F@H seems to be causing PSU power surges?

[Xanthelei]

Hi all, brand new to folding. I have a decent PC that has had no problems with everything from idling for weeks on end to streaming while playing rather intensive games. So I was pretty surprised when I was woken up by my PC restarting night before last, and being greeted with a message saying my ASUS motherboard had detected a "power surge" and shut down to avoid damage. While I'm glad it did so, it's the first time I've ever seen something like that happen. I had only F@H active, with stuff like a few Brave tabs and Discord open in the background, but definitely nothing intensive. I had cranked F@H up to max for while I was asleep, so I figured I'd drop it down to medium and it should be good, but after about 10 minutes it restarted again for the same power surge reason. I'm about to test low, and will add that result after a while.

What I did was shut down F@H for a day and do some stress testing of my stuff, and haven't been able to reproduce the problem. I have an ASUS H170-PRO motherboard, an Intel i5-6600K 3.5 GHz that I haven't overclocked, 16GB 2133 MHz RAM that's XMP'd to 2400 MHz, a GTX 1660 Super, and an EVGA 750 BQ 80 Plus Bronze PSU. The system itself is about 5-6 years old with the only upgrade being the GPU and refreshing thermal paste every so often, so the config is definitely known to be stable. And I've only had the power surge message from F@H, and only twice. Is there a setting in the software I should be looking at that could be causing something to try to draw too much power? I'm hoping to avoid having to swap out the PSU since cable managing in my current case is a nightmare, lol.

Thanks for any help and let me know if you need any more information.

[toTOW]

Nothing seems wrong with your set up : you are not overclocking and the PSU seems big enough to handle what you have installed in your case.

I don't know what "power surge" means in the error you received. Do you have some BIOS settings that could be related and that you could try to disable ? Do you have some documentation about this error that could help us find what caused it ? Is it a surge in power consumption that has been detected, or in voltage supplied to the motherboard ?

If you want to relieve your PSU for testing purposes, you can try to run only CPU slot or only GPU slot alternatively and see if it can run overnight without issue.

[bruce]

FAH does attempt to use ALL of the resouces you're willing to donate so you can expect your GPU to run continuously at close to 100% and your GPU, as well -- although you can adjust those settings. (That should only present a problem if your system's cooling is inadequate.) I wouldn't use the word "surge" to describe that condition but who knows what a bot will choose to report or what words it might use.

[Xanthelei]

I did a quick google search and found an image of the error message I receive. Not sure it's going to be helpful, since it doesn't give much more info, but here it is:

https://images.computerfrage.net/media/ ... 7798649000

Edit: I do want to add that the message is the same despite the PC having been on varying amounts of time. It doesn't mean "during boot up" just "last powered on session" I guess.

I've checked in my BIOS and the only setting outside of default I have is my XMP setting for my RAM, but I can't imagine going from 2133 to 2400 would cause a brand new problem like this. I did do the testing I wanted and dropped things down to low settings, but then the GPU wouldn't fold at all. Despite having it set to always work, it showed the message "Paused: waiting for idle" for the GPU (CPU can't find a WU it seems, no biggie there). I tried all the usual troubleshooting I found on the board, and none of it got the GPU to do anything, so out of curiosity I changed it back to Medium. I immediately heard my fans ramping up (expected) and then got the same crash. So it's definitely tied to the GPU drawing power in some way, and it's definitely tied to F@H getting the GPU to do it's thing. The odd thing is, I'd folded two or three other WU at both Med and High settings with no problems aside from temperature, which in turn was fixed by moving my PC further from the wall. So I'm at a loss here.

Interesting thing was, when I rebooted F@H came up automatically and tried to start folding again, causing another crash before I could do much more than open the menu on it's icon. I didn't realize it was set to do that, so I had to open my Task Manager and watch for the program so I could kill the process before it caused another crash. The biggest problem is I don't have a way to change the setting or pause the WU without it crashing again. I'm guessing I'll have to reinstall the program?

I'm going to hunt for some logs from F@H, see if there's anything there that might be helpful. At this point my only guess is it's something with the WU itself, since this only started on this one. But I did see I'm 15% of the way into the WU so I couldn't really say for certain that's what it is.

Thanks for the suggestions on what to check and how to stress test it. This has been interesting to say the least!

[bruce]

"Paused: waiting for idle" depends on where you've set the Power slider. If you have a strong GPU, you don't need that feature, but if your GPU is weak, folding while using your computer can cause annoying screen delays so FAH's GPU processing is set like an invisible screen-saver, pausing whenever you use the mouse/keyboard.

In addition to resetting the power slider, You can turn that feature on/off with FAHControl.

[foldy]

FAH is more demanding on hardware than games.

