This is the mail archive of the gcc-prs@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]

c++/5001: g++ 3.0.2 getline is too slow



>Number:         5001
>Category:       c++
>Synopsis:       g++ 3.0.2 getline is too slow
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Mon Dec 03 17:06:00 PST 2001
>Closed-Date:
>Last-Modified:
>Originator:     Uwe F. Mayer
>Release:        3.0.2
>Organization:
tux.org
>Environment:
System: SunOS hydra 5.6 Generic_105181-16 sun4u sparc SUNW,Ultra-2
Architecture: sun4
host: sparc-sun-solaris2.6
build: sparc-sun-solaris2.6
target: sparc-sun-solaris2.6
configured with: ../gcc/configure --prefix=/work/hyd01/td/sw/gcc_3_0_2_release --disable-nls --disable-threads --disable-shared --enable-languages=c,c++,f77,objc
>Description:
The getline() call of g++ version 3.0.x is slow if compared to previous
versions of g++ (for example, version 2.95.3). I tried this on various
SunOS boxes, running SunOS 5.6 or 5.8, and g++ versions 3.0.1 and 3.0.2,
having g++ compiled with single-threads only, or compiled with
multi-thread support.

A sample program exhibiting this problem is:

#include <iostream>
#include <fstream>
#include <cstdio>  // for BUFSIZ=1024

using namespace std;

int main (int argc, char * argv[])
{
  if (2!=argc) { cerr << "Please provide a filename\n"; }
  ifstream in(argv[1]);
  char buffer[BUFSIZ];
  while( ! in.eof()) {
    in.getline(buffer, BUFSIZ);
  }
}

If compiled with g++ version 3.0.2 this program reads 100,000 lines with
79 characters (plus newline) each in about 4.5 seconds. On the same machine,
if compiled with g++ version 2.95.3, the program needs for the same input
file only about 0.25 seconds.

A similar program in C using fgets() does NOT have this slow-down problem
with gcc version 3.0.2.

>How-To-Repeat:
See the Description field above.
>Fix:
I am guessing some kind of synchronization problem connected with the
buffering strategy, but I might be way off target...

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


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