Message-Id: <199604031210.HAA08310@delorie.com>
Date:         Wed, 03 Apr 96 14:58:50 LIT
From: Martynas <martynas DOT kunigelis AT VM DOT KTU DOT LT>
Subject:      Re: 4 MB pages
To: DJGPP mailing list <djgpp AT delorie DOT com>
In-Reply-To:  Your message of Tue, 02 Apr 1996 05:25:59 GMT

On Tue, 02 Apr 1996 05:25:59 GMT Peter Johnson wrote:
>
>Does anyone have any ideas on implementing 4 MB paging
>in a DJGPP v2 program?  ASM Sources for realmode were
>recently released along with an article in DDJ (May '96).
>Supposedly they are available only on Pentiums, although I
>have not tested the program on a 486.  Speaking of Pentiums,
>wasn't GCC 2.7.3 supposed to include pentium optimization??
>Or is that scheduled for a later release--might make sense
>to pair it with pentium pro (p6) optimization...
>

 What? is gcc 2.7.3 out? or is this just a typo? anyway, next release should
really have pentium optimization. And also bugs fixed. Like C++ packing. And
I also noticed, that __attribute__((cdecl)) dicards __attribute__((regparm)).
This sucks. Those two combined could make better code. Yeah, and the strenght
reduction fixed as well. Whoops... This is plain off-topic. Sorry... What's
the e-mail address to send stuff like this? gcc-bugs@... or smth?. Is it in
the FAQ, i don't remember?.. Again, sorry for off topic, just felt like saying
this... :)

P.S. and future gcc should also support instruction re-ordering to avoid
     stalls. For now it says: option not supported for the specified target.


Martynas