Re: Possible data allocation and instruction scheduling algo...

Ray Dillinger <bear@sonic.net> wrote in message news:02-02-031...
> It's getting seriously hard for a compiler to really know what the
> hell its target is on the local machine, and even if you know what the
> target is, sometimes the tradeoffs are hard to predict when deciding


        When in doubt, optimize for the Pentium I. Obviously if you're
programming for multimedia (ie: libraries bundled with the compiler),
you'll want to take advantage of the latest MMX/etc instructions. BTW,
I've made optimizations for the Pentium that actually ran faster than
the supposedly optimal equivalent.




> how to optimize something. Lately I've been toying with the idea of
> ...
> along in a config file.


        That's a good idea in theory. One problem is that most programs
have functions that are weaved together with other functions, so you
might end up profiling an entire program several times. Another
problem is that you'll have to "simulate" data; such as dynamic
memory, modifying x bytes in y buffer, etc.




See ya!
Orlando

Return to the comp.compilers page.
Search the comp.compilers archives again.