From: "Andrew Crabtree" <andrewc AT rosemail DOT rose DOT hp DOT com>
Newsgroups: comp.os.msdos.djgpp
Subject: Re: djgpp and optimizations
Date: Wed, 16 Sep 1998 16:00:07 -0700
Organization: Hewlett Packard
Lines: 32
Message-ID: <6tpftm$iif$1@rosenews.rose.hp.com>
References: <000301bde180$e0d71f20$bb4d08c3 AT arthur>
NNTP-Posting-Host: ros51675cra.rose.hp.com
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Precedence: bulk


>So how come I get such a poor performance when using -O7 (and other PGCC
>stuff), whilst I get a huge speed boost using -mpentium (the gcc 2.8.1
>version, not PGCC)?
Dunno.  I can take several guesses, but some or none of them may apply.
Here goes

1) PGCC is very picky about alignment.  If you didn't fix the alignment in
the crt you could have bad problems

2) Some of the options after -O4 are broken and generate slower code.  Some
of them are tailored for PPro and generate slower code for Pentiun, and vice
versa.

3) -O7 can do strange things with code bloat, and you could be running into
cache functions.

For pgcc to really be effective you generally have to go through a
combination of the individual options until you find what works best with
your code and your cpu.   You might want to try -fno-risc
and -fno-inline-functions to see if that helps.

>And why on earth doesn't PGCC cater for the MMX chip?
Adding MMX support into the compiler doesn't happen all that easily.  There
is somebody adding mmx support to gcc but it is separate from the pgcc
project.  There are many difficulties involved in doing this.  Not sure what
state it is in.

Andy