Chapter 5
1961-1975
1963- $1.2 billion (70%)
1965- $ 3 billion
1970- $ 7 billion
Sperry – Rand (UNIVAC)- mid 1960’s -
$ 145 million
CDC (mini) , Honeywell, Philco, RCA, GE,
Burroughs, NCR- mainframe builders
Snow White & 7 Dwarfs
1970’s BUNCH
Stable until 1980’s personal computers
Announced April 1964 “full-circle” of customers – business & scientific
Line, 6 models-sw upward compatible
25:1 performance range
Plus 150 different items
Fortune Magazine: “You bet your company”
1100 orders in 1 st month
In 5 months, 2200 orders
Success was a “threat
”
Could not meet demand
Met daily 2 months in 1961
Review past & decide the future
Couldn’t continue to “upgrade”
Needed a “unified product line”
“Scalability”
Fred Brooks, Gene Amdahl
“can’t be done”
1951- Maurice Wilkes, Manchester
University
“best way to build an automatic calculating machine” was to build its control section as a little stored – program computer of its own
“Micro- programmer”; microcode
Page 148
Allowed common instruction set
Specialization & optimization for models
Via microprogramming
Read only memory
Compatibility with earlier IBM’s
Emulation
Separated design process from control logic
Microprogramming facilitated
Emulated earlier IBM models
7070 and 1401
Up to ½ of 360 time was emulating
Due to new HW, new machines
“emulated” old ones up to 10 times faster
Honeywell H-2000 + liberator
emulated 1401 – sold well!
Software - more permanent than hardware
1401 programs run today
Part of Y2K problem
16 General purpose registers
32-bit words
Finally a power of 2!!
8-bit character code – byte (1956)
EBCDIC – true 8 character code
Originally also included ASCII; dropped
Channels retained
Peripherals
PDP-8: real-time, labs, etc.
1963- American National
Standards Institute
Adopted 7 bit ASC II, not 8
8 holes across paper tape would weaken it too much
Became established in mini’s
System/360=> 360 degrees=> “full circle of customers”
Did not address time-sharing
No dynamic addressing
@ this time – interactive=> one user
Batch jobs “shared” time – one job in memory
John McCarthy- MIT
Each user has illusion that complete machine & SW at her disposal
Page 155
Lots of wasted cycles between key strokes
– instructions
No efficient method for program swapping & keeping track
MIT, IBM 7090
CTSS - Compatible time sharing system
Supported just a few users
Defense Dept.- Project MAC
Chose G.E. over 360
GE became leader in time-sharing
IBM crisis
Problems with program swapping
Had not thought it would be important
Model 67: TSS failure
Anti-trust lawsuit – premature announcement
IBM Stretch
Progress was made
– but still slow to develop workable time-sharing -- Demonstrated feasibility but not practicality
Much turmoil in industry
GE sold to Honeywell
Movement to PDP machines - UNIX
IBM threatened by success
IBM 360 could not compete in mini market
System /3
Incompatible with 360
$1000 month
Small 96 column punch card
Eventually floppy disk
Leasing companies
IBM too expensive
Development costs
Others 20% less
Could withhold technology
1970 - IBM 370
360 compat. w/ time-share, integrated circuits (cpu,mm)
1978 - IBM 4300
360 compat, low cost
1978 – AS/400
Leasing companies in debt due to upgrades
“Go-GO” years
Stocks soared for all electronics companies
Possible attacks on
360 line
Information to numerous companies
1964 - CDC 6600
(above)
Designer Seymour Cray
Supercomputer – Fastest performance
Small sales ~ but impressive ~ serious customers
Sued IBM – announcement 360 model 91
1964 – RCA – Spectra 70 Series
Ran 360 Software (1 st clone)
40% less $ than IBM, 4 models
Used true integrated circuits
Lost in 1970 ~ 370 announced (IC’s)
Out of computer business
Another round of buy-outs
Others targeted 1 part ~IBM system
Tape drive, memory, CPU
These companies soared
IBM price cuts, packaging
1969-75, 10+ lawsuits
Gene Amdahl-1970
Left IBM, own co.
Fujitsu, other Japan
Industry needed SW!!!
Companies providing software
Automatic Data Processing (ADP)~payroll
Computer Sciences Corp. (CSC)
American Management Systems
IBM
Included programming, service in price
1968 agreed to separate (lawsuit)
For 360 success need for others to provide SW & service
Bad IBM SW – OS 360, PL/1
1962 - EDS – H. Ross Perot
Left IBM over “leasing of time”
Success with government contracts
Based in Dallas
Ramo-Wooldridge TRW
Manufacturing + SW + Service
Military, space, etc.
Later into business apps.
Filed January 1969
Lasted 12 years
Sell computers, Unbundle
Dropped January 1982
PC Effect
? Why didn’t IBM pursue PC market?
Unisys = Burroughs + Univac (86)
AT&T hostile take-over of NCR (91)
Honeywell
Partnered with NEC, Bull & Olivetti
Phased out of computing
CDC – up & down – suffered Cray departure
PLATO (p. 175)
Education system failure
Ahead of it’s time
Chapter 5 – 1961-1975