Windows : Kernel 41 error (70368744177664),(2) while folding only

[pyrocyborg]

Hi there!

We have ten PC with Windows 10 and 11 here GPU folding (GPU folding only, I do not fold with the CPU), and all of those with Nvidia RTX cards will crash after a random amount of time. It could be after one hour, it could be after 3 weeks : there is no constant, except the (70368744177664),(2) error in the Windows event log. Sometimes, the rig will simply stop and shutdown, sometimes it will reboot instead. Those with AMD GPU (RX 6xxx) do not get that reboot/shutdown symptom as often.

The only things that are similar between these builds are that they are fairly recent (Ryzen 3xxx, Ryzen 5xxx, Intel 11th and 12th gen), under Windows and have recent graphics card (RTX 3000 and RX 6000). They are all fairly well powered with quality power supplies, and none of these builds have the same power supply model either, so I don't think it's the issue.

These PC almost never crash except while Folding. I know that Folding is ressource intensive, but why are these fairly different builds failing at random while folding only?

Did anyone get a similar issues and found a solution? None of the (70368744177664),(2) solutions I found on the Internet worked, even those that recommended to change the power supply to get rid of the error or to reinstall Windows.

Thanks!

[JimboPalmer]

No one seems to have any experience with it, so can I recommend the latest drivers right from Nvidia?

https://www.nvidia.com/en-us/geforce/drivers/

[pyrocyborg]

Already did that. It happens whatever the driver, even after doing a clean install or using DDU. It mostly happens after a few days of activity, but it sometimes happen a few times in a row (like, after 30 minutes, then again 1 or 2 hours after that, then everything is fine for 2 weeks).

It's not that much of a big deal, but the randomness and the fact that multiple PCs see this error is kind of odd (even the few PCs that are never used by anyone and are folding 24/7 sometimes get that error). I can't say whether it's a BSOD or not, as I'm never there when it crashes (I'm usually out of the house or sleeping when it happens). None of the PCs are configured for deep sleep (or sleep at all), so it shouldn't be the issue either.

I'll continue investigating.

[beaterhellfire]

Have you overclocked your card?

I had the same issue about a month ago and it turned out F@H cores don't like it why you have your GPU overclocked aggressively (in my case +800 Vram and +165 core clock ). After putting my card back on stock settings with only +200Mhz Vram and +50 core clock, everything works fine i have had zero crashes since. (I am running RTX 2060 super)

[toTOW]

This event usually happen after a blue screen ... it should contain the Bug Check code the will be useful to better understand what is going on.

In the event, you should see something like this :

Code: Select all

EventData  
BugcheckCode 159  
BugcheckParameter1 0x3  
BugcheckParameter2 0xfffffa80029c5060  
BugcheckParameter3 0xfffff8000403d518  
BugcheckParameter4 0xfffffa800208c010  
SleepInProgress false  
PowerButtonTimestamp 0Converts to 0x9f (0x3, 0xfffffa80029c5060, 0xfffff8000403d518, 0xfffffa800208c010)

Can you tell us which BugcheckCode you get ?

Anyway, this is the sign of unstable system, so you'll have to proceed to the usual verifications (CPU, RAM, GPU, PSU).

Have you overclocked your card?

I had the same issue about a month ago and it turned out F@H cores don't like it why you have your GPU overclocked aggressively (in my case +800 Vram and +165 core clock ). After putting my card back on stock settings with only +200Mhz Vram and +50 core clock, everything works fine i have had zero crashes since. (I am running RTX 2060 super)

Do not overclock VRAM for FAH, it adds more troubles than it helps on modern VRAM technology (because of ECC mechanisms that will slow down the bandwidth instead of producing errors).

[pyrocyborg]

Have you overclocked your card?

I had the same issue about a month ago and it turned out F@H cores don't like it why you have your GPU overclocked aggressively (in my case +800 Vram and +165 core clock ). After putting my card back on stock settings with only +200Mhz Vram and +50 core clock, everything works fine i have had zero crashes since. (I am running RTX 2060 super)

Nah, I don't overclock at all as it doesn't yield any sensible gains and might even cause bad work units according to what I read.

My graphics cards are stock or under a power limit as I don't mind losing 10-15% PPD for a 35-50% reduction in power. They usually work fine, but there is that odd crash/BSOD/restart with that generic error that only seems to happen while folding.

[pyrocyborg]

This event usually happen after a blue screen ... it should contain the Bug Check code the will be useful to better understand what is going on.

Can you tell us which BugcheckCode you get ?

Anyway, this is the sign of unstable system, so you'll have to proceed to the usual verifications (CPU, RAM, GPU, PSU).

More often than not, there is no BugcheckCode as the PC seems to hang and reboot without a BSOD. I'll make sure to copy-paste it once it happens.

I already changed the PSU on multiple units since then, and the same goes for RAM and SSD. Did a full reinstall in 5 cases (as I changed the SSD) and it still happens, whether it's under Windows 10 or 11. No error on any the RAM sticks after a few hours of test, so I don't think it's the issue per se. I'm currently entertaining the idea of loading F@H on a Linux distro, but I have no idea how to do it (haven't touched Linux in the last 30 years) and whether it would correct the issue or not.

I cannot say for the CPU though, as it's kind of hard to swap it around. What's odd is that it's happening on multiple systems with very different components and I doubt that all our graphics cards (RX 6000 and RTX 3000 under two years old) are failing. It obviously doesn't happen at the same time, but it's happening on very different configurations and seems to happen a little more often lately (since a week or two), but it most probably is a coincidence. It could even be a power issue in the building, even if everything is up to code and almost brand new.

I'll keep an eye opened for bug check.

Thanks!

[toTOW]

Can you try to set the power management profile to "Prefer maximum performances" in nVidia Control Panel to see if that helps ?