This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: Tackling library IO performance
- From: Nathan Myers <ncm-nospam at cantrip dot org>
- To: libstdc++ at gcc dot gnu dot org
- Date: Sun, 5 Jan 2003 23:02:36 -0800
- Subject: Re: Tackling library IO performance
- References: <E18VQov-0006qn-00@tiamat>
On Mon, Jan 06, 2003 at 01:33:09AM -0500, Jerry Quinn wrote:
> Hi all. I've been investigating the source of slowdowns in PR 8761.
> On my machine, gcc 2.95.4 -O2 runs the testcase in about 4 seconds.
> gcc 3.3 -O2 runs it in about 25 seconds.
> ...
> I wanted to get some feedback before I crank out patches that people
> don't want, especially since the format cache had been removed once.
> Are there things I should be careful of or avoid doing in putting
> these patches together?
This is really excellent work. There's lots of room for optimization
when you know that it's the "C" locale.
One thing to watch out about... people get very testy when floating
point values come out (or in) slightly different, in the least
significant bits, from what their C library does. Maybe we're
Big Enough now to just match Glibc, and tell everybody else to like
it or lump it.
Nathan Myers
ncm@cantrip.org