I can reproduce with Blender 3.2.1 on my Dell Precision M6600 / Nvidia Quadro 3000M/PCIe/SSE2.

I can also reproduce on my laptop:
System Information
Operating system: Linux-4.15.0-142-generic-x86_64-with-glibc2.23 64 Bits
Graphics card: Quadro K3100M/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 418.56

No activity for more than a week. As per the tracker policy we assume the issue is gone and can be closed.

Just a poke regarding my inquiry above, otherwise we will have to archive this for now.

Looks like it's working now with newest blender and amd cards on the newest osx monterey.

This problem persists in Blender 3.01 and 3.1.

I still get this problem with a MacBook Pro onboard Vega 20 4gb GPU. So not exclusive to eGPU's but Vega cards.

Just checked with version 3.1 beta on Imac 2019 (intel) with Catalina. The Problem persists, the only "solution" as mentioned in the previous comments is to put a Geometry node set to Position into the texture vector.

In T77572#994309, @Lewis Toohey (Lewis2e) wrote:

I also have this issue.

I believe the root cause is this, which also caused a similar issue for me with Houdini

As of Mesa 19.2, a new OpenGL driver, Iris, is available for testing. In Mesa 20.0, the Iris driver is promoted to be the default for Gen8+.

The Iris driver is causing issues with various OpenGL applications.

I am able to (temporarily) fix this issue by exporting MESA_LOADER_DRIVER_OVERRIDE=i965 in the shell before running Blender. This forces me back to the i965 driver, and opacity of reference images in Blender works again.

system-info.txt12 KBDownload

This report has been unconfirmed for 1 year so far.

I either get corrupt data on the second and third attachment or I get zero data in software mode.
I am not sure if the code is correct to classify this as a bug, however.
All your attachment textures have a type of GL_RGBA32F but you write float and vec3 outputs in the fragment shader. Is that correct?
Can you also attach the gpu module version of your script?

few years in and they still haven't fixed this crap...

yep got the same thing here however ever since I switched to 3.0 the problem has gotten worse and when I render other volumes the problem doesn't occor.

Don't worry @LanaMclaughlin (LanaMclaughlin) this is a bug and it is not related to your graphics card. Sadly is that the devs from blender or invidia are not addressing this for years which makes Eevee unusable with volumes. The only thing we can all hope for is the rewrite of Eevee in blender 3.2+ which has a chance to be better.

Hey hey hey, All here who are willing for a easy solution is that just download the atio6axx.dll version 6.4.10.1339 and just copy an paste it in the /c/windows/sytem32 it will work!
I just solved an issue .
Don't forget to make backup of the previous dll file.

hi i have the same problem
os = windows 10 20h2 pro
gpu = gt 710 driver 472.12

Hi all,

Same/similar issue on 3.0.0. Crashes sometimes upon opening. Error message says Threads have "bad access"
This is on intel based mac.

I have some observations regarding this issue related to Blender v2.92 and Blender v2.93.5.

The problem exist between both iMac and Macbook as I am on a Macbook with an AMD Radeon Pro 5300M 4 GB and also working in EEVEE. Any texture node added, even with a ColorRamp and Mix Shader involved, still has the volume disappear. When the texture is disconnected, the volume reappears.

In T77202#1220692, @Lars (mvipe) wrote:

Blender please let us know how to fix this.

yes and no, it did help kinda to set pixelsize of volumes in eevee to 8px, but honestly it produces to low res renders even with samples of volumes set to 64 so it is not really an option,
also it doesent eliminate the issue completely it still happens just less

@Lars (mvipe) did you try to set the tile size bigger? Did it help?

Actually, the issue did persist even with the latest drivers game ready or studio from Nvidia. I couldnt resolve it in the end :( i jsut got lucky with some short renders that didnt have the flickers..

The problem is still present on 2.93.4 and the experimental 3.0

It seems that I've run into the same problem. Tried the drivers @Lars (mvipe) mentioned, as well as latest 471.96 (both game ready and studio). It didn't make a difference for my RTX 2060 (Win10 x64).
Here's my stack exchange question - I tried a lot of stuff. https://blender.stackexchange.com/questions/237837/flashes-of-light-escaping-through-solid-objects-in-rendered-animation
I really hope this will be fixed. I've ran into this during my first month in blender, I'm making a really cool project, and I'd hate to spend additional time picking out faulty frames from a 6 min music visualizer video by hand.

Same problem happened on Studio driver 471.11, RTX2080super, Blender2.93LTS, Ryzen5950x.

Seems like other people have similar issues.

Update. I notice few things. Firstly, INTEL_DEBUG=reemit environment varriable resolves the problem, but in cost 50% performance in heavy scenes.
Also when you trying to capture blender 2.91 or newer with this tool https://github.com/nowrep/obs-vkcapture , you get strange picture. With Blender 2.90 capture works just fine.

hello, for me installing the latestgame ready drivers (471.41 - WHQL), instead of the studio drivers solved the issue for now.
older studio drivers also produced the same problem for me here.

hey i have the same issue with eevee on nvidia studio drivers. started today after i installed the latest studio drivers from nvidia. very annoying. its definately a nvidia driver issue, since it literally wasnt there 10 minutes ago.

Could this be related to the GPU memory being hit at maximum on the GTX1050? Is there a way to know the used graphic memory at the moment of the crash? This card is being rated at 2GB of VRAM...

For any other poor souls who end up on this page, I had the same problem and fixed it by downloading the latest driver from the AMD website. Windows update and the "update driver" button in Device Manager did nothing. For what it's worth, I selected the "factory reset" option in the AMD installer, although I doubt it matters.

Sorry for the silence; I was hospitalized recently