New Chat with John Carmack Likely to Make Your Head Spin
By Chris on Monday, March 17th, 2008 at 2:38 PM PST In Computer, Game Companies, Game Consoles, Game Platforms, Gamer Life, Games Industry, Hardware
If ray tracing and sprase voxel octree-based technology is over your head, you might want to just skip over this story of John Carmack chatting with PC Perspective. For the two of you who are still with me, the legendary developer had a lot to say on the future of PC hardware, starting with ray tracing and then exploring multi-GPU and multi-core CPU setups, AGEIA (surprise: he still hates them), and much more.
While Carmack was fairly optimistic when discussing many of these new types of technology and strategies, including multi-GPU systems (which I’ve generally been opposed to up until this interview), he realizes that much of this (such as moving physics to GPUs) isn’t likely to produce results in the immediate future. Carmack attributes this in part to the necessity to change basic architecture, but also because developers heavy reliance on the console market – which doesn’t employ multiple GPUs – is making things more difficult.
NVIDIA and AMD driver teams have to hack up games to get them to work optimally on multi-GPU systems and that’s more difficult for them today than in the past. Do you think developers dependence on the console market, which is solely single-GPU today, is a cause of those headaches?
It’s probably making it even harder for the PC card guys because as developers get more sophisticated with the low level access we get on the consoles, the rendering engines are become harder to kind of, behind our backs, automatically split across multiple GPUs. We are doing more sophisticated things on the single GPU – there is a lot more data transfer going back and forth, updated states that have to be replicated across multiple GPUs, dependent sections of the screen doing different things. It’s still possible buts it’s kind of a hairy job and I definitely don’t envy those driver writers or their task at all.
The one thing you can take away from all of this is that high-end PC gaming isn’t on the verge of getting much cheaper.
The Carmack hath spoken…
He’s right about the current strict triangle rendering method fast becoming obsolete. We are going to hit a veritable performance wall of how much detail you can produce from the “break everything down into a triangle” method game design uses today. That’s not to say Ray Tracing and 3D-Pixels/Voxels are the future (the concept of hybrid rendering is fascinating, if a bit daunting), but they are definitely more efficient at describing complex scenes to nth degree of detail. Despite what Nvidia/ATI would have you believe, photorealism will not be achieved through what appears to be their future roadmap of brute force through multi-gpu solutions.
There’s already such a large gap between (PC) CPUs and GPUs regarding scalar performance that dedicated graphics hardware companies like Nvidia/ATI have nothing to fear from the likes of chipzilla. Intel could put 16 Penryn cores on a single die and never come close to the raw throughput of an 8800 when it comes to matrix & vector ops.
And since physics (I’m talking real physics here – not that ragdoll crap) is heavily biased in vector math, physics on GPUs just makes more sense. The problem is that hardware physics APIs right now are in such a bastardized state that hardly any game dev wants to commit themselves to it. We need a single standard for a physics API. Alot of this has to do with the mindset that the el-cheapo ragdoll effects is “good enough” for the average gamer. Someone needs to make that leap, else physics in games will remain in 1st gear.
“Alot of this has to do with the mindset that the el-cheapo ragdoll effects is “good enough” for the average gamer”
Hey now, i enjoy rag doll physics..watching bodies flail through the air after getting launched by an rpg or grenade is always amusing.
Why does anyone listen to this guy, he was relevant 10 years ago, the gaming industry was built on inovation and creativity not bashing companies for trying to create new categories and offer cool features, so what if Ageia wasnt able to meet “John’s expecations.” If he doesn’t like Ageia so what, move on and get over it. He probably wasn’t offerd enough stock by the Ageia exective staff to say “nice things” – the way NVIDIA loaded him up prior to their public offering – Hey John, you still driving that Ferrari the NVIDIA stock bought you?