i cant even play because i have a r9 380x and a fx 6300 6 core processor

im trying it on an integrated gpu because i hate myself

Ran on my 5-5600 with a 3060 just fine on the day of launch/update. Are you for real? Cuz thas no bueno!

it ran on my current installed OS, Windows 3.1x

does it work on windows 95

Has anyone tried to just set max bounces to 0, basically turning reflections off?

I wasn't expecting great performance on my 6950xt liquid devil, but I cant even get one fps on the lowest settings. Even cyberpunk runs at 30fps on full ultra with RTX.

@eversio Yeah, I did't think it was the ONLY way, it's just the main way. So if you don't have it you probably have a 90% or more chance of not doing it. If you are not the developer you don't really have a choice. It doesn't seem like installing OpenCL will automatically port CUDA programs to run on AMD though. Neither does HIP. So I don't know what to say about that. Most people have GPUs, not FPGAs or machine learning specific hardware, that's what makes them the common and unfunny choice for ML these days.

@GustavGlaven yes you can (via HIP) but my point is CUDA is not the only way to run GPU accelerated machine learning models and the like. They don't "just pretend like a whole technology and significant use for graphics cards doesn't even exist". There are open APIs for this stuff, you don't need to lock yourself into one vendor.
Using OpenCL means you can also target non-GPU hardware acceleration like FPGAs. Other non-CUDA hardware for machine learning exists like Google's TPUs and Amazon Trainium / Inferentia chips, all of which make CUDA a funny choice for ML these days.

yes