Re: libstdc++/2211

[Paolo Carlini <pcarlini at unitus dot it>]

The following reply was made to PR libstdc++/2211; it has been noted by GNATS.

From: Paolo Carlini <pcarlini@unitus.it>
To: bkoz@gcc.gnu.org, gcc-gnats@gcc.gnu.org, gcc-prs@gcc.gnu.org,
 	gcc-bugs@gcc.gnu.org, Theodore.Papadopoulo@sophia.inria.fr
Cc:  
Subject: Re: libstdc++/2211
Date: Wed, 27 Jun 2001 23:27:51 +0200

 Hi all,
 
 Benjamin, I have just read your reply concerning the use of
 std::ios::sync_with_stdio(false).
 Therefore I tried my favorite test, that is a short program from the 2nd
 Ed. of Programming Pearls by Jon Bentley:
 
 -----------
 /* Copyright (C) 1999 Lucent Technologies */
 /* From 'Programming Pearls' by Jon Bentley */
 
 /* wordlist.cpp -- Sorted list of words (between white space) in file */
 
 
 #include <iostream>
 #include <set>
 #include <string>
 using namespace std;
 
 int main()
 {
         set<string> S;
         string t;
         set<string>::iterator j;
         while (cin >> t)
                 S.insert(t);
         for (j = S.begin(); j != S.end(); ++j)
                 cout << *j << "\n";
         return 0;
 }
 ------------
 
 
 On my Linux-x86 box, I don't see significant speed differences bring
 about by std::ios::sync_with_stdio(false).
 
 Indeed, I always see a significant run time regression vs libstdc++v2
 (between 2x and 3x, unfortunately) when processing a long test file
 (usually I use electronic version of "The King James Bible",  4445260
 chars, for this purpose)
 
 http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view&pr=2211&database=gcc