[quote name='PHaLaNX GTR']PS2 had, by far, the strongest CPU, but had no GPU. Without a GPU, the PS2 was fighting on an uneven battle ground, since the CPU had to basically do all the graphical rendering via "software". While, on the other hand, the other consoles alleviated all that burden from their respective CPUs by using hardware designed specifically for graphics rendering (GPU). This giving the Xbox the definitive advantage over the PS2 and enabling the NGC to go heads up, and, at times, ahead of the PS2. Microsoft had Nvidia in its GPU corner and Nintendo had ATI. Now with the release of information regarding the next generation of game systems, Sony has just announced that they will have the always venerable Nvidia at its side. If the Cell is as capable as its brethren, the Emotion Engine, for its time, then Sony may just have the one two punch, if all goes well. But then again, from what I surmized MS seems to have something up their sleeves with their next gen processor, as it seems to share a few ideas as the Cell, with multilateral processing, or some thing or the other, where multiple processors are working together as one. So this should be interesting as always. I can't wait to see the combined might of these two processors working together.
Hey, IGN can be useful, at times.[/quote]
This analysis involves a few misconsceptions. First of all, it's just silly to fault the PS2 for not having a GPU when the Nvidia marketing term had yet to be promoted when the PS2 chip set was in its design phase. And that is all it is, a marketing term. I was at the IDF (Intel Developer Forum) in Palm Springs where Nvidia rolled out the original GeForce 256 over a nice lunch. (Still have the nice jersey they gave out, too.) While the presentation was very nice there wasn't anything in the GeForce that anyone found terribly surprising. It was inevitable that any high priority processing intensive function will be offloaded from the CPU as chip companies seek greater value for their products. In the case of T&L it was just a matter of when the transistor budget required became affordable in the consumer graphic card market. For the same reasons Intel didn't add dedicated instructions for codecs, although they did offer SIMD instructions under MMX that helped. It made much more sense for the video chip companies to pursue this.
Keep in mind though, this was a reaction to the problem of enhancing the PC. Intel could easily have added functionality optimized for T&L calculations but this would never have made sense from a business perspective since their primary market in the corporate desktop sector would not appreiate paying for these functions. Sony was not in the same situation. They could and did heavily customize an existing general purpose CPU to be optimized for the needs of a 3D game console. In terms of the processing elements that exist within the PS2, while they don't all reside on a video chip, collectively the console has the equivalent dedicated processing elements of a PC with a GeForce prior to the 3rd generation when the pixel and vertex shaders were added. This is the big standout area of the XGPU over the PS2.
The recently released details of the CEll processor show it to be much less remarkable than originally claimed. It is not as once claimed "entirely new" but a derivative of the POWER architecture (really PowerPC but IBM has gotten in the habit of using the the POWER name even for chip that aren't properly part of that family. THe ISA is very close so it doesn't matter much) with multiple cores and some specilized execution units and optimizations for multimedia tasks. Other than the multiple cores it's looking a hell of lot like like an Emotion Engine designed around POWER instead of MIPS and reflecting the differences inherent in half a decade of semiconductor manufacturing advances.
NVidia has already indicated the chip design they're delivering to Sony (unlike the Xbox they won't be producing the chips themselves through their foundry partners) will be substantially different than what will appear in a Windows PC of the same generation. I think they're being a little disingenuous here. It appears what they'll be selling Sony is their 2005 version of the PS2's Graphic Synthesizer. It'll be a more modern design but it will lack a lot of the onboard processing associated with the major jump to DX9/OGL2 level hardware. Instead, like the PS2, those chores will be performed by the Cell chip. Otherwise you're looking at a system that leaves most of its CPU idle most of the time. There just isn't enough work to go around if you have a Cell and a full-on next generation GPU in the box and the box would be overly expensive as well.
It may sound crazy to suggest that you could have too much processing power but in this instance its true. This isn't a PC running half a dozen applications in the background while playing a complex 3D game. This is a dedicated game system where there are limits to the usefulness of extra CPU power if the video chip is doing 90% of the work. You might think that it would make for better AI but the problem there is not one of CPU resources but rather developer resources. There are still very few people who know how to write decent AI and no amount additional CPU time will let them brute force their way through it.
So, the PS3 should be a nice box and all, but I'm not expecting anything awesomely different or superior beyond what a later design freeze date affords.
