16ga pins. The terminal block in NV's adapter isn't part of that system, so it looks like some custom part that may or may not meet the current requirements. Further discussion here.

ValantarSeriously?

Quite seriously. Please, do no play an apologetic for a failed design and a corrupt company / companies.
Anyone with common sense can tell a good design, that is 8-pin connector rated at 150W and a bad design of this new "16-pin" rated at 600W with its tiny contacts. That's spells bad news from the beginning.

pavleQuite seriously. Please, do no play an apologetic for a failed design and a corrupt company / companies.
Anyone with common sense can tell a good design, that is 8-pin connector rated at 150W and a bad design of this new "16-pin" rated at 600W with its tiny contacts. That's spells bad news from the beginning.

There's no apologism here. If you're a manufacturer and are trying to rectify a failing part, you want examples of that failing part for analysis. Photos are ok, but physical samples are better. Corruption doesn't enter into it.

To your second point, NV's adapter is a poor design. This is not in dispute. But the claim that anyone with common sense can tell a good design, or spot a bad one, is false. Looking like a good/poor design is not the same as being either of those things. The contacts themselves are not the problem, at least that's how it seems so far. Many cards are running on direct-to-PSU connectors without issue. What seems to be the culprit is how the wires are merged inside NV's adapter specifically. That's the poor part of the design, and is absolutely not obvious to any observer, because the flaws can't even be seen without disassembling the connector.

80-watt HamsterFurther discussion here.

Why do we have second thread for same problem? BTW this stuff is no longer an investigation or data collection, it's already figured out. If we need another thread, then it could be about what 4090 buyers should do nowm how to solve adapter problem and how to keep nVidia accountable for their insufficient engineering.

The red spiritWhy do we have second thread for same problem? BTW this stuff is no longer an investigation or data collection, it's already figured out. If we need another thread, then it could be about what 4090 buyers should do nowm how to solve adapter problem and how to keep nVidia accountable for their insufficient engineering.

Because I wanted somewhere to have a useful discussion rather than ... whatever's going on here. Feel free to not participate.

80-watt HamsterBecause I wanted somewhere to have a useful discussion rather than ... whatever's going on here. Feel free to not participate.

There's not much left to talk about, just saying.

The red spiritThere's not much left to talk about, just saying.

Then the thread will die and we'll all move on with our lives. *shrug*

pavleQuite seriously. Please, do no play an apologetic for a failed design and a corrupt company / companies.
Anyone with common sense can tell a good design, that is 8-pin connector rated at 150W and a bad design of this new "16-pin" rated at 600W with its tiny contacts. That's spells bad news from the beginning.

Look, I'm the last person to defend nVidia, but in order to figure out what's going on, you need to investigate the failures. If nVidia is having trouble reproducing the problem, then they need the failing cases to do further analysis. That isn't to say nVidia isn't a morally bankrupt company, they are, but this is what needs to be done sometimes when there is an issue occurring that they can't reliably reproduce.

I'd rather roast nVidia for being stupid by introducing this connector. It was a braindead decision before connectors started melting and this is just further proof of that.

Valantar~Snip~

Not going down further into that rabbit hole with you.

I've said what I've said and I stand by it - Either you get it or you don't.
I've attempted to explain my reasons why I'm thinking all this once already but it's like you're looking for an arguement, so I'll pass this time and indulge your apparent need to bicker later.

Your refusal to see my points about it, whether you agree or disagree with me on those doesn't make it any more or less real, nor proves anything that changes what it is or is not.
I had thoughts, I stated them and it's good enough by me.
It's also something we don't get to decide so any amount of arguing on our part amongst each other ain't changing a damned thing about it so there is no point to it carrying on.

80-watt HamsterThen the thread will die and we'll all move on with our lives. *shrug*

"Please!!! Grant us this sweet mercy - Oh revered mods of the many threads"........ :D

pavleQuite seriously. Please, do no play an apologetic for a failed design and a corrupt company / companies.

Where have I been an apologist for Nvidia in any way, shape or form? You mean by saying things like this?

ValantarIt's completely and utterly unacceptable.

Please don't conflate being an apologist with arguing against the stupid, conspiratorical "oh, they're just asking for these back to hide evidence" nonsense - which is really pure nonsense. My "seriously" was directed at your "It's obvious how it fails" comment, as that shows such a fundamental obliviousness to the difference between seeing "oh, this is broken" and actually figuring out how it failed on the level of materials and construction.

@Aquinus said it excellently above:

AquinusThat isn't to say nVidia isn't a morally bankrupt company, they are, but this is what needs to be done sometimes when there is an issue occurring that they can't reliably reproduce.

Which is precisely the point. There is no contradiction between "Nvidia are generally exploitative a**holes" and "Nvidia are asking for these back in order to investigate the specifics of the failures". None whatsoever. There are scenarios in which they can contradict each other, but there are also (and IMO more relevant) scenarios where these two things support each other. For example, avoiding further liability and avoiding unnecessary regulatory and legal scrutiny by taking action themselves. Even the worst company on the planet will for a significant proportion of the time be "doing the right thing" - as that would often serve their interests more than doing something objectionable.

pavleAnyone with common sense can tell a good design, that is 8-pin connector rated at 150W and a bad design of this new "16-pin" rated at 600W with its tiny contacts. That's spells bad news from the beginning.

The connector is based on an existing industrial connector standard (developed by Molex, called Micro-Fit+, rated for 11-13A/pin), and uses current ratings from that standard. Nvidia might have been a proponent of having this particular variant of it adopted as the PCIe 5.0 power connector, but the connector is based on a rigorously tested industry standard. 13A*6 wire pairs*12V=936W - so 600W leaves a ~50% safety margin, or room to use 11A-rated pins with a slightly lower safety margin.

Also, the 8-pin PCIe connector has massive safety margins, and moving to a 50% margin is overall sensible. It still just uses 3 wire pairs for power (the last two pins are sense pins), so 150W/3 wire pairs/12V=4.17A/pin. Mini-fit Jr. pins (which is what PCIe power connectors use) are typically rated for 9A (there are also 7A variants, possibly also lower grade, shittier ones), which leaves a 116% safety margin. On the other hand, EPS12V cables, which use the same connector and pins but use all four wire pairs for 12V, are rated for 336W, or 7A/pin - 29% safety margin. So, going on (well tested and docutmented) ratings alone, 12VHPWR should be safer than EPS12V - though of course that supposes a competent design and no design fuckups.

The problem here is that Nvidia for these adapters made what seems to be a completely harebrained design decision in adapting four 8-pin PCIe connectors to these 12 pins, leaving them with severe weaknesses. Still, the specifics of these weaknesses need to be investigated so that their redesign can be stronger. But crucially, there's nothing suspicious about Nvidia asking for the burnt-out adapters back for testing now that they're failing - that just shows that they want to see what went wrong for themselves, and to test it in their own labs.

kapone32There is no way EVGA is greedier than Nvidia. Their warranty service compared to other vendors are exemplary.

You paid for that warranty.

Heats up when certain pins become unbalanced.
It is necessary to check with a tester whether solder cracks have occurred inside.