40 Series cards and melted connectors

[starsmine]

2 hours ago, Alvin853 said:

Just because something is rare doesn't mean it is user error. And the failure rates of previous 6 and 8 pin PCIe connectors were much lower.

Also there were no widespread issues reported on 3090Ti cards that used the same connector. There clearly is something wrong with RTX 4000 series cards, and maybe user error can make it worse, but the problem is not user error.

I didnt say that because something is rare that it is user error.

But I do say, that a relatively rare user error and an even significantly rarer non-user error. that it's a nothing burger. 

We have yet to see any evidence of any card failing in a similar matter to the user error failure mode that was not user error, now it would be dumb to say with certainty that there isn't one out ther. But there is not a single shred of evidence pointing to a non-user error failure mode so to assume it exists is just as folly. It needs to be investigated yes, but spreading FUD based off of misinformation is wrong. Don't do that. That does not mean don't discuss it, but when you come out of the gates swinging like this, I can not, nor should anyone take it seriously. 

 

2 hours ago, Alvin853 said:

Jay had a good discussion about this on the RTFM show this week:
 

 

Im going to be potentially unfairly dismissive. Jay does not take any editorial discretion in his normal published videos that he can edit and verify before putting them out there, podcasts in this format are always even worse in terms of being factual just by the nature of it just being guys shooting the shit. And that style of video I am fine with, it can be entertaining, and informative as in making you look at a topic with different perspectives, but they are never something to cite.

[Alvin853]

1 hour ago, starsmine said:

I didnt say that because something is rare that it is user error.

But I do say, that a relatively rare user error and an even significantly rarer non-user error. that it's a nothing burger. 

We have yet to see any evidence of any card failing in a similar matter to the user error failure mode that was not user error, now it would be dumb to say with certainty that there isn't one out ther. But there is not a single shred of evidence pointing to a non-user error failure mode so to assume it exists is just as folly. It needs to be investigated yes, but spreading FUD based off of misinformation is wrong. Don't do that. That does not mean don't discuss it, but when you come out of the gates swinging like this, I can not, nor should anyone take it seriously. 

 

Im going to be potentially unfairly dismissive. Jay does not take any editorial discretion in his normal published videos that he can edit and verify before putting them out there, podcasts in this format are always even worse in terms of being factual just by the nature of it just being guys shooting the shit. And that style of video I am fine with, it can be entertaining, and informative as in making you look at a topic with different perspectives, but they are never something to cite.

Well the card shown in the NorthridgeFix cable mod adapter video looks like it had the connector fitted all the way.

And apparently some AiBs are not repairing burnt connectors under warranty citing the GN investigation that the problem is "user error" which is not covered under warranty. 

 

Not every single person watches GN or other forms of tech news, so how can it be user error when most users didn't even know there's a substantial risk of a not fully seated plug causing the connector to burn? There are no warnings anywhere (except online, which not everybody is going to find)

 

And I wasn't citing Jay as a source, but rather for the point he was making, so I didn't have to type it all out myself. 

[ls612]

This whole issue is making me worried, I have a new 4090 and when I inserted the cable I was extra careful to push it all the way in, leave no gap between the male and female ends of the connector, and I even heard a faint click so it should be latched correctly. But now people are saying even fully inserted cables are at risk. My case has a tempered glass side panel and with it on it still looks like the connector has no gap, just as flush as a normal 8 pin connector.

[Godlygamer23]

2 hours ago, ls612 said:

This whole issue is making me worried, I have a new 4090 and when I inserted the cable I was extra careful to push it all the way in, leave no gap between the male and female ends of the connector, and I even heard a faint click so it should be latched correctly. But now people are saying even fully inserted cables are at risk. My case has a tempered glass side panel and with it on it still looks like the connector has no gap, just as flush as a normal 8 pin connector.

What happens if you pull on it? Does it stay connected?(only do this with the computer off). What kind of cable is it? Is it an adapter? Or is it a native cable? 

 

While most of the situations that popped up might be user error, some of them are also manufacturing defects. 

Edited May 22, 2023 by Godlygamer23

[Slizzo]

10 hours ago, ls612 said:

This whole issue is making me worried, I have a new 4090 and when I inserted the cable I was extra careful to push it all the way in, leave no gap between the male and female ends of the connector, and I even heard a faint click so it should be latched correctly. But now people are saying even fully inserted cables are at risk. My case has a tempered glass side panel and with it on it still looks like the connector has no gap, just as flush as a normal 8 pin connector.

You're fine, there is nothing to worry about.

 

 

[GimmeGaming]

18 minutes ago, Slizzo said:

You're fine, there is nothing to worry about.

 

 

This image would suggest otherwise. This is how the melted connector was fit. As you can see it was seated 100%.

[ls612]

9 hours ago, Godlygamer23 said:

What happens if you pull on it? Does it stay connected?(only do this with the computer off). What kind of cable is it? Is it an adapter? Or is it a native cable? 

 

While most of the situations that popped up might be user error, some of them are also manufacturing defects. 

I’m only using the adaptor which came in the box. When I installed the card I put the adapter in the card before installing the card (the instructions actually recommend this now!) and pushed firmly until I heard a click. It was a bit fainter than the click from an 8-pin connector. Then I yanked and wiggled it a bit and it didn’t seem to move. Only after that did I put the card in the PCI-e slot and connect the power cables to the other end of the adapter. But now people are saying even if it’s in and if it looks straight it can still melt so idk what to believe.

 

EDIT: I just turned off my computer and wiggled the connector again to test. It was firmly seated and did not move when I pulled on it or wiggled it side to side. However it did have a little flex wiggling up/down even when fully inserted and latched. 

