Info-IBMPC@WSMR-SIMTEL20.ARMY.MIL ("Info-IBMPC Digest") (03/19/91)
Info-IBMPC Digest Fri, 15 Mar 91 Volume 91 : Issue 58
Today's Editor:
Gregory Hicks - Rota Spain <GHICKS@WSMR-Simtel20.Army.Mil>
Today's Topics:
DeSmet C Compiler
Game port access
Soundblaster (V91 #48)
Memory Hype? (V91 #48) (2 msgs)
MS WORD 5.5 on an Amstrad 2086 (V91 #50)
Bugs in MS FORTRAN V5.0 using 8087 Update
XTree Gold & McAfee's Vshield conflict (2 msgs)
Today's Queries:
Bugs in MS FORTRAN V5.0 using 8087
Connecting a second keyboard to a PC
Formatting 3.5 diskettes
DOS 4.01 FORMAT trys to recover BAD sectors
Looking for transistor for 8514
Send Replies or notes for publication to:
<INFO-IBMPC@WSMR-SIMTEL20.ARMY.MIL>
Send requests of an administrative nature (addition to, deletion from
the distribution list, et al) to:
<INFO-IBMPC-REQUEST@WSMR-SIMTEL20.ARMY.MIL>
Archives of past issues of the Info-IBMPC Digest are available by FTP
only from WSMR-SIMTEL20.ARMY.MIL in directory PD2:<ARCHIVES.IBMPC>.
----------------------------------------------------------------------
Date: Fri, 8 Mar 91 09:24:53 PST
From: sono!hayden@Sun.COM (Mike Hayden)
Subject: DeSmet C Compiler
In comp.sys.ibm.pc.digest you write:
>Is this beast still in existence? Does anyone know where it can be
>obtained, or possibly how one could get a manual for it?
>I don't so much want the compiler as I want to find out what many
>functions in some DeSmet C code that are called things like _lmove and
>_memset really do. (So that they can be done either directly in 8088 or
>through Turbo C)
He'll probably kill me for doing this, but I know someone who was
involved in the development of the Desmet stuff:
Mike Ouye
(415) 345-7400 (GO Corporation)
Sorry, I don't know a net address
Mike Hayden
sun!sono!hayden
------------------------------
Date: Fri, 8 Mar 91 18:12:45 BST
From: rambaut@castle.edinburgh.ac.uk
Subject: Game port access
Paul L. Kelly (pkel@brazil.psych.purdue.edu) writes:
> I need to access the game port on a PC compatible from Turbo Pascal.
>I know the address of the port (hex 200 to 207), but I can't seem to
>get anything other than readings from the joystick buttons. It would
>seem that if BASIC can handle a game port then Turbo Pascal would be
>able to do so as well. I have searched through the TP archives on
>wsmr-simtel20.army.mil, but nothing in the directory listing says
>anything directly regarding game ports or joy- sticks. Certainly
>somebody out there has written a game in pascal that uses a joystick!
>The application that I'm writing is not a game, and has to operate
>considerably faster than interpreted basic will allow. Any ideas?
Use the Bios routine (I can't remember which one so you will have to
look it up I think it is a subfunction of 13hex) It returns x and y for
both joysticks plus all four button states.
Hope this puts you on the right track,
Andrew.
------------------------------
Date: 8 Mar 91 18:10:01 GMT
From: mir@chorus.fr (Adam Mirowski)
Subject: Soundblaster (V91 #48)
> Date: Tue, 26 Feb 91 07:44:45 EST
> From: jay@zen.cac.stratus.com (Jay Appell)
> There are two additional chips that can be added for additional voices
> on the Soundblaster card. Does someone know if these chips can be
> purchased from a source other than the vendor ?
I suppose you speak about those "CM/S" chips.
I have discovered that they are in fact Philips/Compelec/RTC SAA 1099
chips. Philips send me two as a gift (because the chips already went
out of production) and they work now in my SoundBlaster.
SAA 1099 are expensive circuits, are no more distributed here in France
and there is probably no big deal in not buying them at Creative Labs.
Adam Mirowski, mir@chorus.fr (FRANCE), tel. +33 (1) 30-64-82-00 or 74
Chorus systemes, 6, av.Gustave Eiffel, 78182 Saint-Quentin-en-Yvelines CEDEX
------------------------------
Date: 8 Mar 91 18:37:46 GMT
From: mir@chorus.fr (Adam Mirowski)
Subject: Memory Hype? (V91 #48)
> Date: 26 Feb 91 16:37:00 CST
> From: "90SMWACC" <90smwacc@sacemnet.af.mil>
> I am considering the purchase of a 386DX system. The problem concerns
> *memory hype*. Many mail order firms advertise >>4Mb<< of ram. Does
> this mean 4 MEG of 8 bit memory or 4 MEG of 32 bit memory? Some of the
> motherboards I have seen contain memory sockets for 8 SIMS. Ads for
> SIMs are 1MEG of 8 bit memory. So how do I know if mail order computers
> are 4Mb or 4MEG of 32 bit memory? Also, if these computers have only
> 4Meg of 8 bit memory, aren't they misleading the public? That is 4Mb
> is really only 1Meg of 32 bit memory. Or are these systems arranged as
> 8 bit memory, requiring 4 memory fetchs to load a 32 bit instruction?
386DX memory must be either 16 bit or 32 bit as viewed from the chip.
There is a pin on the chip that is set by the bus to indicate that at
the current address there are only 2 bytes (16 bits) instead of 4
bytes. That occurs when the 386 accesses the ROM bios for example. In
that way the 386 can issue another bus cycle to access the upper 16
bits. Of course, it will do it only if the instruction dealt with 32
bit data. BTW, there are no A0-A1 address pins; instead you have 4
"byte enable" pins.
Each SIMM provides 8 bits, so you address 4 chips in parallel to get a
32 bits wide data access. 386DXs sold with 1 Mb of RAM MUST have four
SIMMs (if it uses them...). So you automatically get 256Kb SIMMs. I
would prefer not to have them, because they occupy precious sockets,
and you ever want to add more RAM sooner or later. There are no 512
SIMMs (because of multiplexed adressing reasons).
It is theoretically possible to use only 16 bits wide RAM, but this is
highly inefficient (no bus access pipelining is possible, by 386DX
design).
To sum up: 386DX normal RAM is always 32 bit wide and is NOT accessed
through the AT bus.
Adam Mirowski, mir@chorus.fr (FRANCE), tel. +33 (1) 30-64-82-00 or 74
Chorus systemes, 6, av.Gustave Eiffel, 78182 Saint-Quentin-en-Yvelines CEDEX
------------------------------
Date: 9 Mar 91 00:24:13 GMT
From: motcid!yeates@uunet.UU.NET (Tony J Yeates)
Subject: Memory Hype?
> Date: 26 Feb 91 16:37:00 CST
> From: "90SMWACC" <90smwacc@sacemnet.af.mil>
Info-IBMPC@WSMR-SIMTEL20.ARMY.mil writes:
>I am considering the purchase of a 386DX system. The problem concerns
>*memory hype*. Many mail order firms advertise >>4Mb<< of ram. Does
>this mean 4 MEG of 8 bit memory or 4 MEG of 32 bit memory?
That means 4 Mega bytes, where a byte is 8 bits.
[What out when buying chips tho', you *may* need nine 1 Mbit chips for
1Mbyte of RAM (= 1Mbyte of user RAM + 1Mbit for parity). I have only
seen mega-bit used in the context of chips tho'].
>Some of the motherboards I have seen contain memory sockets for 8
>SIMS. Ads for SIMs are 1MEG of 8 bit memory. So how do I know if mail
>order computers are 4Mb or 4MEG of 32 bit memory?
If you're concerned about that, call them and get them to put it in writing.
> Also, if these computers have only 4Meg of 8 bit memory, aren't they
> misleading the public?
I don't think so, 0.5-4Mbytes seems to be standard equipment for PCs.
I think they should always specify *bytes* tho'.
> That is 4Mb is really only 1Meg of 32 bit memory. Or are these
>systems arranged as 8 bit memory, requiring 4 memory fetchs to load a
>32 bit instruction?
A 386DX has a 32-bit wide data bus (a 386SX a 16-bit databus), so just
1 fetch.
>Maybe I am off base, but this stuff is a little confusing to me.
I don't think they do it on purpose (although they'll cover up their
weaker features). I would recommend creating a list of features that
you want want and then getting the companies that you're interested in
to send written quotes for their nearest offering. (Watch out for
obscure video cards, lack of expandability in terms of slots/RAM/extra
disk drives/etc., etc.). gateway 2000 have a well spec'd, affordable
range of PC that come with 4Mbytes of RAM, you should check them out.
------------------------------
Date: Sat, 9 Mar 91 14:24 EST
From: Michael Gordon <GORDONM@MAX.CC.DENISON.EDU>
Subject: MS WORD 5.5 on an Amstrad 2086 (V91 #50)
Andrew McLean <PHR050@ibm.southampton.ac.uk> wrote describing some
problems he was having with Microsoft Word and mouse drivers:
Regarding mouse drivers and Word 5.5: the readme.doc that comes with
Word is quite insistent that one use the mouse driver supplied with
Word. Its version is 7.05; try using it.
Michael D. Gordon
Denison University
gordonm@max.cc.denison.edu
------------------------------
Date: Fri, 8 Mar 91 14:05 EST
From: <LANDMAN%MSUPA.BITNET@CUNYVM.CUNY.EDU>
Subject: Bugs in MS FORTRAN V5.0 using 8087 Update
For those people inquiring about my results with fortran from
Microsoft, I am including a sample bit of test code from the SAVAGE
benchmark, that gives strange results. I compiled this under MS
Fortran v5.0 with as many different permutations of switches as I could
try. It reported the SAME results, independant of any of the switch
settings. Since I only use the 8087 library I cannot determine whether
or not this will occur in the other libraries.
I compiled this under MS Fortran v5.0. V4.1 gave me the correct
answer.
program savage
real*8 a
integer*2 i
a=1.d0
print *,'start timing'
do 10 i=1,2499
10 a=dtan(datan(dexp(dlog(dsqrt(a*a)))))+1.d0
print *,'the answer is ',a
end
My compilation of it on an 8088/8087 based PC/XT gave me the
following output. (from output redirection) Using MS Fortran v5.0.
start timing
the answer is 2500.277099609375000
The people at Microsoft did call me to confirm my find. It
appears that the 8087 control word was not properly initialized. I
suspect that the faulty initialization has to do with rounding, and the
treatment of roundoff errors.
They are sending me a patch disk, so I will report to the list
the outcome of those tests shortly. I hope Microsoft will allow me to
post the patch to the SIMTEL-20 archive, or even better, post it
themselves. I think that it would be a wonderful idea to have USENET
access to these things.
Joe Landman Graduate Physics Student
(Computational Physics/Condensed Matter Theory)
email: Joe_Landman@MTS.cc.WAYNE.edu
ZV50@MTS.cc.WAYNE.edu
07480JLN@MSU.edu
SnailNet: Wayne State University,
Physics and Astronomy Bldg,
666 Hancock Avenue, Detroit, MI 48202
ATTNet: (313)-577-2752
------------------------------
Date: Sat, 9 Mar 91 16:03:47 CST
From: Jhinuk Chowdhury <FF76%UNTVM1.BITNET@uga.cc.uga.edu>
Subject: XTree Gold & McAfee's Vshield conflict
DOS users (users of XTree & McAfee's Viruscan, in particular):
XTree Gold Version 2 and McAfee's Vshield (the RAM resident shield for
viruses - part of his famous Viruscan package) seem to have one
conflict.
If you load Vshield in RAM in its entirety, there's no problem.
Vshield occupies about 37 K of RAM. However, if you load VShield and
use the swap option (/SWAP F:\ suppose, to swap to a RAM disk), the
amount loaded in conventional RAM is minimized (assuming your RAM disk
uses extended memory). In such a case, VShield occupies about 3 K.
The problem is, in this latter case, XTreePro Gold V4.3 or the latest
version XTree Gold 2.0 will lock up your computer (requiring a cold
boot) the momemt you strike the F1 key for it's help screens.
Has anyone else run into this irritating situation? Considering the
popu- larity of both the packages, I would imagine that it would
attract the attention of many people. If you know of any tricks or
tips that may solve this, could you please post or let me know by
e-mail.
Thanks, in advance. _ Jhinuk.
Jhinuk Chowdhury
Internet : FF76@VAXB.ACS.UNT.EDU
Bitnet : FF76@UNTVM1
------------------------------
Date: Sat, 9 Mar 91 17:02:00 CST
From: "Ronald Hahm" <HAHM5247@ISCSVAX.UNI.EDU>
Subject: XTree Gold & McAfee's Vshield conflict
Date: Sat, 9 Mar 91 16:03:47 CST
From: Jhinuk Chowdhury <FF76%UNTVM1.BITNET@uga.cc.uga.EDU>
> ........ However, if you load VShield and use the swap
> option (/SWAP F:\ suppose, to swap to a RAM disk), the amount loaded in
> conventional RAM is minimized (assuming your RAM disk uses extended memory).
> In such a case, VShield occupies about 3 K. The problem is, in this latter
> case, XTreePro Gold V4.3 or the latest version XTree Gold 2.0 will lock up
> your computer (requiring a cold boot) the momemt you strike the F1 key for
> it's help screens.
I don't use XTree Gold, but I am tempted to ask whether you have tried
to leave out the RAM disk swap but just leave the /swap option. I use
the /swap option but not the RAM disk option for Vshield.
| Ronald Hahm Internet: HAHM5247@ISCSVAX.UNI.EDU |
| Graduate Assistant |
| College of Business Administration |
| University of Northern Iowa BBS InterLink: RONALD HAHM |
| Cedar Falls, Iowa 50613 USWest: 319-273-3822 |
------------------------------
Date: Fri, 8 Mar 91 14:56:34 +0100
From: C102TA%AINUNI01@pucc.PRINCETON.EDU
Subject: Bugs in MS FORTRAN V5.0 using 8087
Like Joe Landman, I also found a strange behaviour of MS FORTRAN 5.0
working under 8086/8087 machines.
I have several large FORTRAN programs that excessively use the math
coprocessor. I have access to several PS/2 model 30's with 8086/8087
processors. On some of these computers my FORTRAN programs crash before
they have executed any line of code. When I call CodeView or any other
debugger, the programs work correct- ly. Needless to say that the
programs also run on any 286/386 machine without any problems.
The strange thing is that this behaviour is not completely
reproducable!!! Sometimes the same program runs, sometimes it doesn't.
On some PS/2 model 30's the problem has never been observed. So I
suppose that the whole thing could be a timing problem between 8086 and
8087 that only arises together with certain instruction sequences.
Maybe the 8087 startup code of MS FORTRAN contains such a critical
sequence.
The problems described above are amazingly similar to those of Joe
Landman (8086/8087 combination, behaviour under debuggers) though the
effects are different (wrong results versus program crash).
Has anybody made similar observations?
Does anybody know what to do in this case?
Andreas Tranquillini
University of Innsbruck, Austria
C102TA@AINUNI01.BITNET
------------------------------
Date: Sat, 09 Mar 91 11:44:41 +0200
From: Omer Zak <XLACHA1%WEIZMANN.BITNET@CUNYVM.CUNY.EDU>
Subject: Connecting a second keyboard to a PC
I need to connect a second keyboard to a PC and run a program which
accepts input from both keyboards, yet recognize the origin of each
incoming character. One possible way to implement this is to locate a
keyboard with RS232C interface and connect it to one of the COMn: ports
of the PC.
However, I can't locate such a keyboard.
I would appreciate if any one of you can lead me to such a keyboard
and/or suggest alternative solutions which can be implemented. Note:
due to cost and space considrations, use of a terminal with display and
with RS232C interface - is not desirable in this application.
--- Omer (BITNET: XLACHA1@WEIZMANN)
------------------------------
Date: Fri, 8 Mar 91 13:48:36 EST
From: "Dennis L. Williams" <DLWILLIA%OCC.BITNET@uga.cc.uga.edu>
Subject: Formatting 3.5 diskettes
I have a HeadStart III AT-286 clone at home, and a IBM PS/2 55sx here
at work. For both machines, the "FORMAT A:" command formats both DD and
HD diskettes at 1.4m. To format a DD at 720k, I need to add "/N:9
/T:80".
Other clones, with less tolerant drives, reject DD diskettes formatted
at 1.4m. Now for the controversy and question: What are the risks of
formatting DD diskettes at 1.4m, or visa-versa?
On one hand, I have read that DD and HD disk platters are physically
different, as is the drive recording techniques. A diskette formatted
incorrectly WILL result in eventual data loss.
On the other hand, at least 3 vendors selling "hole punches" to convert
DD to HD diskettes claim that there is no difference, except the extra
hole in the case.
Any comments and references will be appreciated. Thanks in advance.
DLWILLIA @ OCC . BITNET : Oakland Community College
Dennis L. Williams : 2480 Opdyke Road
Manager of Technical Support : Bloomfield Hills, MI 48013
(313) 540-1583
------------------------------
Date: Fri, 8 Mar 91 12:44 EDT
From: "BRUCE S. WINTERS" <BSW%UTRC@utrcgw.utc.com>
Subject: DOS 4.01 FORMAT trys to recover BAD sectors
I have found that when I format a hard disk with MS-DOS 4.0 it tries to
recover blocks that were marked bad during the low level format. Is
there a way to disable this function with a command line switch? A
search through the manual was no help.
Thanks
Bruce Winters
EMAIL: BSW%UTRC@UTRCGW.UTC.COM
Tel: 203-727-7089
United Technologies Research Center
E. Hartford, Ct. USA
------------------------------
Date: Fri, 8 Mar 91 13:01 EDT
From: "BRUCE S. WINTERS" <BSW%UTRC@utrcgw.utc.com>
Subject: Looking for transistor for 8514
I have been working on an IBM 8514 VGA display and found a bad
switching transistor in the power supply Q801. It is made by
Matsushita it has C3527 written on it but Matsushita can't cross the
number over because it is an OEM part number. Does anyone have
Schematics for the 8514? A data sheet for the transistor or an ECG
replacement?
Bruce Winters
UTRC
EMAIL:BSW%UTRC@UTRCGW.UTC.COM
TEL:203-727-7089
------------------------------
End of Info-IBMPC Digest V91 #58
********************************
-------