From: "Werner Stock" <west AT xpoint DOT at>
Newsgroups: comp.os.msdos.djgpp
Subject: AMD-K7 speed optimization
Lines: 16
MIME-Version: 1.0
Content-Type: text/plain;
	charset="Windows-1252"
Content-Transfer-Encoding: 7bit
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Message-ID: <s_cD6.79$D3.4288@news.cwxpoint.at>
Date: Wed, 18 Apr 2001 11:14:44 +0200
NNTP-Posting-Host: 195.2.4.25
X-Complaints-To: abuse AT xpoint DOT at
X-Trace: news.cwxpoint.at 987585560 195.2.4.25 (Wed, 18 Apr 2001 11:19:20 MEST)
NNTP-Posting-Date: Wed, 18 Apr 2001 11:19:20 MEST
Organization: Cable & Wireless Xpoint GesmbH
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Reply-To: djgpp AT delorie DOT com

For developing and training a neural network (lot of float math) in conjunction with a Monte Carlo simulation (lot of int math) I
used DJGPP version PGCC from the Pentium compiler group on my old Pentium-200MMX. I found best speed optimization by using the
compiler flags:
-O6
-mpentium
-fomit-frame-pointer
-fno-force-mem
-funroll-loops
-mpreferred-stack-boundary=2

Next week I'll get a new AMD-K7/1200 powered machine to run the major part (millions) of training cycles. Has anybody experience how
to optimize for speed using DJGPP on AMD's K7 cpu?

Thanks in advance for your tip!
-Werner