Posted by: kurtsh | October 1, 2005

NEWS: Victory for the Sony PS3 looking more unlikely as support for Kutaragi drops

To those that have experience in both industries (computer and console) it would appear that the battle between XBox360 and Playstation 3 is looking very similar to the battle betwee Intel’s Itanium 2 processor and AMD’s Opteron processor:  The one console with the lowest integration costs while maintaining an acceptable level of architectural innovation, will win.
And we all know who won the architectural war between Intel & AMD.  For those of you not privy to the war stories of EITHER the "64-bit CPU war" of 2004-2005 or the "Next-gen Console war" of 2005-2006, let me tell a quick story. 
"The 64-bit CPU War"
Intel’s Itanium II was released on the premise that people wanted 64-bit memory addressing to enable massive amounts of directly addressable RAM while also increasing the register size to quicken typical computations by minimizing the number of CPU cycles required for a single calculation.  It provided no backward compatibility for native 32-bit instructions at all, depending instead of Microsoft’s 32bit application emulation engine.  (Inherited from DEC Alpha, back in the days)  The result was that 32-bit applications when run on Itanium II processors over Windows 2003 for 64-bit Itanium ran at a 30%-40% performance hit.  Now because the processor speeds increased and because the overall chip architecture provided faster computation this was something of a wash in many cases.  However suffice it to say, to take full advantage of the chips speed, you had to have a 64-bit application written and compiled  specifically for Itanium II’s instruction set.
How many applications do you know of that are written to Itanium II, hmm?   Well, there’s SQL Server 2000 for 64bit… and… uh… well… uh…
Now there’s also the AMD Opteron.  This processor was released with a sort of hacked architecture that wasn’t at all designed for ‘elite, top shelf’ performance like the Itanium, but rather to simply provide fast 64-bit memory addressing for larger servers with lots of RAM, and to ALSO provide 100% backward compatibility with 32-bit applications on Windows.  As a result, people are able to run 32-bit applications at nearly 100% of the performance they’d expect from a traditional 32-bit processor on the AMD Opteron x64 processor without any penalty in performance.
Never mind that AMD Opteron’s are not as fast overall tha Itanium, or that they don’t scale nearly as well as Itanium.  The end result has been that Itanium has been relegated to a very niche market segment and Opteron has become mainstream… so mainstream that Intel has produced a chip that’s very similar to Opteron called the Intel Pentium EM64T that does virtually the same thing as Opteron, except with a Intel logo on it.  (Okay – jackass stickler out there will rant and rave unless I mention that Opteron is technically a superior implementation of the x64 bit instruction set and architecture and that EM64T doesn’t scale as well… but that being said, I’m impressed that Intel responded so quickly to market demand.)
NOWWWWWWWWWWWWWWW let’s look at the upcoming Console War.
"The Next-gen Console war"
Sony’s Kutaragi has been ranting and raving about the fact that he believes that PS3’s architecture with it’s 7 Graphics CPUs and single General Purpose CPUs will advance or as he calls it, "evolve" the console market pushing it forward in a positive direction.  This architecture will force developers to think in a different way and create more sophisticated, more comprehensive games.
Meanwhile, back at the Hall of Justic… I mean at Microsoft, the XBox Hardware Guys (see Major Nelson’s blog: for their comments) have been talking about how XBox360’s 3 General Purpose CPUs and graphics bus bandwidth not only provides the best sandbox for both today’s and tommorrows console games, but it makes development of these games, simpler, easier, and less cumbersome for the Game Developer, unleashing their creativity. 
"Why XBox360 will win"
What Kutaragi doesn’t seem to understand is that Game Developers don’t want to be manipulated.  Game Developers already face tremendous challenges when it comes to game development costs.  They already have development schedules and labor costs that go far too long.  They already face too many risks in developing games that, frankly, most of which will flop and not allow them to recoup their costs.  While it might be in the industry’s best interests to change the way games are made, Sony’s approach to game development will not endear them to game developers themselves.
I’m a game development veteran of sorts.  I worked for Electronic Arts for a period of time as a Game Tester
and I tried such games as one called "Radio Baseball", another called "Starflight" (oh god that was good game – written in Forth too, for those of you who know what that is) and I’ve see the expressions on game developers faces at both Microsoft and at Electronic Arts: 
I once aspired to be a game developer as a teenager until I looked into the eyes of a guy that worked at Electronic Arts and helped to create a game called Space Simulator, which was like Flight Sim except in space.  He’d visited my high school on "career day" when I was a senior and was talking to a group of 75-100 geeks like me.  I looked at his posture, his soft spoken voice, his messy hair, and his clothing.  He just looked so sad and downtrodden.  I got to speak with him one-on-one for a long time and he told me that, in a nutshell, no matter how much you LOVE developing games and no matter how creative a mind you have, you need to have a near-Godlike amount of willpower, a physical consitution that would repel Ebola, and the organizational mind of a organic Franklin Dayplanner to be successful in this industry.  Your schedules are so tight, and expectations are so high, and the emotional highs and lows are so great, seeing your creation get warped by executives, hyped in the public, and more often than not, rejected by gamers, reviewers, and even your own publisher.
And now Sony wants to make things "harder" for game developers.  It’s not as if Shinobi, the old PS2 SDK was simple to develop to (or cheap)… oh no.  Now Sony wants to teach everyone in the game developing industry "a lesson" in hardware and software evolution.  They want developers to write code in a different way to take advantage of the 7 GPUs of the PS3 and limit the single General purpose CPU’s usage.  Never mind that the developer then has to figure out a way that the game will port over to other consoles and platforms with totally different architectures (XBox with 3 General CPUs, Revolution with a rumored similar approach, the PC with 1 CPU and 1 GPU, etc)  because in this day and age a game when created can only really make great money if it exists on multiple consoles.
Meanwhile, XBox360 folks are saying, "Hey – come on board.  We’ll make it easy for you.  You’ll develop to a multi-CPU architecture that you’ll be able to use for any purpose – not just graphics computations – and you’ll have a HUGE sandbox to play in with all the memory and graphics you could want, and we’ll even throw in a little thing called XNA that will allow you to easy port your game over to the PC once you’re done with it on XBox.  You won’t have to deal with drastic architectural hardware changes.  We’ll make your job easier – not harder by allowing you to work in the same way that you used to work back in the PS2/Original XBox Game Development days."
Doesn’t this all sound familiar?
Here’s another article that talks about this very premise:
Victory for the Sony PS3 looking more unlikely as support for Kutaragi drops


%d bloggers like this: