Reasonable said:
The original idea was, I believe, to not have a conventional GPU at all (imagine the developers then!). However, in the end they struggled with Cell doing the lot and realised they needed to add a GPU of some sort. Some articles I read at the time indicated that while the Cell might have done the lot (I find it doubtful myself) it was also a case of having something more familiar for developers. Either way a GPU was added. The upshot is that, in the worst cases, games just use the PS3 GPU limitations and all (which is why we saw the really bad ports early on, which effectively were using the PS3 in the worst way possible). Things improved as developers changed code to better use PS3 architecture, but even now this is still an issue - for example I'd be willing to bet some of the Bayonetta issues are due to taking code for 360 GPU and just dumping it on PS3, with the resulting performance being poor because only the GPU in PS3 is being used and the Cell wasn't being leveraged at all for graphics. The usual quick fix is reduce graphical fidelity rather than actually change the core code, hence the muddy port. OT - good article, nothing new but at least clear and no bias I could sense. |
For me, the worst offenders were the Ghostbusters devs, saying that they used the PS3 as the core dev platform, and then the game comes out with a lot of graphical problems on the PS3 version, and they say that it's because they had problems with the textures that they didn't have with the 360 version, and the cause was the RAM and the SPEs, revealing that they were lying and didn't use the PS3 as the core platform at all...
