Conclusions and AnalysisThere is a lot of information to take in here as talking with John is never a disappointment. Let’s try to take a few minutes and summarize what his comments reflect in the various areas of ray tracing, consoles, physics and more.
The first point to analyze is on the discussion about ray tracing; John obviously feels that Intel’s current stance on using the traditional ray tracing algorithms for gaming isn’t likely to win out for the next-generation rendering technology. This obviously isn’t great news for Intel since even though John isn’t the only game programming genius around, his opinions seem to have more weight than just about anyone else in the industry. John more or less backs Dr. Kirk’s view that rasterization has too many innate performance benefits to be taken over by ray tracing UNLESS Intel is successful in stretching the Larrabee architecture to incredible performance levels. John mentions Intel would need something to the effect of a 3-4x clock rate advantage on their design versus conventional GPU clock rates to achieve that goal – and while we can generalize and see that he means Intel would need 3-4x the general purpose computational power over competing products, that means a GPU speed of 3.0 GHz or higher when we use today’s AMD and NVIDIA GPUs as a reference.
What John does see ray tracing useful for is a very specific data model he has created called “sparse voxel octrees” that allow him to store immense amounts of data in a fashion that is easily accessed using ray tracing methods. That doesn’t mean that John sees direct rendering occurring using those rays, just that the theory and algorithms of ray tracing are useful for his specific purpose. This new data model and algorithm being worked on for id Tech 6 would allow, according to John, nearly infinite amounts of geometric detail in the world without the problems seen with tessellation engines or trying to store gigabytes of data locally. Though we asked for some kind of diagram or screenshot that might demonstrate what kinds of detail John was referring to, we weren’t able to get anything quite yet; that’s to be expected this early into development though.
John Carmack, 2007 – courtesy of Quakecon.org
John’s comments that a proof-of-concept is required for any new technology to be adopted agrees with my theory on why Intel has been spending money on companies like Havok and Project Offset; they need to create a demo or engine that will make people jump up and be impressed enough to warrant new hardware and software design. Intel has only been showing previous generation media on ray tracing and thus has not caught the attention of the software community at large. John however has said that he plans to have a proof of his technology and its benefits up and running sometime this year using the sparse voxel octree approach to geometry. I know most people will be eager to see what kind of creations come from this research and you can put us on that list as well.
John and David Kirk also agree that mixing and matching traditional rasterization rendering with ray tracing is not only possible, but very likely, despite some reservations that Intel’s team has discussed previously. Not only that but John, as both a modern console and mobile developer, completely dismissed the idea of moving ray tracing engines to portable devices or UMPCs as being a “ridiculous argument.”
Changing the topic somewhat, John and I discussed the status of multi-GPU systems. With the push that both NVIDIA and AMD are making with two, three and even four GPU computers John thinks that in some cases where the game engine is simplistic enough to be sliced up easily, the benefits of being able to double or quadruple the power of a system are fantastic to developers at the very least. Where the trouble comes in though is with the introduction of DX9c and DX10, with features like feedbacks, sub-buffers and conditional queries, slicing up the work for multiple GPUs to address is getting much harder. Take this as justification for why scaling in titles like Crysis is just not shaping up to be what we had hoped even when paired with four of AMD’s high-end products.
The move for developers to focus on the console and port to the PC rather than vise versa is also the culprit for what many people see as the lessening of PC gaming. Because the developers are financially required to target the larger and more profitable audiences of the PS3 and Xbox 360, it doesn’t make sense for them to spend a lot of time or money to change the media or algorithms in a way to take advantage of the extra power a high-end PC can offer. John says that while increasing the resolution and frame rate on the PC is easy to do, really changing things for the PC market isn’t going to make much sense for a developer.
Finally, John has always had a public distaste for AGEIA and their hardware physics push and he sums it up quite nicely here for us today. Obviously from both a personal and professional level he disagreed with what AGEIA was doing but does see the future general purpose computational power of the GPU being utilized for physics and other gaming related problems. For the immediate future the ability to switch back and forth from graphics rendering to GP computing is too slow to be easily utilized in current engines but as either the hardware changes or developers update their software to adjust for the slightly increased latency, that processing can be included.
I’d like to thank John Carmack for taking the time to speak with me on these topics and I hope that our readers find this information as revolutionary and as exciting as I did. As is always the case in talking with John, you get a lot of fantastic information in a short period and taking the time to digest everything he has commented on here might take longer than this article today.
UPDATE: We have posted a podcast (recorded radio show) of my discussion with John Carmack that covers some of the topics discussed here and our analysis of it. If you are interested in hearing Carmack speak about these topics and get some more opinions on them, head over to the PCPer Podcast show page and check it out!!
You can find John’s work over at id Software or Armadillo Aerospace if you want to further follow his exploits!
Please join us in the forums to discuss this incredibly interesting information!!
More Reading on Gaming and Ray Tracing:
- NVIDIA Comments on Ray Tracing and Rasterization Debate
- Ray Tracing and Gaming – One Year Later
- Rendering Games with Raytracing Will Revolutionize Graphics
- Ray Tracing and Gaming – Quake 4: Ray Traced Project
- Intel buys Project Offset, makers of the Offset engine
- Intel demonstrates ray tracing on ultra-mobile PCs
- Playstation 3 Runs Real-time Ray tracing
- His future’s so bright … he’s gotta ray trace shade
Be sure to use our pricing engine to find the best prices on NVIDIA and AMD graphics cards and anything else you might need: