SUPERCOMPUTERS

• Document Type: CREST
• Collection: General CIA Records
• Document Number (FOIA) /ESDN (CREST): CIA-RDP85B01152R000700910068-9
• Release Decision: RIPPUB
• Original Classification: U
• Document Page Count: 25
• Document Creation Date: December 21, 2016
• Document Release Date: June 27, 2008
• Sequence Number: 68
• Publication Date: January 25, 1983
• Content Type: MEMO

Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
ROUTING AND TRANSMITTASLIP
TO. ,Name,, office sytmbol, room number,
Mme., Agency/Post)
1. "
Inlt'als
Data
IA
a 96
3.
31 S issa
4.
.~ ~gv~
ion
File
Note and Return
roval
For Clearance
Per Conversation
_ s Requested
For Correction
Prepare Reply
irculate
For Your information
See Me
ment
Investigate
Signature
Coordination
Just i
FROM: (Name, org. syml Room No.-Bldg.
$041-1o2 n FORM 41 (Rev. 7-76)
0 t8. o - -52s, (1: ~ k FR) 101-1L2OA
Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
ri
Y)
 Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
EXECUTIVE SEC TARIAT
Routing Slip
ACTION
INFO
DATE
INITIAL
1
DCI
2
DDC
3
EXDIR
4
D/ICS
5
DDI
6
DA
7
DDO
8
DDS&T
9
Chm/NIC
10
GC
11
IG
12
Compt
13
D/EEO
14
D/Pers
15
D/OEA
16
C/PAD/OEA
17
SA/IA
18
A0/DCI
19
C/IPD/OIS
E
22
Remarks:
att.~.: ;er, ccays ya}' iuat a ,; _L nom.:!.
.,q i o t u e
Execu`tive` Secretary
Dot.
Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
 Approved For Release 2008/06/27: CIA-RDP85BOl 152R000700910068-9
0 The Director of Central Intelligence ?
DD/A Registry
DCI/ICS 83-4224
a) U W lJ,
f~r o
MEMORANDUM FOR: Director, Defense Intelligence Agency
Director, National Security Agency
Executive Director, Central Intelligence Agency ,
Jl?i 1
----...' ., .. ..v. , .?, ?. ?. , ?yc?~~.c VUIlnnUU I Ly JLai I
Chief, Policy & Planning Staff
SUBJECT: Supercomputers
1. The ad hoc committee on Supercomputers convened by the Office of
Science & Technology Policy (O/S&TP) met on 24 January 1983 to review the many
STAT
dramatically; consequently, no consensus could be developed. STAT
Assistant Director, O/S&TP, tasked the members to respond to the following
questions by Monday, 7 February 1983.
A. Do you currently own and operate or time share a
supercomputer? How many or how much usage?
B. Do you have firm plans to acquire the next generation
supercomputer? What specs? When? How many? Approximate dollar
funds per supercomputer?
C. Is there a difference in your requirement between a "big number
cruncher" and an even bigger and more complex Al-based machine?
D. Have you identified a US source or sources?
E. What impact would a successful Japanese fifth generation and
supercomputer with the approximate performance characteristics
have on your agency? (See attachment)
F. Is there a role for the federal government in stimulating,
partially funding, or actually developing the fifth generation
supercomputer. One body of opinion asserts the US Government
should only indicate the. range of performance characteristic
desired, the best estimate on quantity required and probable
time frame for delivery with a tolerable price range.
G. Do you have an opinion on the probable success or range of
performance the Japanese are likely to achieve? Basis for
opinion?
Approved For Release 2008/06/27: CIA-RDP85BOl 152R000700910068-9
 Approved For Release 2008/06/27: CIA-RDP85BOl 152R000700910068-9
?
2. My assessment is that there is not now a strong commitment for US
Government involvement in any aspect of the next supercomputer, nor to any
concentrated reaction to the potential Japanese computer project.