Edited May 22, 2023 by ls612
More info.

[starsmine]

10 hours ago, ls612 said:

This whole issue is making me worried, I have a new 4090 and when I inserted the cable I was extra careful to push it all the way in, leave no gap between the male and female ends of the connector, and I even heard a faint click so it should be latched correctly. But now people are saying even fully inserted cables are at risk. My case has a tempered glass side panel and with it on it still looks like the connector has no gap, just as flush as a normal 8 pin connector.

And you are the prime example of why reporting on the issue needs to be handled with care, ethics, and editorial discretion. 

1 hour ago, GimmeGaming said:

This image would suggest otherwise. This is how the melted connector was fit. As you can see it was seated 100%.

God damnit dude. IS that what that image suggests? No other possibilities? None at all that you are capable of thinking of?





It's literally cable Mod eating crow for good PR

These failures that are not user caused are not shown to be any more common than normal melted PCIe power cables. 
Like this

or 

Or this

 

Be AWARE of Recency Bias and Selection Biases existing (like how even though violent crime is down by magnitudes over the last few decades, because its in the news people think it's more rampant than it really is/historically was because reporting on it sells ads, not even a malicious thing, just context and framing are missing)

Northridge fix going, oh shit there is a large influx of this issue, is because people with that issue, and cable mod, dumped that issue onto him when they learned he exists and is capable of fixing that issue rather then it being distributed more evenly to RMAs or other repair shops like it would in the past. AKA a form of selection bias he has no control over.

Im not saying dont look into, im saying be more aware of the claims you are making

[GimmeGaming]

I'm not saying its a 1 in a million problem or a 1 in a billion problem. I'm saying its a problem as I seen one perfectly seated connector that failed. Before this would have been classed as use error and if it went to the vendor they may not have honoured the warranty as they have a get out of jail video with Gamer Nexuses video and Nvidia piggy backing off it. Luckily Cable mod did honour it in this case.

 

You could use your selection Bias example the opposite way around also. Gamer Nexus replicated the issue with the connector not fit correctly so it must be all are caused this way. (They even pointed out that the above card had heat lines that suggested it could have been out which means their testing was flawed as even a fully closed connector caused that heat line based on their last video.)

 

Its to do with that one specific card that is the big problem and not the multiple other cards as luckily it melted and seized in place so you could get a perfect shot of how it failed while fit correctly.

 

Also if a vendor is not going to honour warranty on a melted card, how are they going to differentiate between user error or a card failure? The line that Gamers Nexus pointed out which Gamers Nexus also said was on that perfectly seated card?

 

 

 

[ls612]

I have attached some close up pictures I took of the cable to document how it is plugged in. It certainly looks like there is no further in it can physically go and the latch is engaged. Is there anything else I need to beware of with how it's plugged in?

20230522_162745511_iOS.heic 20230522_162806854_iOS.heic 20230522_162853175_iOS.heic 20230522_162912254_iOS.heic

[Godlygamer23]

3 hours ago, ls612 said:

I have attached some close up pictures I took of the cable to document how it is plugged in. It certainly looks like there is no further in it can physically go and the latch is engaged. Is there anything else I need to beware of with how it's plugged in?

20230522_162745511_iOS.heic 1.18 MB · 0 downloads 20230522_162806854_iOS.heic 606.81 kB · 0 downloads 20230522_162853175_iOS.heic 1013.54 kB · 0 downloads 20230522_162912254_iOS.heic 984.26 kB · 0 downloads

The photos are in the HEIC format, so only people with that codec installed will be able to view them. 

[ls612]

Yeah sorry about that idk how to get my phone to save them in a different format.

[Godlygamer23]

2 hours ago, ls612 said:

Yeah sorry about that idk how to get my phone to save them in a different format.

You'll either need to convert them or go to Settings>Camera>Formats and retake the photos. 

[ls612]

https://imgur.com/a/6G2WL8t That should work I think.

 

Also, another user on reddit posted about another connector melting; 

 

[Montana One-Six]

21 hours ago, GimmeGaming said:

I'm not saying its a 1 in a million problem or a 1 in a billion problem. I'm saying its a problem as I seen one perfectly seated connector that failed. Before this would have been classed as use error and if it went to the vendor they may not have honoured the warranty as they have a get out of jail video with Gamer Nexuses video and Nvidia piggy backing off it. Luckily Cable mod did honour it in this case.

I'm not a fan of the 12VHPWR either but saying it is only Nvidias fault is untruthful. There are more than two possible reasons to why that happened. It could be neither Nvidias fault nor user error but something else like a manufacturing defect on the cables/adapters.

 

I also prefer 8pins over 12VHPWR but there are also cases of 8pins melting in the past. Since 12VHPWR is rated for much higher power draw the failures are just more catastrophic than that of 8pins which makes for much better clickbait and headlines.

 

On 5/22/2023 at 12:26 AM, Alvin853 said:

Also there were no widespread issues reported on 3090Ti cards that used the same connector.

Which is fairly easy to explain why. It's a simple numbers game. The amount of 3090Tis sold vs 40series sold. There are a lot more 40series in use than 3090Tis so there will naturally be tons of more cases of 40series.

 

On 5/22/2023 at 4:53 AM, Alvin853 said:

Well the card shown in the NorthridgeFix cable mod adapter video looks like it had the connector fitted all the way.

Which realistically doesn't tell you anything other than it was fitted all the way when that dude looked at it/filmed it. To rule out any user error you would need to know how it looked like installed during failure or would need some failure analyst to examine it thoroughly. Even if it was connected all the way, who says it wasn't an issue with cable mods adapter?