Info-IBMPC@C.ISI.EDU (Info-IBMPC Digest) (06/20/87)
Info-IBMPC Digest Friday, 19 June 1987 Volume 6 : Issue 47
This Week's Editor: Billy Brackenridge
Today's Topics:
Status of Info-IBMPC
toadlong.shr Updated
PS/2 comes with Disk Cache Software that Runs on AT
Microsoft Bus Mouse Works on Zenith 248
Double Width/Height print utility
Speaker Disabling
Replacement malloc for Microsoft C 4.0
MARK/RELEASE
H2INC.C
After the Funeral
MOUSE.E Mouse Support for Epsilon
PHONE.C Electronic Address Book Program
STICKY.E Updated
WSSINDEX Disk Cataloger
Turbo C Rumors
80386 Mailing List vol1 #2
The 386users list has moved
20MHz 386-based PC
varieties of 386 AT upgrades
80386 UN*X -- MicroPort, et Al
PC Designs GV386
80386 Smalltalk
Dhrystones on OS/2
Mice and COM Ports
DOS 3.3 ECHO OFF Patches
Zenith Z-248 Com3 Access
Today's Queries:
Serial Port Driver for 4 Ports
Toshiba 1100+ Async Port Strangeness
Lattice C forkl
OS/2 and 80386
SCCS under DOS?
AT Clones
Fortran Compiler Recommendations (2 Msgs)
INFO-IBMPC BBS Phone Numbers: (213)827-2635 (213)827-2515
----------------------------------------------------------------------
Date: 19 Jun 1987 14:00:10 PDT
From: Billy <BRACKENRIDGE@C.ISI.EDU>
Subject: Status of Info-IBMPC
Things aren't looking hopeful for the continued existence of
info-ibmpc. Frank da Cruz of Columbia University and Kermit fame has
volunteered to take over the digest. He had done this several weeks
before I made my resignation announcement on the first of June.
The management at Columbia thinks this is a wonderful idea if the
responsibility of running info-ibmpc comes with $50K a year to pay
for one full time staff member.
This is a terrific price! For this we get a staff with a good track
record, excellent bitnet, usenet, and arpanet connectivity, lots of
computer cycles and disk space, and plenty of student volunteer
labor.
For the last month I have been attempting to find money for Columbia.
Many readers have volunteered to talk to the "suits" in their
companies to ask them to help support info-ibmpc by giving a grant to
Columbia. This has been terribly frustrating. The managers have
universally thought this would be a great idea and would benefit their
companies and the PC industry as a whole, but to a (wo)man they all
agree that they personally can't make such a decision. I wonder how
decisions do get made at these companies? In some companies the jury
is still out, but I take their silence as a negative sign.
It is tasteless to ask for money like this, but I figure I only have
this forum for another two weeks so I might as well try. If anyone
has any leads on possible funding please let me know. I have never
published my phone number, but I make an exception for good news and
money. (213)822-1511
In the mean time people are rushing in their last minute submissions.
I am particularly pleased that program submissions are still
arriving. The library will remain here on C.ISI.EDU until they cart
the PDP-10 out the door. (predicted to be sometime this fall)
There are several messages advertising other network news groups and
sources of PC information. You have ten more days so keep those cards
and letters coming. There will be life after info-ibmpc.
------------------------------
Date: Thu, 18 Jun 87 01:52:04 EDT
From: David Kirschbaum <kirsch@braggvax.arpa>
Subject: toadlong.shr Updated
Gentlemen,
Please totally remove the TOADLONG package.
Here's a better version I am a little more satisfied with:
TOADLONG v1.1, 17 June 1987
(Totally replaces v1.0, which is buggy and stupid.)
Author: David P Kirschbaum, Toad Hall
TOADLONG.INC (and its accompanying test program TOADTEST.PAS) were
written to provide MS-DOS and PC-DOS Turbo Pascal users with a package
of long integer utilities. They are released to the public domain
for noncommercial application (e.g., no sales).
The long integer functions presently supported are relatively primitive
and limited: long adds, long subtracts, adding and subtracting longs
with integers, NOTs, Shifts, Rotate with Carry, etc., plus a BUNCH of
unsigned integer math from Jos Wennmakker of Holland (the Mad Dutchman
and author of REFORMAT.PAS). Hopefully this package will provide a
base to which others can contribute. At least the price is right!
Also included in TOADLONG.INC are several string routines that fiddle
with long integers in various convenient ways.
In a separate file, TOADTIME.INC, you will find a number of system
date/time routines to provide clean, fast interfacing with DOS and
its peculiar date/time formats. A nice elapsed timer function is
used throughout TOADTEST
Most of this is in inline assembler (and 10,000 thanks to the author
of INLINE!).
I've gained surprising advantages by using long ints in my applications,
as opposed to reals and other awkwardness. The speed test of longs vs.
reals in TOADTEST.PAS should convince you of that! And I won't even
talk about accuracy .. I can't test much of it except by logic and
the primitive tests in TOADTEST.
And you won't BELIEVE how much sounder my knowledge of assembler is
after this hack! You'll never appreciate a microprocessor and its
instructions until you try to emulate them in code.
