Hi, Where it’s taking me? i don’t understand
on the Substance Painter issue: it reads like your gpu can’t handle multiple 8k textures and starts overloading the blocks with 0 bytes (after threshold) until the driver crashes. It usually recovers itself after 5 minutes… at least with the latest AMD chipset drivers. Windows from what I understand doesn’t allow Apps to total overload the RAM or override of the allocated memory… it would simply throw Access Violation, shut down the App but not reboot. Also AMD CPUs no longer have OpenCL support… or show up as OpenCL device …so I assume that it is not the CPU or RAM.
I think it’s a bug that clashes with the AMD drivers… Maybe testing newer builds of Blender can help this… or you switch to CUDA (buy 1070ti … that will very likely solve the issue).
edit: …stupid question, but have you updated windows completely? There have been some major packages released which fixed all sorts of hardware compatibility issues in the AMD sector.
Looks like new updates just came in, i updated it 2days ago.
Btw i ran memtest86, passed the pass 1 out of 4, error 0, i don’t think ram is the issue here since i tested with another slot of rams.
I can’t buy a new GPU right now, and i can understand that SP overloading my rams, but that blender scene doesn’t. it doesn’t even 2GB of ram properly,
here you can see in this stress test gpu pulling so much power still pc is running. But i’m looking for something to stress the GPU vram but i can’t find any.
One thing i just noticed. But it doesn’t happen all the time
I was stress testing but it didn’t restart like always
Then worked in marmoset toolbag that was pulling 180W but it was fine
Opened blender and was surfing around with 4px tiles size but it didn’t restart
Worked for 2-3 min
Then closed blender and opened substance painter and exported 8k maps still didn’t restart though it maxed out my rams
But after closing it everything went to normal.
Then again opened that scene in blender and turn on rendered view with with 2px tiles size but stillpc was running. Until i changed the tiles size to 4px
And pc restarted instantly 
That would indicate that it is very likely that this is due to Hardware issue.
This would be a good test to rule out GPU issues.
Could you share some .blend file where you can reproduce the computer restart with 4px size?
Here is the file
Let me know if blender is crashing or others,
Turn the viewport samples to 50, and volume tiles size to 4, and move around the scene, do this couple of times, get out of render view again turn on render view change tiles to 2PX, or 8/16PX again go back to 4px,
If i do this my pc restarts, let me know if it’s doing the same to you
https://mega.nz/#!d9xSkYLK!NcH7rUiLzkHFQ6yzNVyk3UjpkRtSMgD7URwirU6PURs
I don’t know man, I’m guessing it’s hardware issue but weird.
Here testing on Linux with nvidia. GPU power consumption and temp is working as expected. 2px is more demanding than 4px and 8px. 4px is more demanding than 8px:
1 Like
Yes, and the weird thing is my pc doesn’t restart when i’m on 2px but it restarts when i choose 4px.
Try turning on collection 5 (the whole scene)
It works similarly to what is shown in the video with all the elements not hidden.
Then i guess something wrong with my hardware, But it’s getting very hard to find out, If it’s mobo then i can change it no problem, but for the gpu i have warranty, and someone from that shop told me that if it doesn’t restart when stress testing or playing games they won’t replace it.These are just local shop,They’re gonna send it to the main dealer to claim warranty, And they won’t care if it’s only causing problem in this scene or not.
Yes, everything indicates that it is hardware issue. But it is really strange that there are problems with 4px and there is no problem with 2px. You report the problem with developers just in case. You must share the scene with developers in the report.
By the way, to properly test you must orbit the view to demand GPU.
Yeah also exporting 8K maps with so many layers from SP causes restart as well,
I’ll report the problem with the developers, and check the local shop again tomorrow to test my GPU on another mobo.
Btw someone mentioned about Bios, can bios could be an issue? cause i don’t think so, otherwise PC won’t even turn on i think.
Not sure if motherboards have options for discrete GPU overclocking. By default BIOS profile/settings should not cause problems, unless the motherboard is damaged.
Edit:
If your motherboard has an extra PCI-e slot for GPU available, you could try plugging the video card into that other PCI-e slot.
These symptoms are quite common when it comes to slowly failing RAM stick 
If you have multiple, try taking one out, then boot PC, stress test, put it back in, remove another one, and so on… The faulty one will usually fail stress test very quickly, sometimes it won’t even boot. When it comes to stress testing RAM sticks, just make sure to do something that takes lots of memory (but at the same time not more than you have installed at the given moment so that you don’t crash the system )
1 Like
That system (event) log error message should have a number. Never ignore numbers. Especially when that number does not mean anything to you. It is probably reporting the must critical detail.
Is that number a 41? Then the power controller has cut off power. Nothing in software is relevant. Press the front panel poweron button. That does not turn on any power. That is only a request to a power controller. Controller then decides whether to honest that request. If it does, then it orders the PSU on. Power controller also decides when power is turned off. It also decides when the CPU can execute - when other hardware is setup - when software can execute.
Ram, heat, and other internal parts are irrelevant to AC line noise. Most want to blame only the few things they understand. Nobody demonstrates any knowledge of what makes all power on and power off decisions - that power controller.
A computer must have filters so that electrical noise on AC mains does not get inside. Apparently that computer is missing the essential filters. A common problem when a computer assembler constructs a computer. Since only he (not the PSU manufacturer) is responsible for providing that filter inside a PSU. Since most have no idea what they are responsible for (and do not even know what a power controller is), then essential functions get “forgotten”.
The blender is probably creating AC line noise. That noise is getting into the computer. One can fix the blender so that it stops making noise (that is probably also heard on AM radios tuned to distant stations). That involves a snubber.
Or one can install / fix the AC line filter that should already be inside that computer.
Other mistakes can also compromise what must be robust noise protection inside that computer. For example, a motherboard should only have one conductive connection (ie standoff) to the chassis plate.
But again, this assumes an error code number from event logs that was withheld.
1 Like
Thank you very much, Though i didn’t fully understand what you said about AC line noises.
What can i do about it? In short what you’re actually telling me to do right now ?
But in that shop i tried different ram but still pc restarted.
and i can understand what you’re saying also know about ram issue, but also a ram should cause restart when playing games, games also eats up too much ram, but i’m not facing any restart when playing games,
but when exporting 8K maps from SP with so many layers and in this blender scene
Hi one thing i forgot to mention, pc only restarts when i change the tiles size to 4px, but only in this scene
i have another scene with volume as well but that doesn’t cause any restart when i change the tiles size to 4px,
though that scene is not big as this one, but it doesn’t matter even if the whole scene in turned off expect the car, still i face restart problem with 4px tiles size
I don’t think that is relevant, one scene can be heavier than another for different reasons.
The point is that it sounds illogical that in that scene there are problems with 4px and not with 2px. Although all this looks like a hardware issue, you make a report sharing the problematic scene with developers.
For me to be sure of what you say. With 2px while you continuously orbit the view for a long time, does the problem never appear and the PC is never restarted?