Re: libstdc++/9606: performance regression in operator<<(ostream&,int)

Re: libstdc++/9606: performance regression in operator<<(ostream&,int)

[paolo]

Synopsis: performance regression in operator<<(ostream&,int)

State-Changed-From-To: open->closed
State-Changed-By: paolo
State-Changed-When: Fri Feb  7 00:49:21 2003
State-Changed-Why:
    Unfortunately, known, complex issue (see libstdc++/8761
    and libstdc++/7076).
    Benjamin, Nathan and Jerry Quinn are on it. 3.3, hopefully
    will be quite a bit better. Thanks for your patience!

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=9606

Re: libstdc++/9606: performance regression in operator<<(ostream&,int)

[Joe Buck]

The following reply was made to PR libstdc++/9606; it has been noted by GNATS.

From: Joe Buck <jbuck@synopsys.com>
To: paolo@gcc.gnu.org, gcc-bugs@gcc.gnu.org, gcc-prs@gcc.gnu.org,
   nobody@gcc.gnu.org, gcc-gnats@gcc.gnu.org
Cc:  
Subject: Re: libstdc++/9606: performance regression in operator<<(ostream&,int)
Date: Thu, 6 Feb 2003 17:43:22 -0800

 On Fri, Feb 07, 2003 at 12:49:21AM -0000, paolo@gcc.gnu.org wrote:
 
 > State-Changed-From-To: open->closed
 >     Unfortunately, known, complex issue (see libstdc++/8761
 >     and libstdc++/7076).
 
 It's appropriate to close as a duplicate of 7076, but it doesn't
 seem to have anything to do with 8761.

libstdc++/9606: performance regression in operator<<(ostream&,int)

[jbuck]

>Number:         9606
>Category:       libstdc++
>Synopsis:       performance regression in operator<<(ostream&,int)
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Thu Feb 06 23:46:00 UTC 2003
>Closed-Date:
>Last-Modified:
>Originator:     Joe Buck
>Release:        gcc-3.2.1 (actually, all gcc-3.x releases)
>Organization:
>Environment:
i686-pc-linux-gnu
>Description:
Compile and run the test program below with
g++ -O3 prog.C -o prog

and time it.  For gcc 3.2.1 on a Xeon 2.2 GHz box,
it takes 17.39 sec, but for gcc 2.95.3, it takes
2.06 sec.

It seems that the difference is that snprintf is
now used.

>How-To-Repeat:
// this is the test program.
#include <fstream>

int main(int argc, char** argv) {
    std::ofstream sink("/dev/null");
    for (int i = 0; i < 10000000; i++)
	sink << i;
    return 0;
}
>Fix:

>Release-Note:
>Audit-Trail:
>Unformatted: