Carmack unplugged

by on Dec.09, 2006, under computer graphics, consoles, games programming

Carmack Quakecon 06

Is there a point where pure expertise takes you far from the elemental principles of your subject? Such that your perspective actually becomes skewed? It’s an interesting question, and certainly a difficult one. History is replete with experts that get blindsided by innovation or worse, redundancy.

When listening to someone as clearly expert as John Carmack, it’s absolutely clear that there are few that can exceed his knowledge or experience in graphics technology. Every year, John delivers the keynote at his own QuakeCon convention. It’s somewhat of an industry joke that no one really understands what he says there. His whole manner exudes someone totally steeped in technology (more specifically, graphics technology). It’s always interesting, even if you don’t fully grasp every last ounce of his wisdom. He gets up, delivers a clinically fascinating speech, and leaves.

Last August, it seemed different. He was more relaxed, conversational, witty. I enjoyed the talk some time ago, but I realise that most people haven’t caught up to it. It tends to be very technical, and an hour-long oration on computer graphics isn’t how most people spend Friday night. So, I decided to share my attempt at “John Carmack distilled”. I recorded several of what I felt were the most representative clips of his speech and bundled them here. An hour-long talk boiled into five minutes of your time, how efficient! (for you, at least)

Note: I wrote the audio application in openLaszlo, which is unfortunately rather bug-prone. If you have problems, reload the page. I think I’ll wait a few more releases before attempting this framework again. Again, apologies for any bugs you encounter.


  It’s clear that he’s missed the point of the debate (or is choosing to ignore it). There is a vast chasm between someone asking for innovation in games and someone who whines that they’re not given money to do it. Perhaps Carmack hears from those aspiring developers more often than we do, but do they comprise the audience in attendance at QuakeCon?
  This is another familiar chord from the game community, one where Carmack clearly comes on the side of gamers. Yet, it’s a marginal contradiction. At the time of Wolfenstein and Doom, that was the cutting edge graphics in gaming. Furthermore, to pay tribute to the gameplay aesthetic and in the next breath to say, “of course our next project will look totally stunning’ seems somewhat fallacious.

It’s true that iD’s games have traditionally blended graphics and gameplay with excellent success, but I know I’m not alone in the thought that as iD’s games start looking better, they actually start playing worse.

  !!!!! How deeply into computer graphics do you have to delve in order to emerge from your laboratory and claim that ‘Chicken Little’ and ‘Cars’ are equally good graphically? It’s clear he knows a great many things that I don’t, so I’ll cast that ridiculous statement aside.

He continues to say that content is more important than technology, and to me that he is speaking more as a visual technologist than a game designer. See, there are hundreds of things that games are “missing”, and none of them have anything to do with graphics. He’ll repeatedly stress in his presentation that we are near “the end” of advancement in graphics technology. Yet, what of more procedural approaches to content and gamespaces? We’d surely benefit from an enlightened perspective from John, but what’s clear from his speech is that he seems no longer to be thinking outside of the box in these matters.

  The tools are obviously becoming more and more critical. Asset-heavy media extravaganzas are totally reliant on a solid pipeline and as much automation as can be flung at the process. My hope is that as we gradually shift to content that is more procedural in nature, the tool-artist relationship might actually change. In other words, it’s the opposite in the long haul than what Carmack is saying. Then again, he lives in the reality of how games are made right now, whereas I… only have my (hopefully) unique point of view.
  How can I dispute the master of graphics? He wins technically – he can back up what he’s saying in such detail that I could spend a week dissecting a single sentence. But one thing I have on my side (yes, I disagree with him) is history. I’ve seen proclamations of the end before, of inherent limitations, of exhausted avenues. But we always seem to persevere and progress. It’s just how the world seems to operate (at least to me). Dimishing returns, sure. But I also say, innovation can come from the strangest places. One way or another, either through more “active worlds” or procedural methods, in five years we will look back at today’s graphics and scoff. It just makes sense to me. What do you think?
  I can relate to this topic more directly. I’ve been on both sides of the argument, and I do agree wtih what he’s saying, but there is one aspect that he failed to address. Releasing the sourcecode is also an invitation to develop exploits and unleash them on existing communities. For a griefer or cheater, the more active the community, the sweeter the reward.
  I detected a certain resigned commitment in Carmack’s words. He says, “hope” to take advantage of the Cell… while we go back and refactor all our code”. From what I’ve read, the Cell is a monumental challenge for programmers. One aspect is the tools. Foremostly, Sony is not adequately providing for the needs of the programmer. The very architecture in VLIW-hybrid systems like the cell explicitly put the burden on both the compiler and the programmer to take branch-heavy code and optimize it for execution. Without adequate support, a difficult situation becomes much, much more difficult.

There are clearly no easy wins with respect to creating highly parallel code for games. It just doesn’t suit how humans think about problems. This is not the sole contributing factor to the problem or it’s solution, but I think it’s an important barrier for entry into a system that contains 8 separate processing cores (actually, 7) and how they can be explicitly managed by the programmer.

  The Xbox360’s tools are far better as well, leveraging Microsoft’s considerable expertise in development environments. But also remember, the Xbox 360 and PS3 have been architected such that their very innards are bent, understandably, towards the greater realisation of streaming and dynamic graphics. A lot of Carmack’s talk centers on the easy benefits of parallelization on the GPU-level of smaller packages of graphics tasks (like tesselation). That’s what the Xenon and the Cell SPE’s are totally designed to do.

What is most surprising however, is that the multithreaded advancements have been done literally at the expense of other things like larger cache area. The chip configuration enhances the processor’s ability to “write-stream” visual info into this somewhat limited cache. When this is happening on an architecture that has already made other sacrifices, that doesn’t leave a lot of leeway for more dynamic, non-parallelizable code like A.I. or physics. This is the interactive stuff! The “gameplay”! All of that comes from highly branchy, meandering skeins of code, which is brazenly handed off to the deftness of the programmers and the compiler. Left to cram all of the unravelled decision code into a limited storage region, we can easily see where the 360’s and PS3’s emphasis lay. Graphics, graphics, graphics.

Carmack clearly sees the world through the eyes of a graphics card. I won’t dispute his expertise on the diminishing returns of graphics innovation, I firmly believe that the major leap forward won’t be in graphics at all!

Despite focussing on my misgivings on aspects of the speech, I feel that his wisdom on these matters still clearly shines. In some ways, I hope that innovation will jolt a surprise or two his way. But, if any can take advantage of great leaps in tech, it’s the master innovator himself.

The actual video is available here. Totally worthwhile if you’re interested.

Technorati tags: , ,

10 Comments for this entry

  • b1alpha

    So funny, so what else is this guy gonna say other than the xbox rules over the psIIImillion. I think we all know that sony hit the crack pipe a few years ago a little to hard, when they went for "control"/

    I hate them to, thats what he meant.
  • covert.c.

    More of this is needed, not more detailed decals on my virtual racing car..

    Funny you should say that. In Korea, you'll be buying those decals with their microtransaction model.

    No really. Their most popular online racing game has you buy decals for real money.
  • covert.c.

    @b1 - the difference between you and Carmack is that you are *joking*. 🙂

    @Nazgum - that's as well stated as I've ever seen it. Cinematic gaming tires me out - just from the hype alone. Take CliffyB for a good example. Gears of War, indeed it looks amazing. But isn't it just Contra? Isn't it extremely short?

    If you look on my forums, you'll see a link to a gameplay video of Assassin's Creed. And in this, I'm in full agreement with you Nazgum: if graphics' progress stopped at Assassin's Creed, and gave us the interesting gameplay I see there, then I'd be totally happy.
  • Nazgum

    Personally I think the best thing that could happen for the gaming industry would be for graphics to stop progressing.

    Honestly, graphics have been 'good enough' for a few years now, and this continually graphic upgrade as an excuse to rehash old games just makes things dull.

    Without graphics being upgraded games would progress in the directions they need. Just look at Nintendo's Wii, those guys are on the right track. So what if its no better visually then a gamecube, they've enhanced their system in the ways that count with true innovation. More of this is needed, not more detailed decals on my virtual racing car.. =)
  • b1muglar

    I usually pull the later off, so I guess thats why I can appreciate the comment..
  • b1muglar

    I know! It is that sort of statement that can make you look realy funny or like a complete idiot. I think he pulled it off though, it was super edgy to say so, hes funny!
  • covert.c.

    You hit on the quote that sort of moved me to do this whole posting experiment.

    Chicken Little versus Cars? Is it because there are some super-specifc commonalities in their tools or something? To me, that statement is like saying a grade 5 art student looks the same as a Degas because they both use a paintbrush. Not a smart comment.
  • b.1.alpha

    That was entertaining, I have never listed to a quakecon talk, but you know he almost has a speech impediment. Is that dry humor or is he tripping out? He is so neat to hear though, even if just because he's our dungeon master. You know some of it was funny like when he said cars was not as good as chicken little, and its no big deal the ps3 is virtually impossible to program for, thats hilarious!
  • Uno

    I would rough this guy up for his Milk-Money.
  • Uno

    The wealth of knowledge this guy has is only matched by his milk-money.

Looking for something?

Use the form below to search the site:

Still not finding what you're looking for? Drop a comment on a post or contact us so we can take care of it!