Enjoy the code, let the credits remain where they are.
V1.1, 17 Jun 87:
Vastly improved inline assembler in TOADLONG.INC.
(Was doing v1.0 for relaxation and musta been REALLY tired
to write such Freshman assembler (no offense to freshmen).
Fixed a logic error in Long_RcR and Long_RcL,
and a math error (stupid damned signed integers) in real_long.
Added a bunch more long integer functions in TOADLONG.INC.
Added some more flashy tests, nicer interface, in TOADTEST.PAS.
Split out timer stuff to TOADTIME.INC.
David Kirschbaum
Toad Hall
7573 Jennings Lane
Fayetteville NC 28303
kirsch@braggvax.ARPA
(voice: 919 868-3471,
data: same number, via Kermit, xmodem, ymodem, wxmodem,
and the world famous CommCrypt secure communications CRYPCOMM)
[TOADLONG.SHR has been updated in the info-ibmpc library. -wab]
------------------------------
Date: 18 Jun 87 10:56 EST
From: 26-324%HARVBUS1.BITNET@wiscvm.wisc.edu
Subject: PS/2 comes with Disk Cache Software that Runs on AT
I've been experimenting with the new PS/2s and have found something that
may be useful to us ordinary AT users. The model 50 and above come with
a feature that allows memory to be used as disk cache. However, the
feature is very well hidden, in more than one sense. There is no mention
of it in the Model 80 quick reference (I don't recall seeing it in the
50/60 book either), there is no mention of caching in the nifty menus
that come up when you boot the Reference Diskette, and the two relevant
files on the Reference Diskette, IBMCACHE.COM and IBMCACHE.SYS, are
hidden files. To install the cache, one boots DOS, defaults to the A
drive where the reference diskette lives, and types IBMCACHE. A menu
comes up offering to configure the cache. Its main purpose appears to be
putting a (visible) copy of IBMCACHE.SYS on the boot drive and modifying
CONFIG.SYS to start it up. The good news is that IBMCACHE.SYS appears to
be usable on ATs as well. Just copy the file off your favorite 50/60/80
hard drive, put it on your AT, make the appropriate modifications to
CONFIG.SYS, and watch your hard disk fly. Check the PS/2's CONFIG.SYS
for the correct syntax, or drop me a line if you manage to get a copy of
IBMCACHE.SYS without an accompanying CONFIG file.
I've used IBMCACHE on my AT for two days now, running Desqview, Fastback
and other fairly demanding programs, with no sign of problems. There is
a small improvement in speed the first time a file is read, but the real
advantage comes with repeated reads of the same file, which can be up to
80% faster. It is possible that problems will crop up after some use, so
be very careful if you try using IBMCACHE on an AT.
By marketing IBMCACHE with the PS/2 reference disk rather than with DOS
3.3, it appears that Big Blue is trying to give the PS/2 one extra, if
spurious, performance shot over the AT. Maybe IBM will make caching
officially available to AT users in the future, but there's no reason
not to try it now. One note: I have tested caching only with PCDOS 3.3.
It probably should not be run with earlier versions.
Rich Stillman
ARPA/EDUnet: 26-324%HARVBUS1.BITNET
BITNET: 26-324 at HARVBUS1
(617)495-6135
------------------------------
Date: 18 Jun 87 17:57 GMT
From: esi @ Walker-EMH.arpa
Subject: Microsoft Bus Mouse Works on Zenith 248
We have bought four Microsoft Buss Mice for our four Z-248s
They work great (maybe a little sensitive)
and don't take up any com ports.
Mark Meaders
------------------------------
Date: Thu, 18 Jun 87 11:13:42 pdt
From: reynolds@ames-prandtl.ARPA (Don Reynolds)
Subject: Double Width/Height print utility
One font control package that does this is NicePrint, available at about the
$100 range from:
Spies Laboratories
P. O. Box 336
Lawndale CA 90260
(213)538-8166
Best,
Don
------------------------------
Date: Thu, 18 Jun 87 11:06:36 pdt
From: reynolds@ames-prandtl.ARPA (Don Reynolds)
Subject: Speaker Disabling
Version 2.0 of SILENCE was written in 1985, and is in the public domain.
However, I found it did not eliminate speaker clicks from some software.
Further, I had some "RAM-Cram" problems with it, with the public domain
SPEECH.COM program, with SideKick and other TSR's. The author of SILENCE
Version 2.0 at that time had added some whistles to change clock speed, etc.
SILENCE is on several BBS systems (Maybe in PC-BLUE on SIMTEL-20), but the
author of SILENCE gave the following address:
Carl Burtner
435 Brandon Road
Rochester, NY 14622
I opted for the solution Billy recommended--a switch in the speaker line.
Actually, I added a DC blocking capacitor and an earphone jack for controlled
volume. Otherwise, 5 volts is a bit strong for earphones and audio inputs.
