HFISCHER@USC-ECLB.ARPA (08/17/84)
From: Herm Fischer <HFISCHER@USC-ECLB.ARPA>
This article discusses the PC/AT's new keyboard layout,
presents some technical observations on the compatibility of
PC/AT floppy and hard disk drives with previous generation
drives, identifies new PC DOS 3.0 commands, and provides
comments resulting from scanning a listing of the new BIOS.
KEYBOARD SCREWED UP AGAIN, DIFFERENTLY
Jerry Pournelle will still have something to write about.
The PC/AT keyboard seems like an attempt to placate human
factors engineers who never used a text editor in their
daily work. They put enormous return and shift buttons,
which forced them to move the tilde to the escape position
(above the tab), the PrtSc to the top of the grey plus
(which is now smaller), and HORRORS, the escape to the left
of num lock (which is now sigle width).
Keytronics will have a field day. Any user of any editor
which intensively uses the escape button will be enormously
upset, because, unless the user has the hands of a basket-
ball player, reaching a numlock-positioned escape with the
right pinkie requires removing the hand from the keyboard,
while for the left pinkie to reach the old position above
the tab is even possible for hands on petite women gymnasts.
Lest IBM question what editor uses the escape button, with
vi, which the editor in Xenix, advertised as IBM's only
operating system to take advantage of the 286's features,
the escape is used as the most frequent key (more than the
return, if you do lots of correcting and inserting of text).
Also, many other editors, such as EMACS clones (now quite
popular on the PC) make heavy use of the escape.
On the nice side, they added LEDs for caps lock, num lock,
and scroll lock. And above the grey minus there is a new
key marked System Request (to wake up your multitasking ker-
nel while Lotus is churning away forever).
The keyboard cable still has the same plug and wiring
arrangement, but since the new keyboard has a faster oscil-
lator and additional functionality, I doubt if previous
plug-compatible devices will operate properly in the new
PC/AT (e.g., old keyboards or Logitech mouse adapters).
DISK DRIVE INCOMPATIBILITIES
According to the new Technical Reference manual, disk drive
cables have been redesigned, and I question whether the $175
Teac's will be able to fit in without some changes to their
interfaces. The floppy cables no longer have separate motor
enable lines for each drive; they have a new signal called
reduced write, and a drive select 3 line. I guess this
means that you might be able to squeeze three of the 1/3
height drives shown at the NCC into the space available,
given the third select line.
The hard disk seems to still have a standard interface, but
now the 10 Meg disk's reduced write current line is the 20
Meg disk's head 3 select line, and there are lines for drive
select 3 and 4 (not present in the old cables). One will
have to be careful in buying brand-X 20 Meg disks because of
the head select 3 line.
ASYNC MULTIPLEXOR NEEDED
The interrupt structure still only provides requests 3 and 4
for the async ports. Since this machine's software promotes
use of async-connected mice, and multiple users, obviously
more than two ports is needed. It would be dumb to give
each port a separate interrupt, so I hope somebody builds an
async card, with a single interrupt line, which supports a
bunch of UARTs. I doubt if there would be any serious
technical difficulty in writing either PCDOS or Unix
handlers which share an interrupt and poll a bunch of UARTs.
It certainly would save on interrupt overhead in a multiuser
environment (see my net-note on PC/IX problems at 9600).
NEW ASYNC UART CHIP
The Intel 8250 series UART used in the older PC line has
been replaced with a new National product, the 16450 (said
to be higher speed on the computer interfaces). Users of
8250 clones from foreign suppliers have, in the past,
experienced slight differences from Intel's product opera-
tion. Thus, handler writers should be cautious of the port
part change.
NEW BUS
Six of the eight slots can accomodate new accessory cards,
which have the regular 62 pin connector, and in addition a
supplemental extra 36 pin connector. The problem, however,
is that most of the more recent accessory cards do not have
cutouts which will clear the additional 36 pin motherboard
sockets. I suspect that the reason old async cards (and
presumably also modem cards) won't work is due to changed
oscillators and bus timing.
DOS 3.0 COMMANDS
There are several new DOS commands in 3.0. These are:
a. ATTRIB sets or displays a read-only file flag
b. LABEL adds/deletes or changes volume labels
c. SELECT sets keyboard layout and national date/time
format
d. SHARE installs file sharing support (which is not
described in the DOS manual)
e. COUNTRY sets national date/time format
f. DEVICE installs a RAM-disk
g. FCBs set number of file control blocks
BIOS MODIFICATIONS
The most important change to the BIOS (to my viewpoint) is
that the wait loops were all changed to provide hooks for a
multitasking kernel. The BIOS now accomodates up to 15 Meg
of RAM, in its tables, though I doubt if the > 1 meg stuff
can be used by anything except RAM disks and spoolers.
I quickly read through some of the BIOS code to see if I
could gain any secrets about forthcoming display controll-
ers. Alas, they still support the same old graphics and
monochrome cards, and there are not obvious hints of JR-
style features or 3270 APA-style features comming.
-------jko@ecsvax.UUCP (08/22/84)
Rumor has it that Vectrix of Greensboro, NC is working on a 640 x 400 multi-bit plane 1 million plus color card with fast graphics primitives (arc, box, cube, rubber band, fill, collision detect for points) built in to ROM. Because I am Because I emory on the PC to be reserved for future graphics expansion, is it not reasonable to expect that IBM would provide advanced color graphics on the AT, if they foresaw that possibility on a now inferior machine, the PC?
jko@ecsvax.UUCP (08/22/84)
Rumor has it that Vectrix of Greensboro, NC is working on a 640 x 400 multi- on of such capabilities on a more advanced machine? jonathan k. ocko decvax!mcnc!ecsvax!jko dept. history, north carolina state university
kds@intelca.UUCP (Ken Shoemaker) (08/26/84)
>The Intel 8250 series UART used in the older PC line has >been replaced with a new National product, the 16450 (said >to be higher speed on the computer interfaces). Users of >8250 clones from foreign suppliers have, in the past, >experienced slight differences from Intel's product opera- >tion. Thus, handler writers should be cautious of the port >part change. No offense, but I think the 8250 is also a National part. Intel has a UART called the 8251a, which is kinda close in number, but quite different in functionality, for what it's worth. -- Ken Shoemaker, Intel, Santa Clara, Ca. {pur-ee,hplabs,amd,scgvaxd,dual,idi,omsvax}!intelca!kds ---the above views are personal. They may not represent those of Intel.