I could imagine that maybe the mainboard has some extra 12V 8pin plug for power supply which is not plugged in? Or the GPU is not connected properly to power supply or a cable is broken hiddenly? In both cases CPU or GPU try to get all power from mainboard which will shutdown if overloaded showing this BIOS error message.

Maybe you can try to disable FAH CPU slot and run FAH GPU slot only or vice versa to see which one brings the overload.

Or use tools like MSI Afterburner to limit power usage of GPU.

[Xanthelei]

Bruce: I didn't have any issues with it on Medium while using my PC, I'd only turned it down in case it affected how hard it hit my GPU and only after the first crash. In this case, it seems it just didn't hit my GPU at all. Wasn't quite what I wanted. The current issue is it starts folding immediately on start up, which in turn causes an immediate crash. So I can't have the program running long enough before the crash to make any settings changes.

Foldy: I double checked that my GPU still had it's power cables firmly seated, and it does. So it certainly shouldn't be trying to draw everything through the PCIe slot! I also can't see any missing header connections for the mobo itself at the moment, and it hasn't had this issue in the past at all. I also don't think it's a problem with the CPU at all, since I couldn't get a WU for my CPU before moving the slider back to Medium, but had a WU in place for the GPU that it started folding right away just before the crash. At this point, I'm going to load up a game and do a benchmark that'll hit the GPU and see if it happens in general, or if it's just with F@H. That's the only other way I can think of at the moment to rule out a hardware or driver problem in general vs something with the F@H program/WU.

[Xanthelei]

Well, I just ran Furmark and I think it is the GPU itself in some way. It didn't even last three minutes before shutting down again. Though I think my mobo has given up warning me about power surges, I had to do a manual shut down/power on to get it going again.

So now the only question is how to get F@H to run only CPU for a while so I can get started on an RMA for my GPU. Any ideas how to do that without reinstalling or causing another crash? I'm not sure how much time I'd have from when I could access the menu to when it would crash, or even if it would register a pause before it did so.

[toTOW]

I googled the error message you posted before, and I found that :
- you can disable this feature in the BIOS, but you might experience other symptoms if the PSU is bad. I guess that's what you're seeing with your GPU which might not be bad ...
- most people solved the issue by replacing the PSU. It is not a bad PSU, but usually ageing ones that are starting to deliver unstable voltages.

To run FAH only on CPU, you can pause the GPU slot or delete it.

[rwh202]

So now the only question is how to get F@H to run only CPU for a while so I can get started on an RMA for my GPU. Any ideas how to do that without reinstalling or causing another crash? I'm not sure how much time I'd have from when I could access the menu to when it would crash, or even if it would register a pause before it did so.

Unplug / disconnect your network
Install
Advanced control > configure > slots
Delete the GPU slot
Save
Connect network
Fold!

[Sven]

The system itself is about 5-6 years old with the only upgrade being the GPU

I would say here we have it. Your PSU is 6 years old and you're using a current gen GPU. The problem is not about the wattage of your PSU, but a 6 years old PSU is not prepared for the fast load changes newer graphics cards produce.

I would say a good actual PSU will solve your problems.

[treckin]

PSU is definitely the issue, your MLB is actually telling you as much.

One significant difference that I havent seem mentioned between FAH workloads and gaming/rendering tasks such as furmark:

Your CPU is using AVX istructions to fold, which use much more power at the CPU. The CPU is most sensitive to voltage fluctuations.

Your 1.1v rail is likely drooping during high instantaneous load produced by the AVX instructions, causing your board to auto-protect.

If you want to verify, try running Prime95 with AVX/torture test settings and see if its reproducible.

To be safe, run Furmark simultaneously to try to replicated the load conditions on the PSU.

FYI a bronze rated PSU was never that high quality to begin with, as the lower efficiency standard means they can really cut corners and for the price point, customers are generally OK with it.

Try to stick with Gold/Platinum rated PSUs and remember to spec them such that your load is about 50% of the rated capacity of the PSU, otherwise you will be inefficient, generate excess heat, and kill the PSU quicker.

Good luck!

[wayne_j]

I'm getting the same issue with my 1660 super but I have a seasonic X-series gold 1250. machine runs fine if I have the gpu unplugged and im running the onboard but once the gpu is plugged in and im using that it'll shutdown after a few minutes even when browsing the web.

any help would be appreciated.

[Ichbin3]

Try to use your gpu with a little less power
At cmd use this:
nvidia-smi #it shows you the max power usage in W
nvidia-smi -i 0 -pl 100 # max power usage is 100W
You can increase this as long it is stable without loosing a lot of performance or decrease it down to 70W

[foldy]

Maybe one GPU power coord of power supply is broken. Try to use another one.