Best,
Don
------------------------------
From: kent@ncoast.UUCP (Kent Williams)
Subject: Replacement malloc for Microsoft C 4.0
Date: 16 Jun 87 20:04:49 GMT
Organization: Cleveland Public Access UN*X, Cleveland, OH
Here is a replacement for the standard malloc, realloc, calloc, free,
sbrk, and brk for Microsoft C 4.0.
It was written to satisfy the need for a 'debugging malloc' in the MSC4.0
environment on the PC.
It can be used in production code with impunity, though it isn't as
efficient space-wise as the standard set of routines.
The core routines have been tested in all memory models - but as noted,
there will need to be some work on the debug code to make it function
properly in other than small code, small data.
[MALLOC.C has been added to the info-ibmpc lending library. -wab]
------------------------------
Date: Thu, 18 Jun 87 11:33:23 pdt
From: reynolds@ames-prandtl.ARPA (Don Reynolds)
Subject: MARK/RELEASE
Turbopower Software has added FMARK, RAMFREE, EATMEM, and MAPMEM programs to
Version 2.1 (dated 07/18/86) of MARK and RELEASE. MAPMEM is particularly
useful to display what is loaded. MARK and RELEASE now take parameters to
unload after a desired point.
Executables and source are in the public domain. MAPMEM and RELEASE are in
Turbo Pascal Version 3 and the other programs are in CHASM. Source is
available in TSRCRC.ARC on the Compuserve Borland SIG (Go BOR-100) in data
library 1 (DL1).
[Will the Info-IBMPC Library be maintained? Will SIMTEL-20 step in?]
For a $5.00 shipping fee, Turbopower will send you the latest diskette with
the utilities. If you use Turbo Pascal, you might inquire about their
Turbopower Utilities. They include some useful tools for serious developers.
Turbopower Software
3109 Scotts Valley Drive #122
Scotts Valley, CA 95066
(408)438-8608
Compuserve: 72457,2131
Best,
Don
[Yes the library will be here until Tops-20 is declared gov't surplus
and our PDP-10 is shipped off to the great warehouse in the sky. This
is expected sometime in the fall. The BBS will also remain in operation.
Perhaps giving up info-ibmpc I will have time to upgrade the BBS and
at least get all the source files up to date. If I really get my act
together I might get the BBS on the net by fall in time to provide
FTP access when the PDP-10 goes away. -wab]
------------------------------
Date: Wed, 17 Jun 87 16:59:28 pdt
From: Robert Lenoil <lenoil%apple.csnet@RELAY.CS.NET>
Subject: H2INC.C
The enclosed file, H2INC.C, does a primitive job of translating C language
header files into assembly language include files. It is a valuable tool for
systems that include both C and assembly components that use common constant
declarations. With H2INC, only the C header file needs be maintained; the
assembly include file can be generated automatically via a make file. This
removes the chance of having inconsistent constant declarations between your
C and assembly modules.
Syntax: H2INC [infile[.h]] [outfile[.inc]] [-Didentifier[=[string]]]...
H2INC recognizes the #ifdef and #ifndef preprocessor directives and acts
accordingly, based upon any identifiers defined on the command line or
previously in the header file. It understands decimal, hexadecimal, and
octal constant declarations; all others are converted to string literal
equates. One exception is expressions involving already defined constants,
which will be translated to expressions, not string literals. H2INC
unfortunately does not translate structure declarations (any volunteers?).
If the output file is omitted, stdout is used. If the input file is also
omitted, stdin is used for input. H2INC was written in Microsoft C version
4.0, and contains several workarounds for bugs found in that compiler version.
Example input: Example output:
#define BAR 1 BAR equ 1
#define BLECH 16 + BAR BLECH equ 16 + BAR
#define PNAME "H2INC" PNAME equ "H2INC"
------------------------------
To: info-ibmpc@c.isi.edu
Subject: After the Funeral
Date: Thu, 18 Jun 87 18:27:10 EDT
From: Joe Morris (jcmorris@mitre.arpa) <jcmorris@mitre.arpa>
Assuming that INFO-IBMPC fades into the sunset after 1 July, there will be
many people who still need a place to post general-distribution questions
and comments. While it won't help everyone, I would recommend that readers
of the digest who work for organizations with large IBM mainframes investigate
the PCSHARE bulletin board. It is NOT a public forum; access is restricted to
companies which are members of SHARE, the large IBM mainframe users' group.
Contact your local computer center staff to find out if your organization
has access to PCSHARE.
What other forums (fora?) are there to ease the withdrawal symptoms after
our regular INFO-IBMPC fix is cut off?
Joe (and a tip o' the hat to [wab] and his crew) Morris (jcmorris@mitre.arpa)
------------------------------
Date: Wed, 17 Jun 87 15:26:13 pdt
From: Robert Lenoil <lenoil%apple.csnet@RELAY.CS.NET>
Subject: MOUSE.E Mouse Support for Epsilon
The enclosed file adds limited support to Epsilon for the Microsoft Mouse, or
any other mouse that adheres to the Microsoft Mouse programming interface.
Some of the functions provided are: setting point and mark, deleting a region,
yanking text, scrolling several lines, and a "scroll bar" (a la Macintosh).
Enjoy!
