On Sun, May 15, 2005 at 11:50:02AM -0700, Mike Stump wrote: > how we are progressing gcc to deal with hardware like this. To the > extent technologies like that can make it perform well, I think that is > the direction gcc is headed, beyond that, if you are interested in > doing the work, you'd nee to start a discussion on the remaining > unsolved problems and how to advance things to solve those issues. i think you may find that a less stringent goal - of doing "outsourcing" - may result in an intermediate useable compromise that would keep most people happy or at least a whole damn lot more happy than they are at the moment.
after six months of training to use an ASP, i was able to write 50-line programs in about three weeks, after at LEAST that long in talking to the experts at Aspex as to how to arrange the DMA scheduling with the data spread across units [utilise bit-parallelism across many units or utilise individual APEs over many clock cycles]. the Sony Playstation's 4-wide floating point vector unit is JUST as xxxxing awkward to make effective use of, and that's just 4 wide, not 4096 wide!!! and the key reason why all these things are a pain in the arse is because you cannot "abstract" it out to c++ - you _have_ to go to assembler, you _have_ to make use of macros (which don't mix with c++ templates). ... the other thing i should mention is that i haven't really a clue why i'm bringing this up (well, i do, but it's many _many_ steps removed from becoming a reality - funding, blah blah) as i stopped working for Aspex about three years ago :) ... you can tell that i really loved the processor design and the opportunity to work with something that radical, though. l. -- -- <a href="http://lkcl.net";>http://lkcl.net</a> --
