How far along are RPUs?

eggrock

Supreme [H]ardness
Joined
Dec 1, 2003
Messages
4,102
I caught an article in this month's Scientific American about RPUs. Apparently some car manufacturers are using them to model interiors and test interior lighting. Boeing modelled an airplane (350 million polygons), and various projects dating back to 2003 have modelled 1 billion polygon scenes.

Anyone have one installed yet? ;)

Sounds like it still requires a farm to render, but the article implied that it's moving towards desktop PC hardware.

Links:

http://www.openrt.de/ - openRT project

http://www.gris.uni-tuebingen.de/hvc/hvc_ws/HVC-Woop.pdf - Old (2005) PDF on RPUs.

[edit] I look forward to an SLI system with a quad-core CPU, PPU, RPU, 50 Tb storage array and my SB Live Value card. Oh yeah baby.
 
eggrock said:
[edit] I look forward to an SLI system with a quad-core CPU, PPU, RPU, 50 Tb storage array and my SB Live Value card. Oh yeah baby.

I can't wait for SLI Live! Values :drool:

twice the processing power
 
I've always said polygons are just a stop-gap for raytracing, just like the Voodoo Graphics (with no hardware vertex processor) was a stopgap when all workstation (read: expensive) graphics cards had vertex processors. The introduction of high-powered shadow and radiosity engines on grahics cards are just the tip of the iceberg.

The good thing is, the same advances that make fast polygonal graphics cards possible (smaller process feature sizes, faster memory) also make possible fast raytracing engines.
 
Raytracing in games is a long, long way away I think. It's far more likely that GPUs will evolve extremely parallel, extremely programmable processing units, like a 128-core CPU that will be able to run almost any kind of code that a dev requires: physics, AI, game logic, graphics (either polygonal or raytracing). In the near future though, D3D10, and the chips for it, will be as close to that as we've ever gotten on PCs, and it's likely that some very serious research into doing raytracing on D3D10 GPUs will occur fairly soon. In fact, I think it's very likely that within the next 5 years or so, some developers will be using actual raytracing in games for, at least, either reflections or refractions.

Right now, RtPUs are a bit of a catch-22: In order for consumers to justify buying one for gaming, games have to be using them. In order for game devs to justify the work for them, consumers have to buy one.
 
Cypher19 said:
Right now, RtPUs are a bit of a catch-22: In order for consumers to justify buying one for gaming, games have to be using them. In order for game devs to justify the work for them, consumers have to buy one.

From the sounds of it though, production with RT is much quicker (and therefore cheaper) than rasterizing. If a dev can cut a big chunk of R&D out of the development cycle they'll jump on board fast enough.
 
eggrock said:
From the sounds of it though, production with RT is much quicker (and therefore cheaper) than rasterizing. If a dev can cut a big chunk of R&D out of the development cycle they'll jump on board fast enough.

With the current market though, a dev would be doing a RT game with a total audience in the double digits, instead of in the tens/hundreds of millions, and would make approximately zero profit regardless of how much, or how little, R&D the dev has done. Also, people will not buy an RtPU for a single title. Right now, a lot of consumers are apathetic about the PhysX chip just beause it's one more device they need to buy to even play some games, and it works across several titles now and the effect is quite obvious. For an RtPU, there are a couple existing tech demos and no games, and the effect is used occasionally at best, not to mention the fact that you'll get lower performance and image quality in general than what you get on a rasterizer-based GPU. It simply is not worth it for a consumer to buy now and forever, and hence it will not be worth it for a game dev to even give it the time of day.

And fwiw, I find it hard to believe anyways that raytracing would help reduce production costs of a game by any significant amount. A ton of other work is still required, such as texture creation, animation, modelling, game design, level design, graphics programming (<- the only point that is really affected by a move to RT), game logic programming , sound design, sound programming, scripting, and editor development.
 
Interesting points, but I believe GPU manufacturers will spearhead the advancement of realtime raytracing, not developers pushing GPU manufacturers into this technology.

It's a real chicken or the egg type deal. Did GPU manufacturers instigate the shader revolution, or did developers push GPU manufacturers into this arena? And what role does Microsoft play in the advancement in PC graphics? I believe the formula for advancement is two parts Microsoft, two parts GPU manufacturers and one part game developers. If you're missing one of these ingredients, you end up with something fairly useless.
 
phide said:
I believe the formula for advancement is two parts Microsoft, two parts GPU manufacturers and one part game developers. If you're missing one of these ingredients, you end up with something fairly useless.

Don't forget the 2 eggs, 1 3/4cup flour, and 1tsp baking soda.
 
Back
Top