COMMENT ⊗   VALID 00003 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	.Sec(An α-Micro Computer System,Proposal)
C00011 00003	.next page
C00016 ENDMK
C⊗;
.Sec(An α-Micro Computer System,Proposal)

The α-Micro Computer System is an impressive array of hardware and
software developed for the Western Digital WD-16 micro computer,
a close relative to the DEC LSI-11. This note outlines how
the α-Micro Computer  System  will be able to solve three of our basic needs.

First, this system will be able to satisfy many of
our current and projected computing needs.
The sytem is a multi-user time sharing system, including UCSD Pascal,
extended BASIC, LISP, and with a version of C⊗↓C is a moderately high level 
language which is the main implementation language on the UNIX system.
In fact, Unix is itself written in C. If C is made available then it will
be reasonably easy to convert UNIX software to run on the α-Micro.← in
the works. The lower level software
includes a macro assembler and symbolic debugger. For text production
the system includes both a text editor and a formatting program for
document production. 
For example, the text formatter
could be used to prepare correspondence. Given a suitable output device,
high quality memos could be produced internally with a much shorter
production cycle. 

Second, The α-Micro can be developed into  an "in-circuit emulator" for the LCDS.
It will be necessary to debug the hardware and software of the development
system. We can build hardware interfaces from the α-micro to the prototype
LCDS, and can write software on the α-Micro to download to the new machine
and control its execution. What better place to  debug our ideas about
ICE languages than our own project!

Finally, the third use of the α-Micro is the  tool development and integration
phase of the human interface breadboard.
The multi-user terminal system hangs on the I/O  interface.
We can make one terminal  "special" in that input will be  through
the serial interface, but the output  will be through a dedicated area  of
memory, mapped to  a display through a Micro-Diversions Screensplitter.
This device supports a display of 40 lines of 86 characters each.
An additional feature is its ability to maintain 
an arbitrary number of rectangular windows. 

Given the system, the development of the low level tools begins. First,  a
display editor with the basic  characteristics of the Stanford AI  editor.
This will  give  us a  viable  in-house  text production  system  for  both
document and  program  maintenance.  Assuming  system  availability,  this
should take no more than two months.

The other  low level  tool which  should be  produced is  a display  based
debugger for P-code which the UCSD Pascal compiler produces.
 This debugger  will
be patterned after RAID, Stanford's  symbolic debugger.  It will give  the
user the ability to  single step, execute,  breakpoint, etc., through  the
compiled P-code. This system, again, should take an additional two months to
develop.

At this  time we  will have  an in-house  development system  capabable  of
supporting both  traditional (hard-copy)  interaction through  the  
vendor-supplied  software  and  display-based  
interaction  on  the  experimental
station. The day of judgement can begin; but it can then begin with  solid
basis for comparison.

Assuming the immediate  existence of  the purchased  system, the  schedule
says that by April 1979 we will have the basic tools to begin the task  of
integrating the components  into a homogeneous  system.  For example,  the
integration of ICE languages, on-line user documentation aids, and program
modification tools  must be  done  carefully.  A  more complex  task,  the
integration of a true high-level  development system can also begin  using
this collection of tools.

To reiterate the theme of the previous memo, I feel most strongly that  an
integrated system  must be  the result  of much  testing, exploration  and
modification.   There  are  many  parallels   between  the  design  of   a
computational system and the design  of an automobile. Within the  general
constraints of a transportation system  (computer system) there are  many,
many variants; the  engine (cpu)  and driver console  (interface) are  but
components.  Manufacturers  do  not  write  down  specifications  for  "an
automobile" and build it. Many configurations fit the "an automobile"
requirement --everything from an Edsel to a BMW. It is equally true that
automotive designers do not begin haphazardly, without any plan. They will
examine  past experience within the industry, using those components which
have lead to success in the past and adding their creativity,
hoping to improve the final product. 
  The  design process is  incremental and iterative;  necessary  
decisions and judgements
can only  be made  on the  basis of  experience with  a viable  system.  
A product is a combination of economics, ideas, and iteration⊗↓otherwise
Henry Ford might have built a Porsche Carrera in 1928.←.
It is with these considerations in mind that I pattern
my exploration on the tools of the Stanford AI community. I can draw 
on  a firm basic of display-based interactive tools, begun at Stanford
in 1965 on a DEC PDP-1, transferred to the PDP-6/10 family, and most recently,
to an HP desk-top micro-computer. 

Since the α-Micro System is multi-user, the three areas of endeavor
can all be done in parallel. That is, the system may be used for word processing,
Pascal, BASIC, or LISP experimentation, and for interface design to the
LCDS, even  while the human interface breadboard is being developed.
The power and flexibility offered by the α-Micro System are unique in the
current micro-computer market; it offers us a very solid basis on which to
build a successful LCDS.
.next page
.SS(Proposed Budget for an α-Micro System)

The following costs were obtained on Dec 8, 1978 from Art Dickerson,
Manager of the Micro Byte Store at 2626 Union Ave. San Jose.
(408)377-4685.

.BEGIN TABIT2(22,54)
AM-100\Cpu\$1495.
.Begin fill;single space
.indent 10,10,10
The Cpu boards implement a 16-bit processor similar to the LSI-11,
including hardware floating point. I have full specification of the
instruction set available for inspection. Please note;
the Cpu  cost includes  ALL of the software.
.end

AM-200\Floppy Disc Controller\$695.
.begin fill;single space
.indent 10,10,10
This controller uses DMA, supporting up to four drives, and use
either IBM format (256K/floppy) or α-Micro format(315K/floppy).
.end

AM-300\Six-port Serial Interface\$695.
.begin fill;single space
.indent 10,10,10
This board will support up to six terminal-like devices at rates from
75 to 19,200 baud.
.end

64Kbytes Piceon\250Ns memory\$1400.
64Kbytes Piceon\250Ns memory\$1400.
.begin fill;single space
.indent 10,10,10
I have specified two blocks of memory to allow better support for the
multi-user system. The floppy-based operating system does not
support swapping, rather it partitions available memory between users.
An extra $1400. is a reasonably low price to pay for the extended 
flexibility for the users. Please note; α-Micro also offers  hard disc
systems which do support swapping.
.end

1070 Persci Dual Disc Drive\$2000.
\Case
\Fan
.begin fill;single space
.indent 10,10,10
The CDC drives which are available on the TWIN are not compatible with
the α-Micro System. These drives were recommended by the Byte Store.
.end

.group;
Equibox\Mainframe\$800.
\Wunderbuss motherboard
\Power supply
\Fan
\Card guides
\Connectors
\\_________
\\$8485
\\   551.53 tax
\\_________
\\$9036.53
\\  200  labor (assembly and checkout)
.apart
.END


This configuration does not include terminals or printers; I assume
we can supply those.

.group skip 2
.Cent(Additonal Hardware for the Human Interface)

The special display processor requires three additional parts.
.begin tabit2(22,54)

Micro-Diversions\Screensplitter\$395
.begin fill;single space
.indent 10,10,10
This card supplies the  display memory (4Kbytes), the character generator
prom, and the window handling software. It generates the signals
for the video monitor.
.end

TV monitor\Sanyo 15"\~$200.
.begin fill;single space
.indent 10,10,10
We may be able to locate a comparable monitor in-house. Another
option is to modify a home TV set. Stanford personnel have had
excellent results modifying  such sets to the 14MHz range.
.end

Keyboard\\~$200.
.begin fill;single space
.indent 10,10,10
The  Interface involves the interpretation of keycodes in non-standard
manners, using the control bit to indicate the occurrence of commands
to the system. We may be able to supply our own keyboard.
.end
.end