Launch 4090 problems with Pimax 8KX

[Heliosurge]

Actually you have to look at it as 18 publically confirmed. Not everyone whom has had the issue will post it on social media. PSA should not be looked at as clickbait; but as a warning to take care with new Tech like this power connector.

With currently only Nvidia utilizing this new power connector does make it harder in some respects to isolate whether it is an issue on the cable connector or a manufacturing issue on the cards themselves. Either component could have defects being shipped.

I suspect it may have to do with tolerances of the connection point are either not tight enough or being checked often enough to ensure consistent quality.

Imho people should do stress testing for a long duration while monitoring the connector temps. Perhaps even do several of these stress tests with performing removal visual checks.

Since this issue can indeed cause a fire no one wants to discover they are in the low % of potential cards that might have this defect.

This is why it is often best not to be first in line for new tech until it has shown it works well in the wild.

[Djonko]

@Sargon

This isn’t surprising in the slightest. We’re going to see examples of failed 12vhpwr connectors of every kind. There has never been a power connector in the past that has had a 0% failure rate. We should not expect a 0% failure rate from the new power connector either.

Nah, Like Helio just said, the actual number is always going to be higher than the reports on Reddit. But even if it was 0.018%, then we as a world would have a big problem if 0.018% of any electrical device would melt/burn down. Of course that’s not in any way a tolerable amount. It’s bad, almost to the point that you’d expect a mass recall, like Samsung did when some of their users reported exploding batteries. I also had that device and I remember that I first felt like you, that it was all just overhyped and I didn’t feel like sending back the phone. But then you start thinking, what if it happens at night while you’re sleeping. You just don’t want to run such risk.

Personally I will still use my 4090 and don’t worry too much. But man I will double check every time I shut down my gaming PC that it really is powered off before I leave …

[Sargon]

@Heliosurge

Not everyone whom has had the issue will post it on social media.

There was a huge amount of media hysteria about this even before the first failure was ever reported. The public was primed for panic about this. So while I would usually agree with you, I think in the current climate that nearly every individual failure is getting posted.

@Heliosurge

PSA should not be looked at as clickbait; but as a warning to take care with new Tech like this power connector.

That’s the insidious part about it. In some aspect, it is legitimately a PSA. There is a legitimate concern. It’s not my point to handwave it away, but rather to point out that it is being overblown for clickbait. There is a huge amount of disinformation and hyperbole surrounding this issue, and only a small core of truth at the center.

@Heliosurge

I suspect it may have to do with tolerances of the connection point are either not tight enough or being checked often enough to ensure consistent quality.

In the absence of answers from actual engineers working on this issue, the public has been filling the void with its own unqualified speculation like this. Most of it has been drawn from the leaked email to speculate that it must have to do with bending the cables somehow, even though the leaked email described a problem that was already eliminated in production units and was meant as a warning to other manufacturers to make sure their units also don’t have this problem before release. These kinds of speculations keep being repeated until they are “known” by the public as facts, and it is very misleading.

What actual testing has been done that we know about (such as the video from GN) demonstrates that it is not easy to get these new power connectors to melt even when setting up cases which are trying to cause that outcome intentionally. There is not an obvious design flaw to explain the small number of melted connectors that have been reported.

Consequently, any speculations coming from all the armchair engineers out there are just flat guesses which are probably wrong. These are not speculations based on testing and data. Any one of them could be right, but it would be just by luck.

@Heliosurge

Imho people should do stress testing for a long duration while monitoring the connector temps.

I agree with this. But I would say don’t just monitor the new 12vhpwr connector. Monitor all of your power connectors for a while after any new installation. 8 pin PCIE power connectors melt sometimes, too. PSU cables fail sometimes. All you have to do is touch the various power connectors and power cables while they are under load to feel for excessive heat. This is good practice in general.

The 12vhpwr connector currently has the media magnifying glass focused on it which is amplifying the appearance of a high failure rate. We do not know right now whether it is actually failing more often than other power connectors.

@Heliosurge

Perhaps even do several of these stress tests with performing removal visual checks.