3. Please provide me your comments (which may be classified if
b noon, Monday, 7 February. I will provide them to
office and will obtain and circulate all members' comments for STAT
your individual review and comment prior to developing an Intelligence
Community position.
Attachment:
Japanese Computer Specifications
(20 Charts)
.STAT
Approved For Release 2008/06/27: CIA-RDP85BOl 152R000700910068-9
 Approved For Release 2008/06/27: CIA-RDP85B01152R000700910068-9
SUBJECT: Supercomputers
Distribution: (DCI/ICS 83-4224)
1 - each addressee w/att
1 - ER w/att
1 - A-D/ICS w/o att
1 - ICS Registry w/att
1 - ICS/Info Handling Committee Staff w/att
1 - ICS/PPS Subject (Supercomputers) w a
1 - ICS/PPS Chrono w/o att
DCI/ICS/PP~ (26 Jan 83)
Approved For Release 2008/06/27: CIA-RDP85B01152R000700910068-9
STAT
STAT
 Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
JAPANESE GOALS
?
JAPANESE GOVERNMENT (MITI) AND COMPUTER
INDUSTRY WANT TO BE LEADERS IN SUPER
COMPUTING
THREE NATIONAL PROJECTS
COMPONENTS
- HIGH-SPEED COMPUTER
- FIFTH GENERATION COMPUTER
NUMEROUS INDIVIDUAL COMPANY PROJECTS
COMPUTING ILoe Alamos
Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
 Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
JAPANESE NATIONAL
SUPER-SPEED COMPUTER PROJECT
?
DURATION: 1962-69
FUNDING: $200M
OBJECTIVES: 10 GIGAFLOPS
1 GIGABYTE OF MEMORY WITH
1.5 GIGABYTE BANDWIDTH
100 MEGAFLOPS IN DISTRIBUTED
PROCESSING
PARTICIPANTS:
COMPUTING
ETL
FUJITSU
HITACHI
NEC
MITSUBISHI
OKI
TOSHIBA
LosAlanve
Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
 Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
FIFTH GENERATION COMPUTER
STARTS APRIL 1982
OBJECTIVE: A LARGE, INTELLIGENT COMPUTER SYSTEM
FOR
- LANGUAGE PROCESSING
SPEECH AND IMAGE PROCESSING
- EXPERT SYSTEMS
THREE PHASES
1ST
- 3
YEARS
- FUNCTIONAL MODULES
2ND
- 4
YEARS
- BUILD A PROTOTYPE
3RD
- 3
YEARS
- COMPLETE THE SYSTEM
AND SOFTWARE
$45M OF GOVERNMENT FUNDS FOR FIRST PHASE
COMPUTING LosAhmos
Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
 Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
NIGH LEVEL INQUIRY NATURAL LANGUAGE.
LANGUAGE SPEECK PICTURE
COREGE
LANGUA
?
KNOWLEDGE
MANAGEkENT~~
SYSTEM
KNOTADGE
~ E43iIii/
RELATIONAL
ALGEBRA
REU~TIONAL DATA
BASE MECHANISM
LOGIC PROGRAMMING LANGUAGE
ABSTRACT
DATA AST YPTE
MECHANISM
DATA FLOW
PROCESSING
MECHANISM
I NNOl01T1 VE
VON NEUMAN
MECHANISM
V
INTELLIGENT
INTERFACE
SYSTEM
INTELLIGENT
INTERFACE
MACHINE
BASIC SOFT mum
SYSTEM
HARDWARE
SYSTEM
DISTRIBUTED FUNCTION
E1rORf( SY51EM1
;~i
V
BASIC CONFIGURATION IMAGE OF THE FIFTH GENERATION
COMPUTER SYSTEM (JAPANESE)
COMPUTING
V
Los Alamos
Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
 Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
5TH GENERATION COMPUTER: SPECIFICATIONS
?