[MOUSE.E has been added to the info-ibmpc EEL lending library. The directory
is <INFO-IBMPC.EEL> -wab]
------------------------------
Date: Wed, 17 Jun 87 15:48:12 pdt
From: Robert Lenoil <lenoil%apple.csnet@RELAY.CS.NET>
Subject: PHONE.C Electronic Address Book Program
The enclosed file, PHONE.C, implements an electronic address book. The syntax
is: PHONE search_string [phone_book]
The search string is a string to search for. Phone_book is a filename of a
text file containing the entries to search through. If phone_book is
omitted, the environment variable PHONEBOOK is used. The only rule for the
data file is that a blank line separates entries. PHONE searches the phone
book and prints any entries that contain search_string.
PHONE was written in Microsoft C for MS-DOS. It will also compile unchanged
on Berkeley Unix systems. (USENET's mod.sources wouldn't post the program,
possibly because its name exceeded two letters and it doesn't require any
data files like /etc/foobar.) For those compilers which already have the
strstr() library function (such as Microsoft C), define the preprocessor
constant MSC; otherwise a program-supplied strstr() will be compiled in.
[PHONE.C has been added to the info-ibmpc lending library. -wab]
------------------------------
Date: Wed, 17 Jun 87 14:48:55 pdt
From: Robert Lenoil <lenoil%apple.csnet@RELAY.CS.NET>
Subject: STICKY.E Updated
I uncovered one more bug in my previous STICKY.E; it seems some of my
recent changes invalidated some previous modifications, so that
completion on entries with ".." in the path was performed relative to
the DOS current directory, not the buffer's directory. Enclosed is
what I hope to be the final version of STICKY.E.
[STICKY.E has been updated in the <INFO-IBMPC.EEL> lending library. -wab]
------------------------------
Date: 1987 Jun 18 21:40 EDT
From: (Bob Babcock) PEPRBV@CFAAMP.BITNET
Subject: WSSINDEX Disk Cataloger
In response to a previous message in Info-IBMPC Digest asking about
disk cataloging programs, and a reply recommending my shareware
program WSSINDEX:
1. The latest version is 3.12; there was a bug introduced in version
3.10 which causes occasional crashes when indexing disks with
subdirectories.
2. It doesn't do disk labels yet, but version 3.2 will.
For more info my E-mail address is
Bitnet: PEPRBV@CFAAMP
or PEPRBV%CFAAMP.BITNET@wiscvm.wisc.edu
------------------------------
Date: Thu, 18 Jun 87 22:31:52 EDT
From: Russell Nelson <bh01%CLUTX.BITNET@wiscvm.wisc.edu>
Subject: Turbo C Rumors
This is totally unsubstantiated rumor, and hence, may be true:
Notice that Wizard C is no longer being advertised?
Notice that Wizard systems moved to California early this year?
Notice that Turbo C is an awfully mature compiler (barring some bugs)?
Notice that Wizard C and Turbo C both have some really funky functions in
their libraries?
I leave you to draw your own conclusions.
-russ
------------------------------
From: rochester!steinmetz!davidsen@steinmetz.arpa (William E. Davidsen Jr)
Subject: 80386 Mailing List vol1 #2
Date: 19 Jun 87 13:19:21 GMT
Organization: GE Corp. R & D, Schenectady,NY
80386 mailing list vol 1 #2
This m/l is for exchange of questions, answers, information and rumors
regarding the Intel 80386 family of chips and the computers using them.
Copies of the mailings will be posted to appropriate newsgroups as time
permits.
The addresses for the list are now:
386users@udel.edu - for contributions to the list
or {ihnp4 | allegra}!seismo!udel.edu!386users
or ...!berkeley!udel.edu!386users
or ...!harvard!udel.edu!386users
386users-request@udel.edu - for administrivia
or {ihnp4 | allegra}!seismo!udel.edu!386users-request
or ...!berkeley!udel.edu!386users-request
or ...!harvard!udel.edu!386users-request
bill davidsen (wedu@ge-crd.arpa)
{chinet | philabs | sesimo}!steinmetz!crdos1!davidsen
"Stupidity, like virtue, is its own reward" -me
[The next few messages are from the 386 mailing list. -wab]
------------------------------
To: 386users@UDEL.EDU
Subject: The 386users list has moved
Date: Mon, 08 Jun 87 14:55:13 -0400
From: BBoards Support (agent: James M Galvin)
I will now be maintaining the mailing list for Bill Davidsen. He will
remain in control of the digest and its contents. I will be redistributing
the digests for him from our site.
This message serves three purposes. The first is to test the addresses,
as they say. The second is to appeal to those of you who have Internet
addresses or know a path to yourself from the Internet to please ask to
have your addresses changed. You should be able to read the "received"
lines of the message to find out how the message got to you.
I did my best to "convert" the addresses Bill gave to more suitable
addresses, but I certainly don't know everything. So, I would appreciate
if you would take a moment of your time to see if you can give me a
better address.
The final purpose is to announce the movement of the list and the new
addresses associated with it. Most subscribers are all UUCP subscribers,
so I will attempt to give a reasonable path from UUCP to me for that group.
The addresses for the list are now:
386users@udel.edu - for contributions to the list
or {ihnp4 | allegra}!seismo!udel.edu!386users
or ...!berkeley!udel.edu!386users
or ...!harvard!udel.edu!386users
386users-request@udel.edu - for administrivia
or {ihnp4 | allegra}!seismo!udel.edu!386users-request
or ...!berkeley!udel.edu!386users-request
or ...!harvard!udel.edu!386users-request
I will also be maintaining archives, which will be available via anonymous
ftp for those of you with that access. That information will appear at
a later date.
So, welcome to the list!
Jim <386users-request@udel.edu>
BBoards Support
------------------------------
Date: Fri, 22 May 87 21:50:57 PDT
From: ihnp4!lll-crg!csustan!smdev (Scott Hazen Mueller)
To: chinet!steinmetz!crdos1!386users
I wrote:
>In conversation with Microport yesterday, they told us that System V for 386
>systems is available in pre release form. The official release is scheduled
>for July 1, pending AT&T approval. [...etc...]
We're running it now on our Compaq. It still has bugs, but it mostly runs.
I'm pretty impressed with the overall quality, considering that it is
pre release. You don't get DOS-Merge with the pre release version, though,
so I can't comment on:
>Microport claims that the performance degradation of SysV+DM is only 5% over
>plain DOS!
...yet.
\scott
------------------------------
From: tslu@oliveb.UUCP (Shang Lu)
Subject: 20MHz 386-based PC
Date: 27 May 87 00:11:19 GMT
Organization: Olivetti ATC; Cupertino, Ca
[ Reposted from USENET. Please post info to this ML if you answer.
he has been added to the list - bd ]
I am doing a survey on 20MHz 386-based PCs. I know Compaq, TI,
and IBM are scheduled to deliver these machines.
Does anyone know if any 20MHz machines is already available ? If
yes, how is the performance ? Any of them use cache memory ?
------------------------------
To: umix!itivax!chinet!crdos1!386users
Subject: varieties of 386 AT upgrades
Date: Wed, 27 May 87 12:51:27 -0400
From: Steve Dyer <itivax.uucp!umix!rutgers!spdcc.com!dyer>
Basically, I'd like to poll the folks out there on how they feel the
best method is to upgrade a 286 clone to the 386. My options seem to be:
1.) Get an Intel 386 motherboard. These apparently aren't sold to
end users by Intel (perhaps to defer the wrath of their customers).
I see a fair number of 386 boxes appearing on the market which are
based on this; it is possible to get the motherboard alone from some
distributor? What price can I expect to pay?
2.) Get the Intel 386AT INcard. I'd like to know what constraints, if
any, are placed on the clone as compared with a stock AT. For example,
286/287 physical placement on the clone motherboard. Also, I have
heard that you have to remove all but 256K of memory on the motherboard,
an operation I'm not absolutely sure I can perform on my machine.
War stories, successful or not, here would be quite useful to many of us.
3.) Replace the machine with a Compaq 386. Does SCO XENIX V for the 286 run
"out of the box" on the Compaq or is a special version necessary?
I look forward to being able to give some answers, instead of questions, in the
future!
Steve Dyer
dyer@harvard.harvard.edu
dyer@spdcc.COM aka {ihnp4,harvard,linus,ima,bbn,halleys}!spdcc!dyer
------------------------------
From: jordan@titn.TITN (Jordan Bortz)
Subject: 80386 UN*X -- MicroPort, et Al
Date: 29 May 87 15:46:49 GMT
Organization: TITN Inc. Hayward, CA
Hello - I'm looking for people who have experience with different
flavors of UN*X for the 80386; microports, AT&T's, etc.
Also wondering what kind of end user products, like ALIS, CAD/CAM, etc.
people have ported, or are porting to the 80386.
Also wondering if anyone has put VM in their 386 kernel, and whether
the 386 has adequate VM performance.
Send EMAIL and I will summarize to the net!
Jordan
=============================================================================
Jordan Bortz Higher Level Software 1085 Warfield Ave Piedmont, CA 94611
(415) 258-8948 UUCP: (decvax|ucbvax|ihnp4)!decwrl!sun!dlb!plx!titn!jordan
=============================================================================
------------------------------
From: davidsen@crdos1.uucp
Subject: PC Designs GV386
Last year I bought a PC Designs GV386 at the end of the year. I
bought it sight unseen based on the following factors:
1) a friend had bought 20+ of their 80286 machines and felt
very good about their compatibility, reliability, and warrantee.
2) if it wasn't compatible I could send it back for a refund in
the first 30 days.
3) another friend benchmarked it at Comdex and said it was
about 20% faster than the Compaq.
4) they were one of two companies willing to deliver by the end
of the year (tax reasons).
Initial problems
================
When the machine arrived, it was obvious that it had been
'drop shipped' (from several feet high). The mounts holding the
motherboard in place were torn loose, and the hard disk mounting
rail screws were sheared on one side. I reported the problem,
and while waiting for UPS to come and return it home for
repairs, I set the motherboard in place, connected all of the
loose cables, and powered it up.
It worked flawlessly, a tribute to something, for sure. The
only real problem was that the hard disk screamed while running
and stopped with a noise best described as ticking followed by a
scrape to halt. PCD was worried that the motherboard might have
cracked, or I could have gotten a replacement hard disk if I
wanted to put it in myself.
The new machine was returned in about seven days and has been
working flawlessly ever since. I ran an AT compatibility suite
I developed to evaluate clones at work on the GV386, and it
passed every test. I could find no problems with it (it doesn't
run BASICA, obviously).
Since then I have booted Xenix and Microport on the machine,
and have had no trouble with either, although no o/s seems to do
anything useful with F11 and F12.
The machine benchmarks about twice as fast as my 9MHz AT, and
for integer, character, and long arithmetic, and logic (compare
and branch type stuff) it's about 2.5x a VAX 11/780. The bit
fiddling (set, test, clear bit) is 13 times faster than a VAX
(!) using a UNIX benchmark suite I have run on about 60
machines.
What's it like?
===============
The machine is in an AT sized box, but offers three half
height slots at the right. I intend to put in a tape drive at
some point. The control panel at the left has a key switch,
power, disk, and turbo lights, and a reset switch. The "little
red switch" has avoided using the "big red switch" when things
hang. The system runs 8 or 16MHz, with or without cache, all
selectable from the keyboard in DOS. UNIX takes over the
keyboard and doesn't let the sequences go thru.
The Macroswitch keyboard has switches to run on AT or XT, and
to switch the meaning of the caplock and control keys (including
a 2nd set of keycaps for those keys). The touch is a bit soft,
but I have no real objections to it. I would really like a true
AT touch, but maybe that's covered by 'look and feel'.
The motherboard holds 1MB of 100ns RAM, which can be replaced
by 4MB using 1mbit chips. There is no 32bit bus. I have an
additional 2MB of cheap 120ns RAM on an AT card (16 bit bus),
and the cache make a vast difference. With cache on the AT
memory is only 20% slower than the 32bit memory, without cache
it's about 35% slower. The 32bit memory runs about 20% faster
with cache.
My disk is an 80MB Seagate, other sizes are available. The
BIOS support about 40 disk sizes, and there is some code to let
you roll your own disk table (which I haven't tried). The 1.2MB
disk is a quiet reliable Japanese model (I believe Toshiba). I
added a Toshiba 360k drive and saved about $50 by getting it in
black.
The BIOS contains the setup routines and some rudimentary
diagnostics. You can also bypass the memory check by pressing
ESC during the test. I usually let the test run when I power the
machine up each time, just in case. It runs about 10 sec/MB,
easy enough to tolerate.
Other goodies
=============
The system comes with a good owner's manual. Not perfect, but
obviously written in English rather than translated. The
technical support is fast, available, and capable.
The machine comes with DesQview 1.3 and an EMS simulator which
uses the 80386 'real 8086' mode to do memory mapping. I have had
no trouble with this, although I don't use it a lot. I have been
using it to allow me to do other things while downloading files
with PROCOMM. The secret of making it run well is to cut the
time slice size WAY down (I use 1). This is not a problem with
the 80386, since it runs fast enough to handle the overhead. The
default is 3 ticks, and this is simply too jerky for practical
use.
Conclusion
==========
I have had no reason to regret buying this machine. If I was
rich I'd buy another to run a BBS. It should run even faster
with the 32bit operating systems, but is a find UNIX and casual
DOS machine. Base price: $4000 the last time I looked (with 40MB
hard disk).
bill davidsen (wedu@ge-crd.ARPA)
{ihnp4!chinet | philabs | seismo!rochester}!steinmetz!crdos1!davidsen
"Stupidity, like virtue, is its own reward" -me
------------------------------
From: jordan@titn.TITN (Jordan Bortz)
Subject: 80386 Smalltalk
Date: 1 Jun 87 17:06:08 GMT
Organization: TITN Inc. Hayward, CA
Does anyone know of, or is writing, a '386 Smalltalk, especially
under X/UNIX??
Jordan
=============================================================================
Jordan Bortz Higher Level Software 1085 Warfield Ave Piedmont, CA 94611
(415) 258-8948 UUCP: (decvax|ucbvax|ihnp4)!decwrl!sun!dlb!plx!titn!jordan
=============================================================================
End of 80386 mailing list vol 1 #2
------------------------------
Date: Fri, 19 Jun 87 00:49:52 PDT
From: Jim Anderson <bilbo.jta@CS.UCLA.EDU>
Subject: Dhrystones on OS/2
For what it's worth, here are some Dhrystone results obtained on OS/2.
All tests were run on a 10 Mhz 0 wait state AT clone. The same .obj
file was used for both real and protected mode tests. Different .exe
files are required. Compilation was done using the 4.50.01 version
of Microsoft C and options -Gs -Ota.
