Message-Id: <3.0.32.19990720135743.010ff0e0@pop.xs4all.nl>
X-Sender: diep AT pop DOT xs4all DOT nl
X-Mailer: Windows Eudora Pro Version 3.0 (32)
Date: Tue, 20 Jul 1999 13:57:56 +0100
To: pgcc AT delorie DOT com
From: Vincent Diepeveen <diep AT xs4all DOT nl>
Subject: RE: pgcc port for WinNT
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Reply-To: pgcc AT delorie DOT com

At 11:05 AM 7/20/99 +0200, you wrote:
>> 	Good luck beating VC++ on MS's own OS.  You're getting 
>> that we're at an
>> unfair advantage here.  MS knows all about the OS and ALL of 
>> it's APIs and
>> such, we don't.
>
>Believe me or not, MSVC generates even better code if you don't use the API.
>BTW. I dont believe that the API knowledge helps in any case, but...
>
>VC 6 is even better than the Intel 4.0 compiler with PIII optimization
>enabled
>in our testsuite (for integer, for FP Intel compiler seems to be faster).
>So it might be a long way to beat VC :-(

intel compiler for my integer chessprogram (completely 32 bits
nowadays) was with all optimizations on (except PRO/PII/PIII instructions)
about 20% slower than msvc

with pro instructions turned on it was only 1.5% slower than msvc,
which makes it the fastest compiler after msvc++ for my program.

However, considering it only runs on the same platform as msvc
this doesn't make much sense.

Compiling with intel compiler takes about 10 minutes.
compiling with msvc takes a minute.

Even if it would compile for 10 hours, i wouldn't mind that,
as long as it produces faster code, but it doesn't!

Secondly all handy tools don't work for it.

>--
>Michael Beck,                        email: beck AT dresearch DOT de 
>
>