X-Authentication-Warning:	delorie.com: mailnull set sender to djgpp-workers-bounces using -f
From:	sandmann AT clio DOT rice DOT edu (Charles Sandmann)
Message-Id:	<10112031444.AA13630@clio.rice.edu>
Subject:	Re: Large memory usage CWSDPMI test available
To:	eliz AT is DOT elta DOT co DOT il (Eli Zaretskii)
Date:	Mon, 3 Dec 2001 08:44:50 -0600 (CST)
Cc:	djgpp-workers AT delorie DOT com
In-Reply-To:	<Pine.SUN.3.91.1011203093834.14864F-100000@is> from "Eli Zaretskii" at Dec 03, 2001 09:39:01 AM
X-Mailer:	ELM [version 2.5 PL2]
Mime-Version:	1.0
Reply-To:	djgpp-workers AT delorie DOT com
Errors-To:	nobody AT delorie DOT com
X-Mailing-List:	djgpp-workers AT delorie DOT com
X-Unsubscribes-To:	listserv AT delorie DOT com

> >   http://clio.rice.edu/djgpp/csdpmi6.zip
> > 
> > This version implements 4Mb pages on hardware that support it (Pentium+
> > without VCPI).
> 
> Should we modify `getpagesize' to sense this and report the right
> size?

No, it's a can of worms.  The test version supports both 4K and 4M pages
at the same time.  If you allocate over 4Mb of memory in a single chunk
each 4Mb aligned page when first touched will be mapped in as 4M page
instead of 1024 4K pages.  If you do small allocations nothing changes
from today.

> Does this change anything else that works by pages?

Well, if you try to use the page protection features (1.0 enhancements)
on a 4K page, but it's in the middle of a 4M page, it silently (I hope)
fails.  My assumption was if you planned to use those features it would
be near a dpmi memory block boundary (which will usually have to be
filled with 4K pages).

This is all just prototype/testing at this point (proof of concept).  
There are some chess enthusiasts who need ultimate speed and maximum
memory who now have 1.5Gb of memory in their PCs and want to use all
of it under DOS to solve chess problems.  They are quite happy with the
new test image (Win 98 is limited to less than 1Gb, even with same memory
DOS is about 10% faster than W98, with more memory they can solve bigger
problems faster).

I'll also start looking at the requested enhancements to speed it up under
VMWARE if I get a chance.  In the past people have told me I should release
the prototypes sooner - to let people comment and test and potentially get
work done while my schedule drags out ...

- Raw text -