This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: gcc compile-time performance


dewar@gnat.com (Robert Dewar) writes:

> > There is room for some speedup in CPP (10-20%?), but not much IMO.
> 
> But CPP cannot be taking up a significant amount of time??

In GCC 3.0.4, the #1 item in the profile was _cpp_lex_token at about
5% at -O0, even beating out find_reloads.  That has now been fixed, so
we're back to find_reloads as the single routine that takes the most
time at -O0, at 8%; reload as a whole is taking 22% of the total time
for the testcase I was using, tree.c, and just the extra time in
reload accounts for a third of the difference between 2.95 and 3.1 at
-O0.

-- 
- Geoffrey Keating <geoffk@geoffk.org> <geoffk@redhat.com>


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]