[comp.os.msdos.apps] frustration with memory problems

logan@neuro.usc.edu (Christy Logan) (04/15/91)

  I am experiencing a great deal of trouble and frustration using 
my new 386sx and would really appreciate some help.  I should begin by
saying I am a novice and don't really understand the ins and outs
of memory (nor do I want to--I just want things to work).  
Windows 3.0 is installed on the system and I have not altered
the way it set up the config.sys file (except to add the mouse
as a driver).  My system has 4 MB RAM.

  The problem I am having is that when I try to run Sigmaplot
(a graphing program that works fine on my XT with no extra 
memory in lab), it says it can't load graphics drivers because
there's not enough memory (therefore it doesn't work at all).
That happens whether I try to run it in or out of Windows
(although once when I tried to run it in Windows
it almost worked, but gave me the same error message
when I go to the point of printing the graph.  I don't have any idea
what had changed in between that run and other tries.)
Sigmaplot seems to think there is 115K of memory available.  
Another program I run, CSS (a statistical package) runs all right
but thinks there is 365K of memory available.  Why the difference?  
But more importantly, why can't Sigmaplot find/use the memory?  
What good is 4M if my programs can't use it?  

  When I called the computer's manufacturer (ARC) with this problem,
they suggested buying QEMM, and said that the problem was with the
way Windows set up memory.  Besides the fact that I feel cheated
that I have to go out and spend more money just to get my fancy
new machine to do what my crummy old machine could do, I am
also concerned about trying QEMM because there are so many
problems with using it posted here.  So what's the verdict on
QEMM?  I can't tell if the problems result when people want
to do more sophisticated things than I'd be likely to want.

  I am also experiencing other more minor problems, like starting
up Windows and not having my mouse work, then exiting and restarting
and having it work fine.  No big deal really, but an annoyance
nonetheless and one that makes me think that the system is
poorly designed and just limping by (whether the "system"
is the hardware or the DOS or Windows software is another
question I'd like answered).  

  All in all, I am extremely frustrated, to the point that
I am considering returning my computer (still in its 30
day free return period) and buying a Mac.  I've never used
one, but everyone who owns one seems to love them, something
that I haven't found among IBM users.  Certainly it would be
preferrable to get my 386sx to work, but I hate the idea of just
patching things together and settling for a way of getting it
to limp along.  Isn't there a way to make it *really*work*?
Is Windows 3.0 inherently flawed?

Thank you for allowing me to vent my spleen.  Any suggestions 
or advice would be greatly appreciated.
--Christy Logan

bchin@umd5.umd.edu (Bill Chin) (04/15/91)

In article <31945@usc> logan@neuro.usc.edu (Christy Logan) writes:
>[some parts of the description removed]
>Sigmaplot seems to think there is 115K of memory available.  
>Another program I run, CSS (a statistical package) runs all right
>but thinks there is 365K of memory available.  Why the difference?  
>But more importantly, why can't Sigmaplot find/use the memory?  
>What good is 4M if my programs can't use it?  

Well, you're caught with the difference between extended and 
conventional memory.  First off, SigmaPlot 4.0 (that's the version
I had to deal with) likes lots of conventional memory... the stuff
below 640k.  The amount of memory available is shown with the "mem"
command under DOS 4.0x.  The amount returned by CSS and SigmaPlot
is the amount left after the respective programs have loaded, ie.
this is data space left.  SigmaPlot likes to have ~540k of 
conventional memory free before it loads.  A 100kb of expanded memory
helps too. 

>  When I called the computer's manufacturer (ARC) with this problem,
>they suggested buying QEMM, and said that the problem was with the
>way Windows set up memory.  Besides the fact that I feel cheated
>that I have to go out and spend more money just to get my fancy
>new machine to do what my crummy old machine could do, I am
>also concerned about trying QEMM because there are so many
>problems with using it posted here.  So what's the verdict on
>QEMM?  I can't tell if the problems result when people want
>to do more sophisticated things than I'd be likely to want.

The problem isn't really Windows... It's using DOS programs
that hog conventional memory.  For example, a Windows program
on your system can use many megabytes of memory, or a DOS
program that knows how to use XMS memory can do the same.
So until SigmaPlot 5.0 that either runs as a Windows app
or uses DOS resources better, here are a couple of solutions.

- QEMM/386-to-the-Max/Netroom etc. are programs that take advantage
of your 386 and can give you more conventional memory through
some memory management tricks.  Typically $75 bucks, if you
run lots of memory intensive DOS apps that insist on conventional
memory or have big network drivers, its money well spent.
Most have good installation scripts that make its installation
rather painless.

- Clean out your CONFIG.SYS and AUTOEXEC.BAT of stray drivers
and TSR's that you don't need.  Your files should be 30,
and if you are running a disk cache (SMARTDRIVE) your buffers
can be put down to 10.  See if this gives you the ~540k needed
for SigmaPlot.

- If you are close to having the ~540k needed, you can run
SigmaPlot from under Windows and have Windows give SigmaPlot
a couple hundred kb of Expanded memory.  Read creating a PIF
file.

>  All in all, I am extremely frustrated, to the point that
>I am considering returning my computer (still in its 30
>day free return period) and buying a Mac.  I've never used
>one, but everyone who owns one seems to love them, something
>that I haven't found among IBM users.  Certainly it would be
>preferrable to get my 386sx to work, but I hate the idea of just
>patching things together and settling for a way of getting it
>to limp along.  Isn't there a way to make it *really*work*?
>Is Windows 3.0 inherently flawed?

