INFO-PC@USC-ISIB (02/24/83)
From: Dick Gillmann <INFO-PC@USC-ISIB>
Info-PC Digest Wednesday, 23 February 1983 Volume 2 : Issue 11
Today's Topics:
C Sieve Times
MODEM7 in Assembler
Hard Disk (2 msgs)
IBM Press Conference
8087 Chip (2 msgs)
Problems With >512K Memory (2 msgs)
VT52.BAS Problem
----------------------------------------------------------------------
Date: 17 Feb 83 17:00:23-PST (Thu)
To: info-pc@isib
From: harpo!ihnp4!ihuxx!hersee@Berkeley.arpa
Subject: C Sieve Times
The Lattice C does the Sieve in 10.5 sec on my PC. I think the
current time for an unmodified Sieve for C86 is greater than 20
seconds. How much has George improved on this time? Please quote PC
times for the Sieve as listed in Byte.
Steve Hersee
ihuxx!hersee
312 979-1872
------------------------------
Date: 19 Feb 1983 1450-PST
Sender: BILLW at SRI-KL
Subject: Re: MODEM7 in Assembler
From: William "Chops" Westfield <BillW @ SRI-KL>
To: hess.unicorn at MIT-MULTICS
Doesn't work well at >1200 baud ???!!! I'm not impressed. The Basic
version I wrote (available in the <INFO-PC> directory) does just fine
at 9600 baud. Of course, it doesn't have the "full environment" of
MODEM7, but it does emulate a good terminal.
BillW
------------------------------
Date: 19 Feb 83 23:44:02-EST (Sat)
From: D. J. Farber <farber@udel-relay>
Subject: Hard Disk Query
To: info-pc@isib
I need to procure a hard disk for the PC. What's the experiences with
various suppliers both on hardware and software? And any idea of what
and when IBM will pop?
Dave
------------------------------
Date: 21 Feb 1983 1147-PST
Subject: Re: Hard Disk Query
From: Billy <BRACKENRIDGE@USC-ISIB>
To: D. J. Farber <farber@UDEL-RELAY>
cc: info-pc
Expect an IBM announcement in March. Our USC IBM salesman even
suggested that we delay our last purchase order to wait for the
announcement. It may not be a hard disk system, but it seems worth
the wait to find out.
------------------------------
Date: 23 Feb 83 16:32:21-EST (Wed)
From: Dave <farber@udel-relay>
Subject: IBM Press Conference
To: info-pc@isib
I gather that IBM will be holding a press conference on 8 March to
make announcements wrt the PC.
Dave
------------------------------
Date: 20 Feb 83 10:51:22-EST (Sun)
From: D. J. Farber <farber@udel-relay>
Subject: 8087 Chip
To: info-pc@isib
I have gotten several 8087 chips for my pcs and now want to put them
in. I remember seeing something about that in some magazine but have
not been able to find it.
Anyone either do it or remember that article? Also any experience
with the "drivers" to the IBM Basic and Pascal compilers for the 8087
offered by Microware in Mass?
Dave
------------------------------
Date: 21 Feb 1983 1142-PST
Subject: Re: 8087 chip
From: Billy <BRACKENRIDGE@USC-ISIB>
To: D. J. Farber <farber@UDEL-RELAY>
cc: info-pc
The chips are easy to install. Just line them up in the same direction
as the 8088 chip and plug them in.
A member of the North Orange County Computer Club sent Microware
several hundred dollars a few months ago and to date has only received
the $20 manual. It is a little tough to sue from California but he
has been pursuing this avenue.
The concept of using a set of 8087 library routines for the existing
Pascal or FORTRAN is in MIT parlance a lose. Benchmarks for 8087
libraries for these compilers have been unimpressive.
Microsoft makes real compilers that support the 8087 and the salesman
tells me these are available for the Victor 9000. Perhaps they are
available for other 8086/8 based machines? I have heard rumors about
the Beta test site versions of these compilers for the IBM-PC.
I believe the most important benefit to be derived from the 8087 is
increased precision rather than speed. The current compilers only
support 32 bit calculations. Changing the library won't help that.
Computation intensive applications such as a FFT run only four or five
times faster using compiled 8087 code, however this factor of four
improvement includes doubling the precision to 64 bit real numbers.
While the internal calculation times of the 8087 are very impressive,
that doesn't translate to lightning fast calculation at the user level
as the overhead of getting data on and off the chip takes most of the
time.
With the addition of the 8087 and a decent compiler it should be
possible to run a program like SPICE. One member of the local IBM-PC
club claims to be running a large matrix inversion program for a
client. He is using the JRAM board with special software to simulate a
(VERY) large FORTRAN array. We used to use these sort of tricks years
ago on a Raytheon 704 to simulate million byte arrays of speech data
on a 64K machine.
I have also heard that FORTH graphics using the 8087 is capable of
very impressive 3D rotations. Of course FORTH allows direct
manipulation of the color display board which helps. If anyone knows
of a Battlezone for the PC that uses this technology, I am a willing
customer.
------------------------------
Date: 23 Feb 1983 0920-PST
From: Richard L. Schwartz <Schwartz at SRI-CSL>
Subject: Problems With >512K Memory
To: info-pc at USC-ISIB
Has anyone heard anything about possible (operating system) problems
in supporting more than 512K memory on the PC? I have a 576K PC-DOS
system. With the entire memory configured, I get sector read errors
when I try to use DISKCOPY to copy a DSDD diskette. When I configure
my memory down to 128K using my RAM disk, everything works properly.
I have also noted that the Infocom Deadline program (and probably the
Zork series as well) does not work with more than 512K.
Richard Schwartz
------------------------------
Date: 23 Feb 1983 1534-PST
Subject: Re: Problems with >512K of Memory
From: Billy <BRACKENRIDGE@USC-ISIB>
To: info-pc
DOS 1.1 only supports 512K of main memory. There are switches on the
motherboard which designate the amount of memory in the system. At
boot time these switches are read and the upper bound of memory is
placed in a low core memory location. DOS then only refers to this
location and never again looks at the switches as that would be
machine dependent and Microsoft wants DOS to be as general as
possible. When programs put a number in this cell larger than the DOS
1.1 limit of 512K programs spec'ed to run under DOS blow up.
The answer is to buy a good memory management/electronic disk package,
such as JEL from Tall Tree. If you buy any software package that
tells you to set switches to indicate phantom disks or tells you to
configure memory at more than 64K throw it away immediately. It will
lead you into real trouble when it comes time to run diagnostics. You
will feel like a real sheep when your friendly Computerland charges
you $50 to turn the switches back. Any decent memory management
system or disk driver system should let you configure the system under
program control, and the switches should be left in a minimal
configuration so IBM diagnostics will only test the IBM gear. When
you run memory diagnostics you should run that manufacturer's memory
diagnostic so you know exactly which chip to replace.
------------------------------
Date: 23 Feb 1983 2219-PST
Subject: VT52.BAS
From: Dick Gillmann <GILLMANN@USC-ISIB>
To: Info-PC
It appears that the new IBM Async. Comm. package does not work with
the VT52.BAS program on Info-PC. Does anyone have a fix for this?
/Dick
------------------------------
End of Info-PC Digest
******************************
-------