This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: cache misses in gcc 3.3
- From: dewar at gnat dot com (Robert Dewar)
- To: ak at muc dot de, gcc at gcc dot gnu dot org, mszick at goquest dot com
- Date: Mon, 10 Feb 2003 09:56:45 -0500 (EST)
- Subject: Re: cache misses in gcc 3.3
> > I suspect using an handwritten top-down parser would be somewhat faster
> > here because it encodes the grammer in code, not data and I would expect it
> > to have a smaller working set and be less affected by data caches
> > overflowing.
Parsers should be so fast that algorithms are not really an issue. It
is how the algorithm is written that is important. If a compiler is
spending any noticeable amount of time parsing (as opposed to lexical
analysis or semantic analyis), then it is certainly a target for
improvement (the parser in GNAT for example takes a very small fraction
of the total time for compilation).
One thing that is really important in going after speed improvements is
to have really accurate profiling information that shows where the time
is being spent. Otherwise you will chase phantoms.