Vysez said:
I must say that anything might end up better than VU0 since it's at best "not what it was supposed to be" or worse it's broken. People said that it was the result of some "last minute" cutdowns in the EE architecture, others (DM) said it was because K.Kutagari hates programmers and wants them to suffer for years, looking for solution for the VU0.
Actually according to Deadmeat VU1 is the broken last minute addition one. And inspite the fact that anyone who ever touched the system tells him otherwise, he continues to claim it.
That said, VU0 isn't broken, but it lacks what APU has (internal ability to trigger in/out memory transfers) to reach full potential.
Anyway, I've only browsed through the first patent so far, but I can't see what Deadmeat is freaking out about (aside for it involving Kutaragi, Ken!). It looks no different then programming various shaders to me, and very much like what I expected based on other patents.
It still doesn't tell me anything about what they intend to do with rasterizing side though. Contrary to some people here, I haven't seen any evidence to support those "sotware rasterizer" claims yet, actually rather the opposite if I am to believe that "layer renderer" patent.
IF the ps3 HAVE TO be programmed without passing throught a "layer" that abstract the "metal" (can't express that right in english), then it might* appears as a serious complication for developers.
Depends what exactly you expect from an abstraction layer to do in the first place.
According to DM, the only abstraction layer that is "good enough" is one that makes your entire program look like it was coded for a single CPU (which means the Xenon won't even come close to satisfy his requirements either).
Again, haven't read the other patents yet, but nothing here implies yet that you will have to explicitly control individual APUs or their load distribution. Although in some cases that might actually be nice to have as an
option.