HIGHER PERFORMANCE LEVEL AT A LOWER COST
"TRIPARTITE BRAIN"
1) INTELLIGENT INTERFACE SYSTEM
ACCESS THROUGH NATURAL LANGUAGE AND PICTURES
2) PROBLEM-SOLVING AND INFERENCE SYSTEM
- HANDLE MANY MORE GENERAL PROBLEM-SOLVING TASKS
THAN TODAY'S MACHINES
- BE ABLE TO LEARN, ASSOCIATE, AND INFER
3) KNOWLEDGE-BASED MANAGEMENT SYSTEM
- BE ABLE TO UNDERSTAND AND USE STORED INFORMATION
"KNOWLEDGE BASES" RATHER THAN "DATABASES"
EACH OF THE THREE SYSTEMS HAS OWN SPECIALIZED MACHINE WITH VLSI
ARCHITECTURE
COMPUTER SIZES TO RANGE FROM MICRO'S TO MAINFRAMES
SOURCE: TOM MANUEL, BYTE, 5/82
Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
 Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
5TH GENERATION COMPUTER: EXAMPLE:PROJECTS AND SPECIFICATIONS
0
PERSONAL WORK STATION
- PERFORM 2 MIPS
- HAVE .5 TO 5 MEGABYTES OF MEMORY
- HAVE 100 MEGABYTES OF DISK STORAGE, WITH AN AVERAGE ACCESS OF
1 MILLISECOND
"SUPER HIGH-SPEED PROCESSOR"
PERFORM 1 TO 100 BILLION FLOATING POINT OPERATIONS PER SECOND
HAVE 8 TO 160 MEGABYTES OF MEMORY
FLOP)
- PROBLEM -- SOLVING AND INFERENCE FUNCTION
- PERFORM 100 MILLION TO 1 BILLION LOGICAL-INFERENCE OPERATIONS PER
.SECOND
(1 LOGICAL INFERENCE = 100 TO 1000 INSTRUCTIONS)
NATURAL LANGUAGE PROCESSING SYSTEM
KNOWLEDGE-BASED MANAGEMENT FUNCTION
RETRIEVE 1 UNIT OF KNOWLEDGE IN SEVERAL SECONDS FROM A BASE OF 100
TO 1000 GIGABYTES
SOURCE: MANUEL, BYTE, 5/82
Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
 Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
5TH GENERATION COMPUTER: EXAMPLE;PROJECTS AND SPECIFICATIONS
- VERY-LARGE-SCALE INTEGRATION TECHNOLOGY
- AT FIRST, HAVE 1 MILLION TRANSISTORS PER CHIP
- EVENTUALLY HAVE 10 MILLION TRANSISTORS PER CHIP
SOURCE: MANUEL, BYTE, 5/82
Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
 Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
Machine translation system
? Translations among multiple languages
? Vocabulary size: 100.000 words
? Machine to guarantee 90% accuracy, with remaining 10% to be processed
through intervention by man.
? System to be an integrated system where computers participate in individual
stages ranging from text editing to printing and of translations.
? Total costs Involved to remain at 30% or lower than for human translation.
