Caustic Graphics Ray Tracing Acceleration Technology Review
Caustic's Business Model and Gaming
From a technology stand point, it’s hard to not be at least somewhat excited about what Caustic is claiming to be able to do. The ability to bring ray tracing closer and closer to real-time in a desktop platform would allow for a lot of potential applications including gaming, rendering, entertainment and more. There are of course some complications not the least of which are a few companies known as NVIDIA, AMD and Intel.
The current GPU kings, NVIDIA and AMD, and yes the third player in all of this with its upcoming Larrabee technology, Intel, would definitely have something to say when it comes to bringing any kind of ray tracing acceleration technology to the PC.
We have already spoken with NVIDIA about ray tracing on its current generation of GPUs and the improvements they are making in utilizing the current GT200 architecture for that purpose. I would obviously also expect the upcoming generation of graphics processors to handle ray tracing even more elegantly and more efficiently as the rendering method gains more and more steam. With the advent of OpenCL I would expect both NVIDIA and AMD GPUs to be improved upon when it comes to a ray tracing implementations and I don’t think I need to reiterate how seriously Intel is taking the idea of ray tracing on its upcoming hardware – we interviewed an individual on the topic while still in college in Germany who is now a project lead for ray tracing at Intel’s Research division.
If Caustic goes as far as to propose a ray tracing API standard with CausticGL (even if it will likely be under a different name in my opinion) then we would also see NVIDIA, AMD and Intel implementing their own ray tracing hardware acceleration algorithms and methods to compete with what the CausticOne and CausticTwo can accomplish. James and company are confident though that while other hardware vendors would definitely encroach on ray tracing acceleration that Caustic’s strength lies in the implementation of new, patented, algorithms and software. We could see a time where each hardware vendor creates their own implementation and accelerations for the ray tracing API much like they already do for new versions of OpenGL.
An analogy that quickly crops up when discussing Caustic is another add-in accelerator start up from a few years back: AGEIA. In that case a new company wanted to sell a hardware and software solution to accelerate a technology for gaming purposes though the idea has never really taken off since the install base for the AGEIA PhysX hardware was always so low. Now that NVIDIA has acquired the company and ported the software to GeForce GPUs there is a larger, but still small, collection of titles that implement the standard. For the desktop market, Caustic’s offering is very similar and has the same problem: game developers might want to add ray tracing-only features to their games that require the speed and acceleration of the Caustic card and thus an incredibly small market. More than likely this means that for the immediate future of Caustic Graphics, the desktop market will be untouchable.
Future of gaming?
The one advantage that Caustic has over a company like AGEIA is a second, much more profitable market to sell to: professional and enterprise users. While gaming is the most immediately visible target for this type of technology from a consumer’s point of view, the move away from rasterization on the PC and console world would be too dramatic a shift to have happen in my opinion. But there are a whole host of media and entertainment designers, product designers, architects and more that already using ray tracing on a daily basis, know and understand the benefits of ray tracing to their work and would likely see great productivity enhancements from the CausticRT platform.
When a car designer needs to make a final render to show to his boss for a new project, the ray tracing draw time for a single frame can be hours, if not longer. Caustic is hoping to move that process to near real-time. Now that same designer can work on the presentation of the project for a longer period and even show changes during that meeting rather than having to agree to return with another rendering after the discussion. The same cycle could be improved upon for any professional that uses ray tracing in their working cycle – and business is willing to pay greatly for computing speed ups of this nature.
Caustic is planning on working with companies like Cinema4D, Autodesk, Blender Render and others to begin implementing support for the CausticGL software into future versions of these rendering applications. If they can achieve a near-global adoption of their software then selling the acceleration hardware to design companies would be an easy sell. Caustic did purchase a company called “Splutterfish” that created the Brazil Rendering System in order to better understand how to work with software ISVs to integrate CausticRT and develop tools to utilize the hardware and software capabilities.
That professional market, where Caustic is spending the majority of its initial development time, is the key to long term success for the company and the technology.
The good news for any gaming implementations of the CausticRT platform is that the software stack between the professional and gaming segments is identical. As more professionals use the technology for design or even for gaming development (as I discussed earlier) the technology could become more and more accepted and developers would want to see it running live in games. Caustic isn’t afraid to acknowledge though that we are still many years from seeing a fully ray traced game engine – they estimate a solid 4-5 years will pass before their hardware iterations will be fast enough to handle 60 FPS at resolutions like 1920x1200.
There is still the possibility of seeing ray tracing segments of game development using a hybrid rendering approach that could be implemented with a simple feature check box in the game configuration. Because the Caustic software really would just be changing the appearance and look of the game, it would not require a hard fork in game development like the AGEIA PhysX processor did.