Quote:
Originally Posted by Zero1
I don't know about you guys, but I'd prefer to see more quality tweaks and x86 optimisations before wasting time with PS3.
PS3's hardware will never change, so you are stuck with a set CPU speed; however with a nice optimised x86 build, CPUs are always getting faster and you will feel the improvements for years to come.
I also doubt you will get the full potential of the PS3 too, since Sony seem to like locking down their hardware.
|
It's probable that we'll never get GPU access. What we do have access to, however, is the SPEs and the PPE - those are core to decoding h264 content through the PS3, and are the ones that can be harnessed for (potentially a lot of) encoding performance. It hasn't really been done yet, and that Japanese x264 site/blog does detail the first happenings in that area. I agree the optimizations for this should be done by an independent or otherwise dedicated group (rather than the actual devs), but I also think there's the possibility for Cell to seriously surpass the performance of today's (and possibly the future's) general purpose CPUs -- in the kinds of calculations dealt with in encoding and the like, anyway. I'm sure there are exceptions.
The problem is how to get there. To my understanding, the cell requires a bit of a different philosophy to program for, stressing throughput over simplicity (in code). See
here:
Quote:
|
On a Pentium 4 HT running at 3.4 GHz, this algorithm is able to check 24-million edges per second. On the Cell, at the end of our optimization, we achieved a performance of 538-million edges per second. This is an impressive result, but came at the price of an explosion in code complexity. While the algorithm in Listing One fits in 60 lines of source code, our final algorithm on the Cell measures 1200 lines of code.
|
I'd hate to have to maintain that.
