Transcript Chapter 4 – 1959 to 1969

Chapter 4 – 1959 to 1969
From Mainframe to
Minicomputer
Russian Sputnik Satellite 1957
1
State of Computing – 1960’s
note: driving forces – pg. 110






Great Demand for Computing
IRS: Mainframe + Punch Cards
JFK: 1961 Moon Challenge
Great economic growth & prosperity
New computers – the MINI
Technology already obsolete
2
IBM

1960





70% Market Share  stability
Research Labs (US, Europe)-slow payoff
Could control market, releases (usually)
Philco – transistor production
Others need to find area unserved by
IBM
3
Government Influence
IBM 7000




Dept. of Defense - needed
computer technology
Researchers began
defining work
Demand for computing was heavy
Govt. had funding – to lots of places

See how this influenced the development
4
Massachusetts Blue Cross




1960 - IBM 7070; 1401 for I/O
6 months- transferred 2,500,000 records
from cards to tape (150 file cabinets)
AUTOCODER (not FORTRAN or COBOL)
1965 Medicare established





Won job of administering for Mass.
Fall 1966 - fully computerized - 1st state
Also rented another 7070 (drove cards daily)
1969 - 3 IBM System /360’s +
COBOL, 24-7
5
NASA- Ames Research Center


Mountain View, California, 1940
Soviet Sputnik 1957 & JFK Moon Challenge


Honeywell H-800 for wind tunnel - real time
+ others for dedicated purposes


Became part of NASA in 1958
DEC, Scientific Data Systems, EAI, IBM
Demands on central computer grew 100%
per year in 1960’s
6
NASA-Ames Research
Center (cont.)

Direct Couple System - 1963



7094, 7040 (I/O), 7740 (Remote Terminals)
Replaced in 1968 with System 360 model 50
1969- System/360 model 67
For time-sharing, not a success
 1971 –connection to new ARPANet

7
NASA-Ames Research Center
(cont.)

Timesharing not a success
Design problems
 Incompatible work patterns
 No longer “real time” for wind tunnel


Table 4.1 p. 118 – computers at Ames
8
IRS- Internal
Revenue Service

During WWII – need $





Taxpayers from 8 to 60
million
Withholding


Punch Card – 1948
IBM 650 - 1955


Kansas City regional
center
1.1 million returns – test
1959 - authorized to
computerize fully



IBM won bid
1401 @ each regional
center
7070 national center
Still 400 million cards
a year
Changeover complete by
1967: cards to magnetic
drum
9
IRS
(continued)
Honeywell H200


No NW – flew tapes/cards around
TAS - Tax Administration System






Late 1960’s – implement new ideas
$650- $800 million
Distributing computing to 10 centers
Direct access via 8,000 terminals
Network
Lot of work in to security plans
10
TAS (continued)



1974 - Nixon resignation / IRS records
Public questioned security
1977- Computerworld – GAO report




“Proposed IRS system may pose threat to
privacy”
Congress held hearings – IRS no trusted
Jan. 1978 - IRS dropped TAS
1985 - IRS system collapsed, bad press

Congress approved change
11
Put a Man on the Moon



Batch would just not work
Needed lots of money for real-time
Mercury Monitor (system software)



Data could interrupt processor to note lifethreatening situations
Trap processor: checked levels of priority;
saved contents of registers
Evolved to real-time version of System/ 360
O.S. a significant advance in system software12
Minicomputer



New, Not a competitor to mainframe
Driven by technology
Factors defining Mini
Architecture
 Packaging
 Third parties
 Price
 Financing

13
Architecture of Mini

Short Word

Implications

Small addresses, values, instructions
Instructions more complex
 Instruction “modes”
 With new transistors, processor still
simple inexpensive and fast

14
First Mini at CDC


1957 - CDC (Jack Norris) developed 1604
Seymour Cray – CDC 160

1960 – CDC Model 1604






Designed model 160 for I/O
Word = 12 bits
8,000 word memory
Fast clock
160A sold for $60,000 (stand alone)
1963- Jack Scanllin, Scanllin Electronics

2-160A’s to provide online stock quotes to
brokers
15
Digital Equipment Corp (DEC)


Founded 1957; K. Olsen, H. Anderson
Olsen: MIT Whirlwind;
MIT’s TX-0 (transistors)


1957-Most advanced in world
1959 - PDP-1



designer Ben Gurley
Designed to take full advantage of transistors
(not a “re-fit”)
100,000 adds/sec; 4,000-6,000 8-bit words
16
PDP- 1 features


Transistors
DMA: defined mini architecture




Interrupts: up to 16 levels
circuits to handle in order
1 IBM channel cost more than PDP-1: $120K
Sold about 50, only moderate success financially, but
17
very innovative.



No I/O channels (with own memory)
Fast, little effect on 1717processor
Cheap and simple
DEC’s Policies



Customer Relations Contrast to IBM
Sold Computers
Encouraged customer modifications



Had sophisticated customers
Necessary - small company
Manuals


“A Sears- Roebuck catalog” for DEC products
Sold cheap or gave away
18
PDP-8 -1965

Sold 50,000;






12-bit word
Germanium transistors - faster
Indirect addressing; Paged memory; innovations
DECtape – portable, r/w both directions, like disk
Price??? $18,000; down to $10,000


plus single–chip versions
Very low; shocked industry
Model 33 ASR - Teletype Corp.

Used ASCII, ctrl & esc Keys (Photo Pg. 134)
19
PDP-8 (cont’d)





Term: Minicomputer
Tendency toward assembly language
OEM – original equipment mfg.
 Added specialized equipment
Early use LS-8 (by Electronics Diversified)
 Contained a PDP-8A
 Controlled lights for A Chorus Line
PDP-8 on a potato picker (P. 136)
20
DEC Culture

The Mill, Maynard, Mass.
1965- $15 mil. Revenues
 1970 - $135 mil.

Proud of differences – IBM
 Eventually competed with IBM - VAX


Read P. 136-139
21
Chapter 4 – 1959 to 1969
From Mainframe to
Minicomputer
22