I don’t agree with this. Actually disconnecting the power connectors is not necessary and could actually induce failure. Again this applies to all power connectors, not just the new 12vhpwr connector. Internal power connectors (all of them, not just the 12vhpwr) are not rated for a large number of mating cycles like a USB cable. Plus each time you reconnect it, there’s a chance that you won’t seat it properly. You could test it and see that it’s fine, and then leave it in a condition where it is no longer fine.

@Heliosurge

Since this issue can indeed cause a fire no one wants to discover they are in the low % of potential cards that might have this defect.

The actual fire risk is low. If the power connector fails and melts, there’s a lack of material which could reasonably be ignited in the vicinity (by design). What’s it going to light on fire, exactly? The giant aluminum cooler? The metal or glass side of the case? I’m not saying the fire risk is zero and not a concern at all, but just again pointing out that the fear factor is being hyperbolized.

Please don’t keep your PC case open with your can of kerosene leaning against the GPU.

@Heliosurge

This is why it is often best not to be first in line for new tech until it has shown it works well in the wild.

I do agree with that. It’s rare for me to hop on to a new product at launch for this reason. But I did this time, and so far I haven’t regretted it.

@Djonko

But even if it was 0.018%

Note that this is just napkin math. Just some rough numbers to point out that this is a rare problem, not a 10% type problem. We don’t know the actual failure rate and how it compares to prior power connectors.

To be sure, I think there likely is a problem, but that the scope of it and the danger isn’t nearly as large as the media is pumping.

@Djonko

like Samsung did when some of their users reported exploding batteries.

That was a much more dangerous condition which was very capable of causing fires since smartphones could be and regularly are in the vicinity of easily combustible materials, and also because the failures were rather more energetic than we’ve seen from melted 12vhpwr connectors.

Unless there’s some new development that increases the possible danger, I don’t think we’re going to be seeing (the need for) a safety recall on 12vhpwr connectors.

@Djonko

Personally I will still use my 4090 and don’t worry too much. But man I will double check every time I shut down my gaming PC that it really is powered off before I leave …

I do keep checking my adapter for excessive heat or any other signs of trouble. I doubt there will be trouble, especially since I’ve been running it for a while now without any problems. But I will remain vigilant anyway until we know more. An extra degree of caution is indicated.

[Heliosurge]

@Sargon

@Heliosurge

Not everyone whom has had the issue will post it on social media.

There was a huge amount of media hysteria about this even before the first failure was ever reported. The public was primed for panic about this. So while I would usually agree with you, I think in the current climate that nearly every individual failure is getting posted

Your presuming every failure has been public. Something we do not have data to support. Nvidia and partners taking the Fire potential hazard lottery is encouraging.

@Sargon

I agree with this. But I would say don’t just monitor the new 12vhpwr connector. Monitor all of your power connectors for a while after any new installation

This is good advice. If the psu is modular you can check psu connector and card connection.

Atm from known reports it seems to be at the card connection point.

The media is not really magnifying it. It has thus far been a low number. To which @Djonko 's napkin math shows just that. Now you could make it seem larger or smaller using external ppm defect numbers. Depending on if you see the glass as half full or half empty.

[Djonko]

I don’t have the time nor will to engage in these super long discussions about nothing so I’ll just say I agree to this part:

@Sargon

An extra degree of caution is indicated.

[Sargon]

@Heliosurge

Your presuming every failure has been public. Something we do not have data to support.

I have to grant that point. The idea that the percentage of cases being publicly reported would be much higher than usual is just my read on the situation.

@Heliosurge

Nvidia and partners taking the Fire potential hazard lottery is encouraging.

I agree they should take it seriously and seem to be doing so. Each one melted for a reason. Regardless of how many fail, they need to know the causes of failures in the wild and also to assess how much safety risk these failures actually represent.

[drowhunter]

I’ve actually been turning my PC off before bed and when I leave for work most days, prior to this card my PC hadn’t shut down since 2016.

Sucks trying to remote in to my PC and remembering that it’s off.

