earlier today I wrote:

I am trying now various methods to "delimit" the problem:

- right now, I am crunching SETI@home tasks, so I'll see, whether the problem occurs also there. If so, then I might install

- Folding@Home, which is working with OpenGL (in contrast to GPUGRID and SETI, both of which work with CUDA).

If the problem persists in both of the above cases, then I will revert back to Windows XP (on the same machine, with dual boot) which I have used for the past 2 1/2 years, and see,if the problem also occurs then.

In case it does, then I am afraid that JoergF may be right when assuming that there may be some kind of hardware failure :-(((

I had run first SETI and then Einstein for several hours now, no failure occured.

Hence, a minute ago I changed from Win10 to WinXP and now crunch two GPUGRID tasks overnight, plus a few LHC tasks (CPU only), as I used to do it for long time.

I am curious what I will see tomorrow morning.

I had the same problem, it was a power saving issue with the BIOS and Windows 10

How did you resolve it?

It had something to do with multiple cards and UEFI BIOS, if your monitor was set to turn off after a time limit mine wouldn't come back on. I turned off that feature untill a new BIOS came out, that and the "Fall creators update" fixed it for me.

It seems to me that update and the April 2018 update fixed alot of issues with Windows 10, I know they never mention everything that is fixed because of interdependencies.

I turned off that feature untill a new BIOS came out, that and the "Fall creators update" fixed it for me.

which means: in the Windows Energy settings you switched "turn off monitor" to "never" ? So your monitor was on 24 hours per day?

Also: did the problem occur only when crunching GPUGRID? Or anytime else?

When ever I was using all the cards, I dug into it so far as it had to do with your utilization being high and no SLI bridge being on or present.

My motherboard was like shaking down a battleship, all kinds of problems.


Edit: When I switched to Windows 10 I hated it, the looks, where stuff was. Then I remembered reading about a small applet called Classic Shell that gave you the option of a Windows 7 or XP start menu with desktop icons instead of those stupid boxes.


It has a ton of options like Windows explorer all the menus and pathways once I installed it I was very happy with Windows 10. It would even create backups of the changes I made and ask me if I wanted to restore my changes after installing those major Windows updates.


Anyway, for those people who will be forced to switch 10, it's a pretty cool alternative.

I turned off that feature untill a new BIOS came out, that and the "Fall creators update" fixed it for me.

which means: in the Windows Energy settings you switched "turn off monitor" to "never" ? So your monitor was on 24 hours per day?

Also: did the problem occur only when crunching GPUGRID? Or anytime else?

Only if you leave the monitor on. I just turn off the monitor.

Couple of things, find energy profiles. Change it to Max performance.

I turn off Screen saver, switch power off monitor to never, screen to blank never, disable spin down hard drive (if you are still using one) vs an SSD

Make sure that no where you have put machine to sleep after x minutes of no use.

Those error messages you are getting are coming from when the computer shuts down the GPUs and the drivers crash. You need to figure out why the GPUs are being put to sleep.

As far as keeping Einstein from using both CPU and GPU, that is in your preference setting on your account page ---->Preferences--->Project. select a location, change use CPU to no and use Nvidia to yes. Save and then make sure the location of your computer is correct.

---

... and no SLI bridge being on or present.

this reminds me - after re-installing the NVIDIA driver yesterday morning, on the right lower corner of the screen I got some kind of warning that the SLI bridge is missing. I don't remember whether the warning came from NVIDIA or from Windows.
Whether this (also) has to do with my problem or not - no idea ...

Anyway, all last night I crunched GPUGRID and LHC on Windows XP, no problem at all. It's still running fine.

Conclusion:
1) there does not seem to be any defective hardware
2) the problem clearly has to do with Windows10, and maybe (although I am not 100% sure yet at this point) only when GPUGRID is running.

So today I will take a closer look into the energy savings settings, and I'll again run either Einstein or SETI tasks for a lenghty period of time.

You don't want to run your GPU's in SLI mode while crunching WU's, it doesn't cause problems to have the bridge on. Just make sure SLI is turned off while running GPUGrid.


I'm not saying this will fix your problem, but I'm pretty sure it is a power saving issue with your monitor\GPU's.

...but I'm pretty sure it is a power saving issue with your monitor\GPU's.

as I said, I need to do more testing with SETI and/or Einstein.

But if my first impression from yesterday's testing is right, then the problem might come up only when crunching GPUGRID, and NOT when crunching the other projects.

But this is not quite sure yet, probably I'll know more during the course of today.

... probably I'll know more during the course of today.

well, by now it seems pretty sure that the problem comes up only when crunching GPUGRID, but not with other projects like SETI and Einstein.

I saw an interesting article in the Anandtech Forum:

https://forums.anandtech.com/threads/gpu-tasks-are-causing-win10-machine-to-become-unresponsive-restart-fixed-by-disabling-sli.2526566/page-2

under the headline "GPU tasks are causing Win10 machine to become unresponsive/restart (fixed by Disabling SLI)"

Well, on my machine, SLI is NOT enabled, anyway.
What I am trying right now is to crunch with only 1 GPU instead with 2. So let's see what will happen.

You posted earlier that "I got some kind of warning that the SLI bridge is missing."

That can only have come from the NVidia driver. It sounds like SLI support is enabled in the driver, but not in your current hardware configuration. I'd turn it off from the 3D settings page in the NVidia Control Panel.

... I'd turn it off from the 3D settings page in the NVidia Control Panel.

I looked it up now - it's deactivated there.

What I am doing right now is to crunch only with 1 GPU instead of 2. Let's see what happens.

My understanding is that NVidia cards are designed to power up using the 75W available from the PCIe slot, detect that the additional power cables are unconnected, and refuse to move out of a protective low-power state.

Thus, in a different state from total removal. Possibly safe, but nor very informative.

Depends on the age of the card or family type. Don't try that trick with a Kepler card. I forgot to plug in the PCIe power connectors for the dual 670's and when I turned on the computer, both fans raced screamingly to full rpm's for a few seconds and the computer promptly shut down. Thankfully no damage done but scared me silly.