delorie.com/archives/browse.cgi   search  
Mail Archives: opendos/2000/11/30/07:55:46

X-Apparently-From: <pmoran22 AT yahoo DOT com>
Message-ID: <004c01c05a76$4732db90$c5881004@dbcooper>
From: "Patrick Moran" <pmoran22 AT yahoo DOT com>
To: <opendos AT delorie DOT com>
References: <DDCDC9DE5EC0D411AE7B0090273F74A412C9C7 AT emwatent02 DOT meters DOT com DOT au>
Subject: Re: BASIC & EMS (was: Optimizing CONFIG.SYS...)
Date: Wed, 29 Nov 2000 19:19:26 -0700
MIME-Version: 1.0
X-Priority: 3
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook Express 5.00.3018.1300
X-MimeOLE: Produced By Microsoft MimeOLE V5.00.3018.1300
Reply-To: opendos AT delorie DOT com

----- Original Message -----
From: "Da Silva, Joe" <Joe DOT DaSilva AT emailmetering DOT com>
To: <opendos AT delorie DOT com>
Sent: Sunday, November 26, 2000 7:11 PM
Subject: RE: BASIC & EMS (was: Optimizing CONFIG.SYS...)


> Thanks for the info. on M$ BASIC's origins, BTW ...
> I'll take a look at Darmouth's stuff, soon.
>
> As a side-note, anyone ever seen/used a Hitachi Peach
> computer (ca. 1980)? This had an O/S from M$, partly in
> machine code, partly in BASIC. Very primitive stuff - and
> EXTREMELY slow! For example, say you wanted to duplicate
> a disk - it took 20 minutes to format it, 20 minutes to copy
> it and, finally, another 20 minutes to verify it!!! No, I'm not
> kidding - that's actually how long it took!!!

I have worked with similar systems. We had a system with 2 8" floppy drives
and it would take about 20 to 30 minutes to load a 8k program. These drives
were connected via serial port and had a baud rate of 4800 if I remember
correctly. Never tried to copy a diskette on that thing.

Commodore C-64/128 also used serial port for their floppy drives, but the
DOS was on ROM in the drives. It could take 30 minutes to an hour to
duplicate a copy protected diskette that would take about a minute or less
on my APPLE ][+ for the same program. BTW FYI, the copy protction scheme
used most by Commodore diskeetes was an
APPLE ][ formatted floppy! <BG>


> W.R.T. EMS, I recall reading somewhere that, rather than being
> "Extremely inefficient", EMS is actually more efficient than XMS
> (when based on either real hardware or the 386's magic tricks).
> There were also "EMS simulator" software (286 stuff?), which
> had to copy data to/from the page frame, hence these were
> indeed inefficient - perhaps you were thinking of these?

That could be or maybe I am thinking of the way MS crap for DOS 5 did it.
You first had to have Extended memory then that is converted to EMS memory.
However, no matter how it is done, you have a tiny little 64K window of
memory to shove stuff through from expanded memory to conventional memory
and conventional memory to expanded memory. When you use something like
Desqview and you have a large program over 500k in size and have to swap
portions of it out 64k at a time in 16k chunks.

XMS. When a program is run in protected mode, there is no swapping, it
operates in exntended memory as originally designed to do. In virtual 86
mode you can actually have a full 8086 computers installed many times and
each can share entended memory. This is what the DRDOS Task manger does. It
does not use EMS like Desqview and other so-called multitaskers do via EMS
memory.

I am not positive what XMS memory is, it is not talked about in "DOS Beyond
640K" book. I do remember something about QEMM converting Extended memory to
EMS then converting EMS memory back for use as Extended memory, which is
really, really stupid. Why not just use Extended memory directly like DPMS
does?????

> The only real problem with EMS, is the 64K hole it makes in
> the UMB region (sometimes 32K, as noted by M.P.). Perhaps
> (I don't know how much work is involved), Matthias could add
> "cloaking" to EMM386 - that would help GREATLY (bug fixes
> are more urgent, of course)!

I don't know what you mean by 32K, EMS LIM 4.0 requires 64K page frame. You
can use larger page frames, but EMM386 only uses four 15K pages for 64K page
frame. AST once made a computer with no conventional memory! Everything in
that computer used EEMS memory. LIM 4.0 is mostly based on the AST EEMS
memory specifications but because of all the companies involved and their
stupid egoes we got stuck with an inferior system called LIM 4.0.

> Also, I wonder if, on machines which can use F000-F7FF for
> EMS, this address space can simply be added to the UMB
> "pool" instead?

It sure can. Look at my previous message where I show my CONFIG and AUTOEXEC
and MEMORY MEM/AP readout. You will see that I use that area, although on my
system I can only use F000-F6FF because of the 1A handler. I'll try Paul's
suggestion and see if I can get more.

The only things I have had problems with using Task manager is the stupid
mouse driver and you can't switch windows when you are in graphics mode.
This is understandable because graphics mode uses the A000-AFFF address
range in lower memory. If that could be used in Extended memory then that
would not be a problem. DV also had problems with graphics mode and switch
windows. Some could be fixed but others could not. The bleed through I could
handle in most cases.

Try play DOOM in DV then try playing it in Task Manager on a boarderline
speed computer for playing that game, then you will see why EMS sucks big
time. Have it as your only task. I can probably do that on this system if I
turn turbo off. It should be about a DX2-66 that way, which is the minimum
recommended system for DOOM.

BTW Linux beats them all anyway.

I was reading something on the WEB the other day that said something about
true multitasking systems. They included NT under that category, I had to
laugh out loud when I read that. NT is not a true multitasker. I could do
more and faster with no delays on my old 386DX40 MB with * meg RAM under
Linux that I can do with this 5x86-133 with 48MB RAM with NT and I never had
things slow down they way they do with NT with 6 times the memory and over
100MB of virtual memory. I only had 24MB of virtual memory under Linux until
one day I actually ran out of memory then I bumped it up to 32MB on the fly
and kept on trucking. I still get out of memory with NT and 100MB pf virtual
memory! Bloated crap for software!!!!

This system under Linux really screams. For MS crap for bloated software I
would need a 10GHz MB, 4GB RAM, FSB of 1GHz and a 40 terabyte HD sapce on a
160MBs RAID controller to get even decent speed and still probably still
would not keep up with Linux on this system!!!!

Pat



_________________________________________________________
Do You Yahoo!?
Get your free @yahoo.com address at http://mail.yahoo.com

- Raw text -


  webmaster     delorie software   privacy  
  Copyright © 2019   by DJ Delorie     Updated Jul 2019