Looks like my idea of running it with 3 connectors was probably a good one.

I’ll likely drop the power limit to 80% as it makes little difference to frames. But drops the wattage into the mid 300’s

Edit:

So much for my theory that 3 connectors is fireproof.

In the second test, we set up a 450W test load using an NVIDIA GeForce RTX 3090 Ti graphics card. Here, we used the reference adapter that comes with the card, a 12VHPWR to 3x 8-pin connectors. Here, instead of splitting the load equally through all three connectors, a single 8-pin connector is running 23.5A or 282W through it while the rest of the connectors are pulling 6-8A (80-100W). That means that there’s still a safety issue on a single 8-pin connector even if you are using three plugs.

source

[Heliosurge]

New theory

Corsair’s PSU Expert Jonny Guru Weighs in on Nvidia Connector-Gate

[Sargon]

@Heliosurge

New theory

This seems like a good article to me. It’s another theory, but it properly presents itself as a theory with no clickbait or hyperbole. I think it’s a very plausible theory, too.

I particularly like that he talks about bending the cables and trying to reproduce the problems that way not causing any issues. I want to keep repeating that the leaked email was about a pre-production issue that was already solved prior to release. Even now there are still news sources reporting that bending the cables is the problem and that PCIE already knew about it and released the new connector anyway with this problem which is false information. Whatever melted the connectors that have been reported is something else.

Partially seating these connectors is the only way so far that anyone has been able to replicate heat rises in these connectors which could lead to melting.

Note that you can cause 8 pin PCIE connectors to heat up and fail that way, too.

This whole thing could be much ado over nothing. Just some improperly seated power connectors due to user error.

However, we still haven’t heard from nVidia which suggests that the failure analysis is ongoing and that it’s not easy to determine the cause(s) conclusively.

[Heliosurge]

All theories at this point are valid and if using a Fishbone necessary. None of the theories have truly been eliminated as Failures can be a quality issue, Incorrect assembly(as mentioned above).

The adapter cable could have defects not caught during manufacturing process, as we now also have native 16pin Atx 3.0 failure as well.

The dielectic(sp?) grease is not a bad idea as mentioned in the Article and double checking that the conector is properly seated.

[mike4845]

are all you guys running afterburner or precision x1 , ive tried both and when doing so my pimax 8kx runs poorly with the 4090 , once i delete any overclocking software it runs better , but guess i cant overclock the 4090 then

[Heliosurge]

There is another 4090 thread. Some are experiencing issues without any extra programs just running bade Nvidia display manager and driver with no vendor extras.

For some it seems not running things like Afterburner and the like resolves the performance issue while others it does not.

[mike4845]

ok as of now , I thought deleting msi afterburner was the solution but turns out it wasnt , tried my 8kx today and it was stuttering and only getting 50 to 60 fps , soo wierd lastnite it was running great at steady 90fps now today its back to working poorly and having micro stutters almost worse then my 3090 , what the heck is going on
system specs
i9 9900k
32g ram

[Heliosurge]

Can you share the vendor of the card

[mike4845]

msi trio 4090 , not the X
yesterday it was running at 90fps and today its back to stuttering , i didnt change anything weird

[tykey6]

i have that card. i tried the latest drivers which lowered the gpu’s utilization and fps so i’m using the previous drivers for now.

[mike4845]

I’m wondering if my riser cable pci 3 is a problem, I’ve read that they could have latency issues , anyone using these

[dstar]

I use one out of necessity because I haven’t gotten a waterblock yet. But it’s of the better kind. Some riser cables can certainly cause latencies.

There you should also be able to identify bottlenecks via benchmarks.

[mike4845]

Well I think if there’s any chance of latency from rider cables , I’m just gonna find a way to connect it right to the MB , unless there’s a good proven pci 4 riser cable I could get

[dstar]

first i would try without, if it works… this is cheap and ok. https://www.amazon.de/gp/product/B08S6QJ3WC/ref=ppx_yo_dt_b_asin_title_o02_s00?ie=UTF8&psc=1

the same thats Asus sell for more money