In T52802#460544, @Brecht Van Lommel (brecht) wrote:

Perhaps the denoising should work on log(glossy) or some other kind of remapping to bring the extreme values into a smaller range.

I just booted up my linux box, downloaded the current stable release and tried the tower_packed.blend file. Here it doesn't crash when I switch viewport, but I haven't updated nvidia-drivers yet, using 378.13 here. Haven't used linux for some time. I'm running gentoo, there's no relevant output on the terminal.
I'm off until sunday, when I'm back, I'll go and update the nvidia driver and report back on the behaviour.
HTH

Here's the output when starting it from the terminal:

Here's a comparison using a shot from Agent, where branched path is the clear winner:

Path	Branched Path

Yes, you can change the default in master and blender2.8 now, but note:

Thank you for the report. Currently we are aware of many issues in 2.8 and actively working to fix them.
But since replying to reports takes time, we have decided to limit bug reports to module team members.

I figured out how to pack all files into the .blend file. So please try the attached file.

There's only one texture used that I'm aware of, the environment texture. I used one from the sIBL archive at https://www.hdrlabs.com/sibl/archive.html and append it here, if it's what you're referring to.

Changing priority to Normal, until @Lukas Stockner (lukasstockner97) confirms that this is a bug.

Hi, I have this too.

@Maeldor (Maeldor), yes, should be bmw27_gpu. And you're right about --factory-startup. Let's do this way then: backup your settings, load factory startup, enable CUDA and then try rendering from terminal.

Hi.
I have read before that the mere fact of having installed some GPU overclocking application can bring conflicts with Blender, even for CPU render. I say this because I read above about MSI Afterburner. So just in case to do reliable tests, if you have any of them installed (MSI Afterburner, ASUS GPU Tweak, etc.), completely remove the application.

In T52572#460567, @Sergey Sharybin (sergey) wrote:

@Maeldor (Maeldor), if you don't overclock, then things should be working stable. Root of your issue might be different from @Stefan Eisenreich (Stef1309).

There is one more thing you can try: render file from console (cmd.exe), something like:

path\to\blender.exe --debug-cycles --factory-startup -y -b path\to\bmw27_cpu.blend -f 1

and see if that works and whether log gives any clues.

@Maeldor (Maeldor), if you don't overclock, then things should be working stable. Root of your issue might be different from @Stefan Eisenreich (Stef1309).

Thanks for your quick reply!

In T52572#460489, @Sergey Sharybin (sergey) wrote:>

@Maeldor (Maeldor), did you try rendering default cube on GPU? Did you try installing driver downloaded directly from nvidia.com ?

Thanks, I'll commit this.

Perhaps the denoising should work on log(glossy) or some other kind of remapping to bring the extreme values into a smaller range.

Yeah, I am currently suffering from the same problem. I think it's mainly a problem with the very bright intensities of glossy reflections. At least in my case it doesn't need any SSS to produce the issue, just a rather reflective surface. I am facing this problem mostly with metals in scenes with bright lamps. I think the denoiser somehow tries compensate for the brightness differences between the glossy reflection and the much darker surrounding areas, and by overcompensating that brightness differences it makes the problematic areas black. Or something like that. No idea how to solve that but maybe as a workaround the denoiser could just keep the super bright highlights just white?
Here's the above file but without SSS and point lamp, but with a metal surface, still producing a black artefact.

Hello Sergey!

This is in fact related on T42888, also rB1549fea, rBc5927cd.

@Stefan Eisenreich (Stef1309), we can not guarantee Blender to work on overclocked GPUs. There is a good reason why vendors didn't use higher frequencies to begin with. Fact that games are stable doesn't really mean much here, CUDA program will stress GPU much more than OpenGL/DirectX.

This file is missing textures in it, so it might affect on reproducability of the bug. Besides that:

I am also getting the CUDA error: Illegal address in cuCtxSynchronize(), line 1372 error on the majority scenes I attempt to render via GPU (e.g. BMW Benchmark, Racing Car, etc.). Unfortunately, it's making GPU rendering unusable for me, as I can't tell from one render to another if it's going to crash, and it requires a complete restart of Blender to be able to attempt rendering again.

Update:
I tried to get to the bottom of this since it seemed very strange that this issue suddenly vanished.
Now, since I use this PC mainly for gaming I have routinely overclocked my GPU as far as I could. I have of course tested the stability of this in games so my setup should be stable at all times. When I made the tests earlier this day I hadn't noticed that msi Afterburner had reset my overclocking settings.
Testing out further I found that only the core clock offset seemed to cause this issue, when I raised only the memory clock it rendered just fine.
So there, that's propably where the problem comes from.
I did all my testing with the .blend file from the previous post (denoiser disabled).

Sorry, my bad. I tested out the denoiser and forgot to turn it off for the error log. The error was 100% reproducable both with denoiser on and off but I tested again today and I couldn't reproduce it a single time. I made no updates to my Windows version, drivers or Blender version so I don't really understand why this is gone. I attached the .blend file anyway.

Can you maybe save .blend file with all the referenced textures etc, pack them to an archive? But also weird why importing doesn't happen correctly. Perhaps worth separating two reports:

Sure. I also did another experiment. I exported a model from Poser as an .obj. I did this twice, once with full texture path names and once exporting the textures to the folder.

@Brecht Van Lommel (brecht) @Aaron Carlisle (Blendify) Can this now be considered for being committed to master as Blender 2.79 is out?

Can you attach an example of corrupted image? I would like to see if we can make Blender to identify such images and prevent crashes (rendering magenta-color as a missing texture for corrupted image is more friendly than crashing)

Thank you both. Your answers helped me to identify the cause. The other program compiling the .obj (exporter) is creating corrupt images of the materials, and I suspect this faulty code is causing Blender to crash. After manually replacing the texture images in the export file with valid images, the Blender import to Cycles worked fine.
BTW, I had tried 4 different objects before posting here and all had resulted in the same issues. My bad for not including the exporter in the possible fault line.
This can be closed, no bug here....

If OBJ file has:

Thank you very much, Vuk. Just tried it and it works. This should get me by while I save up for a new GPU. The main thing I need is the viewport render preview which I use a lot and the CPU is just not fast enough.

In T52752#459286, @Alexander Chaplinskikh (AlexChaplinskikh) wrote:

Wonderful. Any way to whitelist it and build it myself from the source? I don't know Python but I could try. I was excited for this release but this just screwed me up completely.