Xref:	news2.mv.net comp.os.msdos.djgpp:3209
From:	"James R. Ashley" <jimrthy AT cowboy DOT net>
Newsgroups:	comp.os.msdos.djgpp
Subject:	Re: Pentium C.G. URL
Date:	27 Apr 1996 08:02:03 GMT
Organization:	Lenape Information Systems
Lines:	35
Message-ID:	<4lsk9r$nmg@ns.cowboy.net>
NNTP-Posting-Host:	ppp17.cowboy.net
To:	djgpp AT delorie DOT com
DJ-Gateway:	from newsgroup comp.os.msdos.djgpp

Martynas Kunigelis <martynas DOT kunigelis AT VM DOT KTU DOT LT> writes:
> 
>   Some time ago someone posted an URL of the Pentium Compiler Group and
> said there were patches available to add pentium optimization to DJGPP.
> I tried the URL as given: http://www.goof.com/pcg/ and Netscape said
> 'Could not locate host'. Now, did I mis-type the URL or is it invalid?
> Could someone who knows the exact location give me a pointer?

I'm curious, too.

 Pentium
> optimization is something really necessary nowadays, isn't it?

Um, well, kind of.  You're kind of running down the middle of a 
2 way street.  On the one hand, optimizing important Pent code 
in an important segment is a great way to go, since it also happens
to optimize 486 (no ideas regarding pent pro--sorry), not to mention
loads of fun.  On the other hand, indiscriminitely compiling w/ 
"Optimizing For Asm" (or whatever it's called) enabled has a 
tendency to make the object code larger.

Now, admittedly, code size is a lot less significant under p-mode
than real.  But the pent (?) only (?) has an 8M internal cache.
Keeping inner loops under that limit should significantly increase
performance.  And help executable size.  Etc, etc...basically all
the reasons for optimizing in the first place.

jimerthy

Am I leaving myself wide open tonight, or what?
> 
> TIA,
> 
>    Martynas

- Raw text -