Could you do tests with 2px Tile Size in Volumetrics?
My graphics card is a Nvidia GTX 970.
I started getting this problem about two months ago when I upgraded my CPU memory and mobo. Before the upgrade I could render using any tile size in eevee without any crashes in Blender 2.8.
I thought it had to be a hardware problem at first but I have checked all connections and given everything a clean but still the problem persists.
Blender 2.8 is the only thing that causes my PC to crash all the other software works fine without any problems. And it is only when trying to use eevee and volumetrics.
No, my PC has not been overclocked.
Just tried with 2px tile size and as soon as I clicked the viewport render button it crashed. The crash was faster to happen than it was using 4px size.
Very different systems HW wise and even the OS is different. Only common factor then is blender. Might be best off taking this to the bug tracker perhaps.
Wth.
But my pc doesn’t when i use other tiles size other than 4px
You’re using way better pc then mine, still facing similar problem.
I don’t understand wtf.
No it’s not, not even gpu
This result is more logical if this is a hardware problem, since 2px stresses the hardware more than 4px.
The strange thing is that @Md_utsho says to have no problems with 2px size.
According to my tests (at least with nvidia) Eevee consumes much more power than Cycles in heavy scenes. So PSU, power connectors and motherboard would be something you should pay attention to.
Edit:
Just in case it is GPU cooling problem, some nvidia cards have dual BIOS selectable from a small physical switch on the card. One state is Master (quiet/silent) mode where fans try to run at minimum speed, and another mode Slave for Performance where fans run at higher speeds. You find out about it and select switch position for performance mode (you modify the state of the switch with the PC completely turned off and if possible with PSU switch in off)
If your card does not have dual bios, you try to figure out how to manually configure fan speed on Linux (this is not so simple, you should enable coolbits perhaps from xorg.conf)
Today i wasted 2-3h.
Went to that shop after installing another gpu, they couldn’t install gpu driver
It was showing error.
I though something wrong with their gpu
But they kept trying but no luck
And later after coming home i couldn’t install gpu driver. Same error
Well actually windows corrupted after all those restarts.
Now again after installation new windows i can finally can install gpu driver.
But one thing. Like i said psu was fine but now I’m not sure
Those idiots used 450W psu with rx 590 to test it, and still it was restarting.
So I’m not sure if my psu is faulty or not.
I’m testing it again without switching to 4px tiles size.
And no PC isn’t restarting.
I installed Windows again, because of all of thise restart my Windows was corrupted so I had to install Windows again but after selecting 4px tiles size it restarted couple of minutes ago but now I am again testing it with 2px 8px and 16px without switching to 4px and PC isn’t restarting
Edit : read my previous reply,
So you’re saying you were fine before upgrading your mobo and ram? have you tried testing your ram by 1 stick at a time?
and does stress test cause your pc to restart ?
So what mobo did you get?
What is the only thing that turns on and turns off the PSU? Power controller. Ram does not. CPU does not. OS does not. Tile sizes do not. Front panel button does not. Only thing that powers off a computer is the power controller. Other items (OS, front panel button, etc) only “request” a controller to power off. Then that controller (and only that controller) decides whether to honor that request.
Any recommendation that does not only discuss the power controller is misdirected. Often only wild speculation.
Focus only on what is relevant. Othewise nothing useful can become. That explains the so many bogus accusations (OS, Ram, etc) that cannot cause a power off. Ram cannot even request a power off.
How to fix the problem? That also was discussed previously. However long before fixing anything, first the defect (problem) must be defined. That cannot happen is every post does not center on what a power controller does.
Meanwhile, even a number in that kernel error message still is not provided. That number is essential for better (informed) replies. Also stated earlier as being necessary. What is that number? 41?
Informed replies are impossible when essential facts (ie that number) are withheld.
You installed Linux even though the only relevant part (that power controller) is completely separate from the OS. So of course Linux would do nothing. Basic computer knowledge makes that obvious. OS does not control power. Only a power controller does that. Any answer that does not address the power controller and its functions is wasted time.
Power controller was defined previously. Since it was new, then nothing in that post is understood until at least three rereads (as learned in school). Go back to that post to appreciate (learn) the target and strategy necessary for a solution.
Edit: so motherboard was replaced? What did you replace? The power controller.
