Hacker News Reader

Comment by psyc | Hacker News Reader

Graphics programmers have been predicting this revolution for almost as long as I can remember. It's quite exciting that so many of the big players seem to believe it is almost time.

View on HN

> Graphics programmers have been predicting this revolution...

Mostly it has seemed like it's been everyone except the actual graphics programmers.

Yep. There's a billion and a half reasons you wouldn't use full-blown path tracing for games. It'll always perform worse regardless of what you do, the frame budgeting is likely going to be a total nightmare, the tree search you need to trace rays requires too much branching to map well to GPUs (it's perfectly possible, just another absolute nightmare as your complexity goes up), console GPUs are (and have always been) extremely underpowered in terms of compute (but good with textures), etc.

Limited uses? Sure. Otherwise hell no. It's never been more than a gimmick GPU vendors use to show off (in real-time, offline rendering is something else entirely).

Maybe GPU vendors will offer specialized spatial look up trees or something .

To clarify: that's more of a problem in programming the GPU in a portable manner, and that may or may not turn into a problem depending on the implementation. GPUs can do loops that run a different number of times per thread just fine, at the cost of having to wait for the slowest to finish. Things start to turn sideways if you want diverging branches though (at least AMD kind of has support for this, but it's limited). If you want animations to work there's a good chance your search would become pretty complicated and you run the risk of smacking into that.

Also, I doubt GPU vendors would add special-purpose hardware like that. They'd much rather just add some new shader instructions to help doing it.

> Also, I doubt GPU vendors would add special-purpose hardware like that. They'd much rather just add some new shader instructions to help doing it.

The PowerVR guys added special purpose hardware for doing spatial lookups for raytracing.

And it's awesome.[1] Getting shadows right for the specifics of a game is a nightmare; balancing shadow acne and performance. Just being able to do one-size-fits-all shadows on desktops would be a major step forward.

[1]: https://www.imgtec.com/blog/ray-traced-shadows-vs-cascaded-s...

For reference, VSM/4MSM are a set of shadow mapping techniques that don't suffer from acne, at the cost of suffering from bleeding problems that also need biasing.

Shadows are probably one of the places where raytracing makes a lot of sense though. It's one of few shadow rendering techniques that don't blow a ton of precious fillrate, and it shouldn't take too many rays.

In the days of ML based AI, seeing raytracing + revolution is both refreshing and funny :)

Actually, raytracing is a domain with strong potential for AI, as there is a lot of work using ML models to de-noise ray-traced scenes in real time.

Edit: Yeah, NVidia actually just announced a big project on that exact subject: https://www.youtube.com/watch?v=jkhBlmKtEAk

Pixar is also using Machine Learning to denoise path traced images: http://graphics.pixar.com/library/MLDenoisingB/paper.pdf

Ha, I would never have guessed this use case. Very fortunate for Nvidia to provide both GPU and NPU cores for the task :)

I was also wondering if the brief tangent in the article here that the DXR used more Compute than Graphics cores (given availability) may also indicate that they are using something of an ML approach to the Raytracing renderer here?

Obviously, it's hard to extrapolate from just a press release, but with Microsoft's big ML kick of the last few years, breakthroughs in ML steps in raytracing might seem like a reason for the press release.

Reply on news.ycombinator.com