If you bought a Windows app, lets say PowerPoint, then installation
and use is really easy.  Just point and click, pop in the diskettes,
and you can do things that the Mac can't do and for much less.
Windows 3.0 is great for running Windows apps.  It's only fair 
at running most DOS apps.  But its fundamental flaw is that
it runs on DOS.  OS/2 is definitely the way to go in the upcoming
years.  I wish MS will stop looking at this week's ledger and
plan for the future.  As for going to a Mac, then you're opening
another whole can of worms.  They have their own problems and
wierd stuff.  (as a former consultant, I *know*  :-) )
BTW, after working with Macintoshes, PS/2's and clones, Vaxen,
etc, I *personally* would love to own/work on/play with a
NeXT.  But I also love my trusty 386 clone, and would *not*
trade it for a Mac.
--
Bill Chin                            internet:bchin@umd5.umd.edu
MS-Windows Programmer                NeXTmail:bchin@is-next.umd.edu
PCIP, Computer Science Center        CompuServe:74130,2714
University of Maryland, College Park  *Standard Disclaimers Apply*

halpern@casbah.acns.nwu.edu (David Halpern) (04/15/91)

In article <8430@umd5.umd.edu> bchin@umd5.umd.edu (Bill Chin) writes:
>In article <31945@usc> logan@neuro.usc.edu (Christy Logan) writes:
>>[some parts of the description removed]
>>Sigmaplot seems to think there is 115K of memory available.  
>>Another program I run, CSS (a statistical package) runs all right
>>but thinks there is 365K of memory available.  Why the difference?  
>>But more importantly, why can't Sigmaplot find/use the memory?  
>>What good is 4M if my programs can't use it?  
>
>Well, you're caught with the difference between extended and 
>conventional memory.  First off, SigmaPlot 4.0 (that's the version
>I had to deal with) likes lots of conventional memory... the stuff
>below 640k.  The amount of memory available is shown with the "mem"
>command under DOS 4.0x.  The amount returned by CSS and SigmaPlot
>is the amount left after the respective programs have loaded, ie.
>this is data space left.  SigmaPlot likes to have ~540k of 
>conventional memory free before it loads.  A 100kb of expanded memory
>helps too. 
>
I would like to remind you that there is now version 4.1 of
sigmaplot.  I don't have the upgrade yet but Jandel  says that this
version uses around 490K of conventional memory (<50 K than 4.0)and uses both
expanded and extended memory, as much as your computer has instead of just
64k of expanded memory. It is also easier using this version with
windows 3.1 (the real windows version will come out in the fall, that's
what I've been told by Jandel).
There are some other new features which I think make the $50 upgrade 
worthwhile.

David Halpern
Telephone: (708) 491-4308
Office Location: TECH B426 (Center for multiphase flow)
Address:  Biomedical Engineering Department
          Northwestern University
	  Evanston IL 60208

e-mail:halpern@casbah.acns.nwu.edu

e4666881@rick.cs.ubc.ca (richard louie) (04/18/91)

     It sounds to me that you may have gotten a bad system - in regards to
not having the mouse working when you use Windows.  You should ask the dealer
about that one.
     As for the memory problems, they are very common on MS-DOS machines. 
To see why they are common, you need a bit of history.  MS-DOS was first
written with only 640KB of memory in mind - back then 640KB was a lot
on a PC. As 286s and 386s and later 486s come into being, the 640KB mark
became a problem - with the added complexity of programs and the user's wishes
to have TSR (programs which stay resident in memory even after you exit the 
program), that 640KB gets used up very quickly (most users only get about 
500KB of useable space if they are lucky).  To solve these memory problems,
companies like Quarterdeck have introduced produces like QEMM which will
put these TSRs and your DOS drivers into high memory (memory between 640 -
1024KB) and this would free up the valuable memory below the 640KB mark. As a
result, most 386 owners have about 550+ KB of free memory to play with.
     I've mentioned one solution - QEMM - but you can also do it on the cheap.
There are bound to be some extra DOS drivers and TSR which the dealer may
have installed which you don't use which could be gotten rid of to get that
extra memory back.  The first thing you should do is to boot the computer
off of a regular DOS disk with the system on it and nothing else. See if the
programs like Sigmaplot will run.  If they do, you just have to free up some
memory.  If they don't you got a computer which is not truely compatible. 
     There is one other solution, get a product called DR-DOS 5.0 from 
the people at Digital Research.  Its a DOS compatible OS which replaces MS-DOS.
WHY?  Because it will do everything MS-DOS will and give you the memory that
QEMM does and do it automatically.

Richard
e4666881@rick.cs.ubc.ca
 

sag@iplmail.orl.mmc.com (Steve Gabrilowitz) (04/18/91)

In article <1991Apr17.174244.5318@rick.cs.ubc.ca>, e4666881@rick.cs.ubc.ca (richard louie) writes:

|>      There is one other solution, get a product called DR-DOS 5.0 from 
|> the people at Digital Research.  Its a DOS compatible OS which replaces MS-DOS.
|> WHY?  Because it will do everything MS-DOS will and give you the memory that
|> QEMM does and do it automatically.


There is yet another possible solution to this RAM cram problem, which is to replace your COMMAND.COM with a shareware product called 4DOS.  It will keep most of itself swapped out to high memory (or even disk) and save you some of the memory normally eaten up by COMMAND.COM.

-- 




                            Steve Gabrilowitz
                            Martin Marietta, Orlando Fl.
                            sag@iplmail.orl.mmc.com
                            Fidonet 1:363/1701