PC powering down in some games

[strumpusplunket]

My PC has been powering down on its own, but only in some games.

It acts just like the power has been turned off at the wall, and I need to wait a little while and cycle the power switch off and on before it will power up again.

This has only happened in three games, though:
Pillars of Eternity II (only once, in a scene with lots of sigils and lots of enemies on screen at once)
Enderal: Forgotten Stories (again, only once, in the crystal forest place)
Operencia (three times now, latest one being just before I posted this)

I had originally put this down to a likely thermal shutdown. I'm only using the stock cooler on my 2600X, and I could well imagine the phsyics in that PoE II scene, with lots of effects firing off at once, was stressing the CPU.

Oh well, I thought, it's only happened once in that one scene and I am planning to get a better cooler anyway at some point.

In Enderal I felt less confident it was CPU thermals. It's not a new game and I can't imagine it being that demanding. But again, it only happened once and it was in an area with a lot of particle effects, so maybe.

But Operencia? I can't imagine this is a very stressful game for the CPU.

I have tried monitoring my temperatures in these games. I've seen the CPU temp sitting regularly at 74C when coming out of a game, which is fairly hot, but not in the thermal shutdown range. On powering back up after the last shutdown, it was reading 66C in the BIOS. Still hot, but not troublingly so.

Of course, I don't know if there are some effects causing higher spikes in temperature that I'm not monitoring. I guess that's possible.

However, with this system I've played many more demanding games that have run fine (27 hours in RDR2, recently played through Rage 2 and Metro Exodus with no problems at all). Why does it seem to want to power down in Operencia of all things? Three times now. This can't be a conincidence, surely?

Any other potential causes than thermal shutdown? Any diagnostics I should run?

 

[StarShock]

Could be the PSU. I'd also want to double check that the cooler is running properly and mounted correctly, 66c is pretty hot in the bios even for the stock cooler.

 

[strumpusplunket]

Could be the PSU. I'd also want to double check that the cooler is running properly and mounted correctly, 66c is pretty hot in the bios even for the stock cooler.

Thanks for the reply.

I should clarify that the 66C in BIOS is only right after powering up after a shutdown, so it will still be cooling from whatever peak temperature it reached I guess. In Windows it idles at 37C, which seems about right. I'm fairly confident the cooler is mounted okay.

I just ran Operencia for a little bit. Only five minutes or so, but even then alt-tabbing back to HWmonitor I could see the temps getting up to 74C. I've never seen it go above 74, but I tbh I haven't set up an OSD yet. I'm just doing that now with HWinfo.

I did wonder about PSU too (it's a Bitfenix Formula Gold 750W, so it should have plenty of headroom for my build, but you never know). It does make me wonder why it can play games that stress the graphics card much more intensely for hours without issue, though.

 

[EsaT]

Check Event Viewer and Reliability history for any errors which would be close to time of those shut downs.
That might give some hints to cause.

 

[strumpusplunket]

Check Event Viewer and Reliability history for any errors which would be close to time of those shut downs.
That might give some hints to cause.

No, there's nothing enlightening there. No critical events before the shutdown.

I just ran Operencia for about half an hour with OSD monitoring. It didn't shutdown, but I saw CPU temps peak at 82C. This is pretty hot, but still must be 10C or so off the thermal shutdown limit, I'd imagine(?). It doesn't seem impossible that a spike a bit higher might reach the shutdown threshold, but should't the chip throttle to prevent this?

(By comparison, half an hour of RDR2 saw the CPU temps much more slowly worked their way up to 70-72C and stay at that level)

I've looked in the BIOS but I can't see what the thermal shutdown limit is on this system. Can't find it by Googling, either. Where is it hiding?

 

[strumpusplunket]

An update is relevant here.

I didn't have this problem again since I posted last time in this thread, despite doing a lot of gaming over Christmas. Well, until I decided to play Mass Effect: Andromeda the other day. Then, once again, I found my PC shutting itself down. This time it was more conistent - it happened three times, and all within 30-45 minutes of starting the game.

Some more thorough stress-testing let me see that whilst my CPU temps can get alarmingly high, they don't cause a thermal shutdown - even at temps 20C higher than I ever see in-game.

I decided to open up the PC and have a look around inside again, and I noticed that I'd connected both 8-pin PCI-E connectors on my 1080Ti to the same lead and thus rail from my PSU, when two are available. Looking at my PSU, each 12V VGA rail is rated for delivering 35A, so 420W. I would have thought this was plenty. With more hope than expectation, I changed this so that each connector was powered from a separate 12V rail.

Well, the result was that I was able to go from a maximum of 45 minutes or so in the game before having a shutdown to playing for five hours straight with no problems at all.

It's looking like this might have been the problem all along, which may explain why it only happened so infrequently. I guess the 1080Ti is capable of drawing very high loads under certain conditions and triggering the OCP shut-off. I guess also that ME:A seems to cause these conditions more often than any other game (though it does fit also with the shutdowns in Enderal and Deadfire at times when sudden graphical spikes ocurred in certain areas).

Touch wood that that's it, anyway...