Not sure if I'm supposed to read into that or not Archie, but the concept is intriguing.
Still I'd hope on Sony's end that there was a resolution other than full EE+GS inclusion, simply because that would seem to me to an inelegant - if effective - resolution resulting in a non-trival cost being added to the system. Granted, I understand that Sony probably would view such inclusion as a one-gen bullet to bite, since future PS versions will likely facilitate B/C through extending the Cell and NVidia architectures the current console is built upon.
Relative to What? A fabled PS2 title written by monkeys hired from MS?
PS2 average texture size is 4/8bits texel, DXTC gives you the exact Same numbers, and many of us have taken that into lower ranges where we could. Converting an arbitrary PS2 game from the "no compression" to all DXTC compression, would gain you approximately nothing in terms of memory usage.
Also, converting PS2 assets into DXTC would make things look very ugly (double lossy compression) - although on the other hand, it should be interesting how emulator would handle paletted textures because PC/NVidia hardware has no support for them.
Anyway to be fair, given relative cache sizes, I doubt PS2 game texture acceses would even make a dent in main memory bandwith.
xdestroya said:
Not sure if I'm supposed to read into that or not Archie, but the concept is intriguing
Well who knows - at least with PS2 they didn't mind changing the hw when they could keep functionality the same (latest PS2 revision no longer has PS1 main CPU in there).
Well who knows - at least with PS2 they didn't mind changing the hw when they could keep functionality the same (latest PS2 revision no longer has PS1 main CPU in there).
Yeah I thought about that actually when Archie presented the dual-posssibility, but still I mean there's got to be some sort of perma-hw consideration in there to replicate the GS bandwidth. Maybe if there's an EE+GS in there now, that just means a straight GS later. Anyway I don't know... certainly I find all the RSX transistor theories interesting though; I guess I would like the answer to lie there if only for added novelty to the GPU solution.
It's of course amusing to reflect and know there are a couple of people floating around the forum that likely have the answers to all these questions (yourself or Archie perhaps?), but hey I just have to work with what I've got!
Well who knows - at least with PS2 they didn't mind changing the hw when they could keep functionality the same (latest PS2 revision no longer has PS1 main CPU in there).
????
I cant see the logic behind that, the IOP is neccesary for PS1 Games and PS2 Games. They cant remove it without replacement. And why would they replace it with another Processor that in turn will have to emulate the IOP all the time?
Cost. IOP isn't owned by Sony like the other chips.
Afaik the replacement chip is inhouse - cheaper to make by default, and could be integrated with the rest of PS2 chipset onto a single die one time in the future.
Dint knew that. Especially since IOP was a rather unspectacular MIPS R3000 design, and the PS1 specific parts already were emulated (Coprocessor GPU ).
The reason is the quality of emulator. SONY has to add the real hardware for B/C. Once the emulator complete, the EE+GS part can be removed for cost down.