X-pop3-spooler:	POP3MAIL 2.1.0 b 4 980420 -bs-
To:	Vincent Diepeveen <diep AT xs4all DOT nl>
Cc:	beastium-list AT Desk DOT nl
Subject:	Re: executable speed.
References:	<3 DOT 0 DOT 32 DOT 19980712173801 DOT 0098b210 AT xs4all DOT nl>
Mime-Version:	1.0 (generated by tm-edit 7.108)
From:	Florian Weimer <fw AT cygnus DOT stuttgart DOT netsurf DOT de>
Date:	13 Jul 1998 18:52:22 +0200
In-Reply-To:	Vincent Diepeveen's message of "Sun, 12 Jul 1998 17:38:03 +0100"
Message-ID:	<m3g1g5lmm1.fsf@deneb.cygnus.stuttgart.netsurf.de>
X-Mailer:	Gnus v5.6.23/XEmacs 20.4 - "Emerald"
Sender:	Marc Lehmann <pcg AT goof DOT com>
Status:	RO
Lines:	20

Vincent Diepeveen <diep AT xs4all DOT nl> writes:

> I compared speeds between compilers for my chess program at my pentium pro
> 200Mhz running linux.
> 
> Speed of 2.7.2 is 10% slower than msvc++

Did you compare FPU-intensive code?  MSVC uses only 53 bits of
precision for floats, compared to 64 bit which is traditionally used
by GNU implementations.

> Now the weird thing: 2.8.1 is *exactly* as fast as 2.7.2, no matter what
> optimizations i give it. This is weird.

Did you compare the executables? Do they differ?

I've recently done some OS (and compiler) comparisions using the HINT
benchmark.  The optimization switches did affect performance, but
sometimes not in the expected direction.  (No, sorry, no details yet,
I want to make sure that I can reproduce the results first.)

- Raw text -