1 Like
No i didn’t replaced the motherboard, but today i’m going to.
And yes it was Event ID = 41
Also it says “The system has rebooted without cleanly shutting down first. This error could be caused if the system stopped responding, crashed, or lost power unexpectedly”
And i know OS, Ram, doesn’t cause pc to turn off. But i guessing my ram was faulty, i know some problem who faced this restart problem because of that faulty ram, or maybe some people faced BSOD
And please be specific on this, Thank you for letting me know what power controller is,
It exists on the motherboard right? So you’re saying my motherboard i mean the power controller is facing some problem?
First, noise is best eliminated at the generator and not at its victims. Previous noted was the snubber that should be on its switch.
Second, is that noise excessive? Always define the problem long before trying to fix anything. Use a portable AM radio tuned to a most distant station. What makes noise and what does not? If the blender is not making any more noise than anything else, then move on to other suspects … and do not even look back. Because the fact is that definitive - and is provided with perspection.
Computer must already have an AC line filter. As required by a long list include FCC and the ATX standard. But if that computer is assembled by a computer assembler, then does he know that only HE is responsible for the existence of that filter? Most do not even know what a filter is yet alone know they are responsible if it does not exist.
Many PSU manufacturers profit on their naivety. Forget to install that filter. Since PSU manufacturers do not have to meet any standards or requirements. That is 100% on the computer assembler.
So does that filter exist? Again, use diagnostic tools. Does the computer created more noise on that AM radio? If yes, then the computer is built defectively. Now fix that defect.
Plenty more solutions are possible. A long list. But none can (or should) be provided until the defect is first defined. And nothing but a subjective symptom has been posted. The blender causes a power controller to cycle power? Again, start only at the item that is relevant. OS, Ram, disk drive, etc all do nothing for power cycling. Only some can make a request to that controller. Only the controller decides when a PSU can turn on or off.
Why is noise from a blender getting anywhere near that controller? A defect is suspect. But we are far from defining what that defect is.
A snubber was described earlier. It was ignored because is was new. Because nobody even mentioned it before. Because so many have knowledge only from the many myths they were told. That snubber should have resulted in many questions if a responsible strategy was being used. That snubber is probably missing if a blender is making excessive noise. Is it? Use the diagnostic tool to discover what exists long before jumping to solutions.
Could you explain to me how you would attach an electrical filter to a program to prevent EM noise. Please use a diagram. In twenty years of electrical troubleshooting experience in ccws and amcom this is a new one for me.
So the defect centers on the power controller. Is it defective? Or do other defects exist meaning excessive noise is inside that computer? Another post discusses the missing and needed strategy. Find the defect long before even considering any fix.
Replacing the motherboard is classic shotgunning. Keep replacing good parts on wild speculation. Techs can even be fired for doing it.
Long before replacing anything, first define the defect. A diagnostic tool is discussed in that other post.
First, and I’m going to step into my old role as an aviation electronics tech.
First, if you don’t have some degree of training you should not be opening up the powersupply to a computer that you like (I would say at all, but meh…seriously though it is a good way to get hurt)
Next, an AM radio is as tools go fairly primitive by 1950 standards. Yes they are useful for spotting EM noise but there is far better out there.
Next, yes a pc will make EM noise, and while they are designed with mitigating that in mind well…reality and dreams do not always walk hand in hand. But as a rule a program does not have the physical parts needed to generate an EM signal…you know parts such as an antenna to radiate it or even you know a source of electrons to do kinky stuff with.
And last, please for the love of the unholy do not go out buying tools, expensive tools to troubleshoot a machine, while I’m all for education and new experiences it can take months or years of training to be able to effectively use tools for working on digital electronics in an effective way and I doubt most people here have the toolset to enact any repairs of any hardware problems they may have found.
And again, please do not investigate the inside of a pc powersupply, it can be a fire and health hazard to do so. And it can be a fire hazard long after you put it back together.
2 Likes
Do note that the user you are replying to has just joined the forum recently, and as far as I can tell, solely for this topic. Doesn’t seem to know what blender is, either.
I’ll give you that and I’ll set aside general electrical maintenance principles for a moment. But with that said I am still skeptical about the mental state who joins a community to comment on something, without first checking to see what the community is about. That is some internet 101 right there.