From: Donovan Rebbechi <elflord AT panix DOT com>
Newsgroups: comp.os.msdos.djgpp,comp.lang.c++
Subject: Re: ostringstream vs. ostrstream performance
Followup-To: comp.lang.c++
Date: 25 Nov 2001 22:50:03 GMT
Organization: PANIX Public Access Internet and UNIX, NYC
Lines: 35
Message-ID: <slrna02tcr.3q0.elflord@panix2.panix.com>
References: <3C0100AC DOT 93129C4B AT bigfoot DOT com>
NNTP-Posting-Host: panix2.panix.com
X-Trace: news.panix.com 1006728603 20086 166.84.1.2 (25 Nov 2001 22:50:03 GMT)
X-Complaints-To: abuse AT panix DOT com
NNTP-Posting-Date: 25 Nov 2001 22:50:03 GMT
User-Agent: slrn/0.9.7.2 (NetBSD)
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Reply-To: djgpp AT delorie DOT com

In article <3C0100AC DOT 93129C4B AT bigfoot DOT com>, Alex Vinokur wrote:
>===============================================================
> Windows98
> gpp : GNU C++ version 2.95.3 20010315/djgpp (release) (djgpp)
>       compiled by GNU C version 2.95.3 20010315/djgpp (release).
>===============================================================
> 
> 
> Hi,
> 
> Here are some results of comparing performance :  'ostrstream vs.
> ostringstream'.
> It seems that the results are weird :
>  ostringstream::operator<< is improbably slower than
> ostrstream::operator<<.
> 
> Any explanation?
> Something wrong in this experiment?

First, it doesn't surprise me that much, because with strstream, because
there's less indirection.

However, your methods could do with some improvement.

For example, it's probably better to put the two in seperate programs,
and time them using the "time" command. This tells you about how
much CPU the program uses. The issue this addresses is that programs
may slow down over time, because the operating system deprioritises 
them to prevent them eating all the CPU.

Another test that may be interesting is profiling your program, to get 
some more precise insights as to why <sstream> is slower.

-- 
Donovan