In some ways, the PS2 was quite ancient in it's implementations, in some ways it was very forward thinking though.
An of it's "old way of doing things" was tasking the main processor (even if it was designed to do so from the outset) with geometry processing. I'm sure Sony (specifically Kutaragi) was wanting to see what developers could do with the two vector processors, not only from a graphical standpoint, but for physics, and other processes necessary to create high realized game worlds, hence probably why they were put on the Emotion Engine. Also the PS2's development was made during the beginning days of true 3D Accelerator development on PCs. Even the Dreamcast did it's geometry on it's main CPU via the FPU. The other "ancient" issue was the actual "graphics processor", the Graphics Synthesizer. While it possessed an incredible pixel throughput (over 2 GPixels), the lack of texture units severly regulated the system to multipassing when the need arose and I'm sure it made it all the more difficult to implement things like bump mapping, etc, especially when vertex (or vector units in the PS2's case), necessary for such graphics techniques, were not directly available, on the GPU itself. Even still, the PS2 was more than capable of such graphics techniques, and it's unfortunate they were never used to any real extent except in a few select titles.
As for the forward thinking bits, the PS2 was of course, the first DVD based system which obviously made it a prime candidate as a hit DVD player. The PS2 implemented an EDRAM system to give the graphics system a very fast (48 GBPS IIRC) connection to the PS2's 4 MB of EDRAM buffer. Especially for instancing, this was a very good feature to have and made the PS2 quite good with many special effects and helped with the very small amount of main RAM the system had in the first place (32 MB). The crown jewel of the PS2 (to me at least) was the Emotion Enginen main processor. While difficult to work with initially, as a single chip, it was a beast, containing not only a MIPS based CPU, but an FPU and the two Vector Units which of course could and did everything from geometry, lighting, physics, etc. It was also an early experiment at multithreading and multicore processing since it would be a few years later that the first consumer dual core CPUs would be released, so it is highly significant in that regard. It's just unfortunate that the EE was stuck doing so much of the work when you think about it's competitors which opted for a more "seperated" approach, with complete CPUs and GPUs allowed to focus on their own thing. At least the PS2 was a highly capable computing machine at it's release, and even against the Xbox and GC still had the highest raw geometry throughput, and possessed by far the most powerful computing processor (the EE). EE I think was likened to being twice as powerful as 2 of the Xbox's CPUs (basically a 733 MHz Pentium III). Considering the addition of the vector units, that doesn't seem farfetched at all. An EE based console, with 128 MB of memory, and a true 3D GPU a la the Xbox's would've been amazing to see, especially when you consider the physics potential of the vector units. They were also used for fluid physics for a few games like Ghost Hunter and Ico, something I don't remember ever being attempted on the Xbox or GC.
And compared to the Dreamcast, there are only two things I can think of the DC being superior to the PS2 in: Sound (PS2's 48 Channels vs the DC's 64), and arguably the texturing performance since the DC had 8 MB of dedicated video memory as well as dedicated texture compression capability. However, plenty of PS2 games had pretty good texturing.
PC_Otter
XBox1 relies on NVIDIA NV2A GPU and nForce/Soundstorm APU (Audio Processor Unit).
According to wiki, NVIDIA NV2A's geometry engine has a peak of 115 million vertices/second and 125 million particles/second.
Log in to comment