This is the mail archive of the
gcc-prs@gcc.gnu.org
mailing list for the GCC project.
Re: libstdc++/4150: catastrophic performance decrease in C++ code
- From: rth at gcc dot gnu dot org
- To: gcc-bugs at gcc dot gnu dot org, gcc-prs at gcc dot gnu dot org, leitner at fefe dot de, nobody at gcc dot gnu dot org
- Date: 12 Apr 2002 00:19:06 -0000
- Subject: Re: libstdc++/4150: catastrophic performance decrease in C++ code
- Reply-to: rth at gcc dot gnu dot org, gcc-bugs at gcc dot gnu dot org, gcc-prs at gcc dot gnu dot org, leitner at fefe dot de, nobody at gcc dot gnu dot org, gcc-gnats at gcc dot gnu dot org
Synopsis: catastrophic performance decrease in C++ code
State-Changed-From-To: open->analyzed
State-Changed-By: rth
State-Changed-When: Thu Apr 11 17:19:06 2002
State-Changed-Why:
Still present in 3.1. We're seeking unnecessarily, even after
adding a sync_with_stdio(false) to the source (which I now attach
to the PR). Strace shows
_llseek(4, 8192, {8192}, SEEK_SET) = 0
read(4, "AOL (Traffic-Server/1.1.6 [1])\""..., 8192) = 8192
_llseek(4, 8192, {8192}, SEEK_SET) = 0
_llseek(4, 16384, {16384}, SEEK_SET) = 0
read(4, "cur=0&skip=10\" \"Mozilla/4.0 (c"..., 8192) = 8192
_llseek(4, 16384, {16384}, SEEK_SET) = 0
which indeed agrees with what basic_filebuf<>::underflow is
doing. To my mind this is appalling waste. Surely this is
not necessary when sync is off?
http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=4150