Mail Archives: djgpp/2000/10/19/17:15:14
Hans-Bernhard Broeker :
> *How much* slower? It's really hard to speak about optimization
> differences if their size is not known. Is it: a percent slower?
> Fifty? Taking 5 times as long?
No no, be reassured: a few percents.
> And what's that code doing, in the first place?
Basically, a 3D engine and a unzipper without ASM insertions.
I could annoy you with frames-per-second log files but well...
I'm not writing a very serious article in Journal of Computational
Physics about it. I was just curious.
> '-funroll-loops' hardly does you any good, on any x86 type machine. It
> increases the code size, and as soon as the amount of code being
> looped over (the 'active set') goes beyond the size of the 1st level
> Cache, you'll receive a noticeable performance degradation. Same as
> you cross other size barriers. Pentium-class machines are good enough
> at branching (and branch prediction, in particular) that unrolling
> loops doesn't gain you terribly much, anyway, before the cache
> coherency loss strikes back.
I'm well aware about the bad cache windowing when code size increase,
especially when you let the auto-inlining option on, but trust me,
in *that* case '-funroll-loops' has a positive effect (up to 10%).
Anyway, thank you.
G.
Sent via Deja.com http://www.deja.com/
Before you buy.
- Raw text -