Hello, I was wondering if there are any technical details about how NVIDIA fixed the rainbow like artifacts in driver 531.41?
I’m a game developer, and we’ve met similar problem during development. It happens several months ago, randomly after some camera rotation or movement. Some textures become rainbow-like and some meshes are broken, stretching, flickering, flying everywhere. From GPU capture, IA output is broken, containing invalid data like 8e+38, nan…
We are using an inhouse engine and we couldn’t locate the cause… But after updated to driver 531.41, the problem just disappered. (Exact 531.41, former ones still have this problem). I was wondering if it’s an driver issue, or it’s something that we need to fix from engine side. So it would be really helpful if we could get detailed information about the driver update. Thanks!
Hi @dumpling1001,
Which artifacts are you referring to? Recent ones or the older ones from last year like here on reddit?
https://www.reddit.com/r/ForzaHorizon/comments/vysqpf/visual_artifacts_on_road_and_bushes/
While I won’t be able to share the actual code details on how this was addressed, I might be able to give a root cause summary if you can narrow it down a bit. And if it falls into something I am able to disclose.
Thanks!
Hi, MarkusHoHo,
Thank you for your reply. Yes, the texture problem in our game looks similar to the colorful bushes in Froza Horizon, textures of some models (sometimes bushes, sometimes rocks or other models) become blocky and rainbow-like, very cyberpunk. Likethis:
I’ve also found a similar topic here: Rainbow artifacts in many games and | NVIDIA GeForce Forums
While having the texture issue, we also have another problem with random mesh data broken, stretching, flickering, flying everywhere. It’s not mentioned in the article and seems unrelated, but this problem is also solved with the driver update 531.41.
It’s mentioned that it might be related to VRAM. Our scene is pretty heavy, contains about 20 million triangles, 8000 drawcalls and use more than 7GB VRAM. Not sure if it would be the cause…
Thanks!
Hi again @dumpling1001!
Good news to share with you, it was very likely only the driver bug that caused this. So if it is fixed for you now, you can attribute it to the driver update.
Regarding the “how” I am not allowed to share much, but suffice to say it did have to do with high-memory usage scenarios, which means that your second issue might have been because of the same root cause.
I hope that helps at least to a certain extent!