DOS 3.1 2941 (68 seconds for 200000 loops)
Real mode box of OS/2 2898 (69)
Protect mode 1709 (117)
Protect mode (note 1) 1724 (116)
Protect mode (note 2) 1257 (159)
Note 1: This test was run in protect mode with the real mode box
disabled (protectonly=yes)
Note 2: This test was run in protect mode with the real mode box
in foreground, but no real mode activity.
Timing differences of one second are not significant due to a bug
in the code that tries to determine the overhead of the outer loop.
------------------------------
Date: Thu, 18 Jun 87 21:37:05 PLT
From: "Glenn L. Austin" <AUSTIN%WSUVM1.BITNET@wiscvm.wisc.edu>
Subject: Mice and COM Ports
>We want to purchase mice for our new Zenith 248 PCs (AT clones).
>Our goal is to use a mouse with Microsoft Windows without interfering
>with any of the COM ports. There is a Logitech mouse on the
>Zenith contract. Does that plug into a DB25 RS232 port, or does
>it come with a card that takes a slot? Does the non-RS232 Microsoft
>Mouse 'take up' a COM port, or interfere with one in any way.
The MS Buss mouse is configurable to take any interrupt from 2 to 5
(inclusive). Interrupts 4 and 3 are for COM ports 1 and 2, respectively, and
interrupt 2 is in use in the AT. Interrupt 5 is used for the hard disk on PCs
and XTs. There is no reason that you couldn't use the MS buss mouse with both
COM ports available, unless you have another device that uses INT 5 in an AT
(like an external hard drive, the internal hard drive is remapped to INT 2).
------------------------------
Date: Thu, 18 Jun 87 21:43:37 PLT
From: "Glenn L. Austin" <AUSTIN%WSUVM1.BITNET@wiscvm.wisc.edu>
Subject: DOS 3.3 ECHO OFF Patches
>We want to have ECHO off as the default in BATch files and the
>AUTOEXEC.BAT. This has been addressed in a recent issue of INFO-IBMPC,
>with the appropriate patch locations for COMMAND.COM 3.1 and 3.2.
From what I've read, if you prefix a command with an '@' sign, that command
will not be echoed to the screen. Try putting '@ECHO OFF' as the first line
of your batch files.
I've worked a lot with batch files (my machine runs a BBS after hours so
my batch file is set up to automatically run my routines from 8 to 5 and
the BBS at other times). I've been using environment variables for almost
everything (why did MS put a limit on about 100 chars in the environment in
DOS 3.1???), including what is currently loaded, what times to come up as
the BBS, ect. If anybody is having problems with batch files, let me know!
------------------------------
Date: 18 Jun 87 02:09 GMT
From: a-team @ Walker-EMH.arpa
Subject: Zenith Z-248 Com3 Access
Looking for specific information on how to access the Zenith
Z-248's comm 3 port. I have received all sorts of information on
how to allegedly do it but to date have been unsuccessful. I have
tried Crosstalk XVI V3.61, but when I tell it to access COM 2-4 it
says tough. Does anyone have a driver, etc to make it work?
Thanks in advance,
Mark Meaders
P.S. Please respond to a-team 'at' walker.emh.arpa as I have only indirect
access to the Digest. m
------------------------------
From: munnari!wapsyvax.OZ!doug@seismo.CSS.GOV (Doug Robb)
Date: 18 Jun 87 09:04:04 GMT
Subject: Serial Port Driver for 4 Ports
Organization: Psychology, University of Western Australia
I have a four port ASYNC/AT communications card for
the ibm PC,XT,AT and compatibles (made by AST research).
The card is just four ports (NS16450's) using an RS-232 interface
to handle asynchronous data transmission between pc and serial devices.
What I am looking for is a driver for this (or a similar) card
preferably written in C.
Also in the event of having to write one myself if anyone
has a sample C program that does the required setting up and
trapping of a hardware interrupt on the ibm ps/xt/at this would
be most valuable. I have a microsoft C compiler but really I
just need to get the idea of what needs to be done.
Thanks. doug@wapsyvax.oz
[There are numerous such drivers in the library that could be expanded
to 4 ports. -wab]
------------------------------
Date: 18 Jun 1987 08:05-PDT
Sender: STJOHNS@SRI-NIC.ARPA
Subject: Toshiba 1100+ Async Port Strangeness
I'm trying to find info or pointers to info on the Toshiba 100+
laptop serial port. I got one of these beasties and can't seem
to get either Kermit or Crosstalk to work properly with the
built-in serial port.
Symptoms: I've hooked it up to a modem with the proper cable and
am able to get characters from the keyboard to the modem. No
such luck getting characters back from the modem. I don't think
its the hardware because a program called "Brooklyn Bridge" uses
the serial prot in a bi-directional mode and works fine. (I've
also hooked it up to another IBMPC running KERMIT and have been
able to get characters from the laptop to the IBMPC, but not the
reverse.)
Please reply directly as I don't yet subscribe to the list. Mike
[Many lap tops don't use the same serial port chip as in the PC. The
Kermit people have some special versions of Kermit for some of these
lap tops. I don't know what the Toshiba 1100+ uses for a serial port.
