PC shutting off constantly...

[CarlosMiguelKlein]

Hello! Recently with Corona 7 hotfix 1 I've noticed a problem that's becoming more and more commom to me, my pc shutting off during render. Me and my teammates have been looking at foruns but couldn't locate a proper solution, it happens after a certain number of passes of after corona reaches a low value of noise, independent of the image size, also it's not because there's not enough ram, since it happens even on smaller images or after a long time in interactive render. I've never had that problem with corona  5, with corona 6 I only had it on renders with "Corona High Quality" denoiser, during denoise time after a rendering completion, and now it became more commom, even with other denoising modes.

Maybe something to do with power supply? Maybe something in my rams? My pc is a 2700x + 32 gb ram ( 2x 8gb 2400mhz and 2x 8gb 2666 mhz), power supply 650w, lot's of space left both in SSD and HD units, temperature is bellow 75 C when it shuts down so no over heat, max is 2021 and I guess that's it. Anyone had the same issue?

[TomG]

Hi! The good news is Corona can't ask your machine to do anything it isn't capable of doing. So the cause here will indeed be hardware as you suspect, either overheating, or power draw I would say (could be the heat or power demand is peaking so fast the monitoring doesn't see it before the machine shuts down). Could be hitting a bad part of memory I guess, but probably not (you can check whether Corona is using close to max memory).

Denoising uses the most memory, and also none of the cores have to wait on each other, so the CPU runs at a genuine 100% of what it is capable of (often times there are microsecond delays while a core waits for data or for another core to do something, which is just enough for some cooling to happen, but not with denoising, it really is at full throttle!)

With heat, bear in mind that it could be a bad connection on the board too, so may not be hitting the CPU thresholds but some part of the board is getting hot enough that a connection or some other chip on the board goes on the fritz (this is suggested by the "it has become more common")

[sebastian___]

You could try opening up the PC case, taking off the cover or the lid, I'm not sure what's the proper word for that in english. And pointing a big room fan at the computer, maybe on max power. And try to reproduce the conditions in which the pc shuts down. In case maybe there's an obscure component somewhere or the ram, or the PSU, which does overheats and doesn't have a temperature sensor to report back to the OS.

If it's still shuts down in the same way, it could be the PSU, which is maybe on the low size, at 650W, depending on PSU quality.

[CarlosMiguelKlein]

Well, managed to fix it! I had 4 ram sticks, 2x 2400 mhz and 2x 2666mhz, removed the 2x 2400mhz and it works like a charm! This only happened in corona tho, possible bug?

[v.p.vlasenko]

I had a similar issue 4x2400 and added 4x2666, but in my case system didn't even start.
It looks like the system set all of them to 2666 and those 4 (2400) or some of them weren't able to work on such speed.
I have memok button on the motherboard and after pressing it all working fine.

In your case it looks like those 2400 is something better and able to work overclocked up to some point but they fail under load.

Obvious solution for you - replace 2400 to 2666.

Or like in my case - set memory speed for all to 2400