-wab]
------------------------------
Date: 18 Jun 87 11:20:00 EST
From: "James L. Blue" <blue@ceee-sed.arpa>
Subject: Lattice C forkl
I have been escaping to DOS from within a C program (Lattice 3.10,
small, or 'S' model) using "forkl" to start a new copy of command.com.
("forkl" is similar to Microsoft C's "spawnl".) This works if there is
enough extra memory WITHIN THE 64K LIMIT, but not if I am using the
whole 64K. Naturally, I have lots of memory available above the 64K.
Is there a way around this, either within C or through assembly
language? If through assembly language, does anyone have a routine I
could adapt to call from C?
Jim Blue, National Bureau of Standards
------------------------------
Date: Thu, 18 Jun 87 11:09:17 PDT
From: 5103%LSN.MFENET@nmfecc.arpa
Subject: OS/2 and 80386
I have a question about PS/2 Model 80, the one with the
80386, running OS/2. As I understand it, OS/2 is basically an
80286 operating system. It does not itself make use of any of
the 80386 features, particularly the 32-bit registers. Is there
anything to prevent an assembly language program from making use
of the 32-bit registers in sections which do not involve calls to
the operating system?
This would be possible only if the assembler has the
instructions for 32-bit operations. Does anyone know whether the
new generation of assemblers from IBM/Microsoft has such
capabilities? If not, is there any reason why a third party
couldn't market a more powerful compiler with these features?
Similar comments could be made about high-level language
compilers.
[The 386 does 32 bit arithmetic by prefixing a normal instruction
with a "do the next instruction in 32 bit mode" instruction. Even if
you have an older assembler you can insert this prefix with a data
statement. -wab]
A reason why this might not be possible is that the
operating system makes use of 80286 protected mode, which sets
bounds on the memory addresses which a particular program is
allowed to access. Since the 80386 can use 32-bit offsets, this
would violate the operating system's assumption's about
restricted access. I could imagine an operating system dealing
with this on the 80386 by somehow disallowing any use of the 32-
bit registers. Does anyone have any knowledge, guesses, or
thoughts on this matter?
CC:
INFO-IBMPC%C.ISI.EDU@NMFECC.ARPA
5103
[See the programs 386bug.c and 386bug.asm in the lending library. Yes
it is legal to use some of the 386 instructions and yes the Microsoft
C compiler will put out the appropriate code to do 32 bit arithmetic
if you are running on a 386. Microsoft warns users not to use segment
registers as scratch registers. Only load valid segments into segment
registers. If you want to do silly things with segment pointers, go
ahead. Hopefully the system will just slap your hand, but you might
be able to bring the system down in flames if you are real clever. -wab]
------------------------------
Date: 18 Jun 1987 14:38:57-EDT (Thursday)
From: "Victor S. Miller" <VICTOR%YKTVMX.BITNET@wiscvm.wisc.edu>
Subject: SCCS under DOS?
Does anyone know of a version of SCCS (Un*x Source Code Control System)
that runs under DOS? It would seem that the reimplementation from the C
sources wouldn't be hard, with one exception: how would you treat the
manufactured names, like a.c.s. One possible solution would be to
have a special file that would contain the directory that SCCS expects,
and have that file contain manufactured DOS names. Any comments?
Victor S. Miller -- victor@ibm.com
------------------------------
Date: Fri, 19 Jun 1987 11:17 EDT
From: Ray Stell <USDGRS%VTVM1.BITNET@wiscvm.wisc.edu>
Subject: AT Clones
Any suggestions on a great AT clone. Who makes the best BIOS. Also,
this is for small users' group with a small budget...Thanks
------------------------------
Date: Fri, 19 Jun 87 16:10:40 EST
From: Jim Ennis <JIM%UCF1VM.BITNET@wiscvm.wisc.edu>
Subject: Fortran Compiler Recommendations
Hello net,
Can anybody recommend a more complete or powerful FORTRAN Compiler for
the IBM PC family. Are there any recent articles that we could check for
in our library. No one here remembers any articles recently but we could
have missed one due to information-overflow/underflow.
Thanks in advance,
Jim Ennis
University of Central Florida
JIM@UCF1VM.BITNET
------------------------------
Date: 19 Jun 1987 13:35:29 PDT
Subject: Fortran Compiler Recommendations
From: Billy <BRACKENRIDGE@C.ISI.EDU>
To: Jim Ennis <JIM%UCF1VM.BITNET@WISCVM.WISC.EDU>
More complete or more powerful than What? The three major players
are Rayn Macfarland, Lahey, and Microsoft. I gather the new Microsoft
FORTRAN is greatly improved. Ryan Macfarland had the edge for a while
but I don't know how the race is going now that the new Microsoft
compiler is out.
I don't know anything about the Lahey compiler other than that they
have been making FORTRAN compilers for a long time.
Also Microway makes various 8087 routines. For many typical FORTRAN
applications you win big if you code the inner loop in 8087 machine
code. I don't know if any FORTRAN compilers do intelligent register
allocation on the 8087.
We haven't had any discussions of PC FORTRAN in a long time. Is
anyone out there using the latest and greatest of FORTRANS?
------------------------------
End of Info-IBMPC Digest
************************
-------