Consultation systems
? Specimen applications
? Medical diagnosis
? Natural language comprehension
? Mechanical equipment computer-aided design
? Computer user consultation
? Computer systems d as osts
? Number of objects: 5,000 or more
? Inference rules: 10,000 or more
? Semi-automated knowledge auuufsitlon
? Interfaces with system: Natural languages and speech
? Vocabulary size: 5.000 words or more
Figure I. Subjects and 10-Year Targets for Basic Applications Systems
SOURCE: REX MALIK, COMPUTERWORLD/EXTRA
11/17/82, P. 25
Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
 Approved For Release 2008/06/27: CIA-RDP85B01152R000700910068-9
I-n-itial oast Development of basic Computer technology . Middlestale. Dealopmtnt of subsystems Find slate Development of total system
1 Modules of indt.duat functMnal mtehan4ms for 1 (Esptrhnental smW oak tustyttem.l
Inference subsynttm
type Inftrtoce bsatc nwchuutm (Inferenct wbtysitm)
Part8tl
Data no. InttUrttnt I tiflct tnft.atr
I rtvJ.snusn _
Ptoblemsolnn{ u,d
rRcenct,.dr.ste
Abstract WE tree wppu+n mecMttm
I
nluence n.echar..vn
Srenulurxt for e.n.r,r.ene.:.,.....e..-
L ntqu-t for thvtt= in VLSI, I - - - - - ,- - - - _ - IndRtecm prrprrnmrnt tnf~.ur
Nodukt fa tndMd
l f
l
us
unctrotu
mecharusmt fur
I itsrwledte bat nibtytttm
Knwkdte bat base wchan6m -
' P
l
atrot re
ational and knn.kdee
..y~..tJon rt.cntnmw
LSmnulston for eaptrrrremal opernton
Ttciu? qutt for urretrsrron,n VILSI1
Problem rolrmt and mftteace ,oft.ue modukr
Kno.ledte base manatrrnent wfwue mniule,
Intemstnt tnte7fsct pft.ltt mudulet
I InteUgtnr propsrnnunt soft.ue modukt
Infetence and kno.ladte
lute mtchannm
ftaittedin VLSI*)
Kno.kdte bsw manatrnrenr
rof t.sre
kno-krip bate rt,echaru,m
Figure 2. Stages of Fifth-Generation Research and Development
SOURCE: REX MALIK, COMPUTERWORLD/EXTRA
11/17/82, P. 25
Approved For Release 2008/06/27: CIA-RDP85B01152R000700910068-9
 Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
5TH GENERATION COMPUTER: NEW TECHNOLOGIES
NEW .TECHNIQUES TO BE USED
- NEW ARCHITECTURES LIKE DATA-FLOW MACHINES
- ARTIFICIAL-INTELLIGENCE CONCEPTS
- LANGUAGES SUCH AS LISP AND PROLOG WITH MACHINES OPTIMIZED FOR THEM
TECHNOLOGIES CURRENTLY EXCLUDED FROM PROGRAM
- EXAMPLES
- GALLIUM ARSENIDE
- JOSEPHSON JUNCTIONS
- RESEARCHERS FELT THESE TECHNOLOGIES WOULD NOT BE SUFFICIENTLY DEVELOPED
FOR GENERAL USE BY 1990
THEY WILL BE INCLUDED AT SOME INTERMEDIATE STAGE IF OUTSIDE
RESEARCH GAINS OCCUR
SOURCE: MANUEL, BYTE, 5/82
Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
 Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
5TH GENERATION COMPUTER: DESIGN AUTOMATION SYSTEM (PAGE i)
CONSISTS OF THREE PARTS
1) SOFTWARE FOR AUTOMATED DESIGN OF VLSI
PLAN TO INITIALLY IMPLEMENT HSL (HIERARCHICAL SPECIFICATION LANGUAGE)
HSL CONTAINS SEVERAL MODULES INTEGRATED INTO A TOTAL
DESIGN SYSTEM
CURRENTLY BEING USED AT THE i1USASHINO ELECTRICAL
COMMUNICATION LABORATORY OF NIPPON TELEGRAPH AND
TELEPHONE PUBLIC CORPORATION
2) SYSTEM 5G - THE COMPUTER SYSTEM TO RUN IT
- PLAN TO USE CONVENTIONAL 40 MIPS GENERAL-PURPOSE COMPUTER
UNTIL FIRST 5TH GENERATION COMPUTERS ARE AVAILABLE
SOURCE: MANUEL, BYTE, 5/82
Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
 Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
5TH GENERATION COMPUTER: DESIGN AUTOMATION SYSTEM (PAGE 2)
3) 5G PERSONAL COMPUTER - LOGIC-PROGRAMMING WORK STATION FOR DESIGNERS
0
REQUIRES HIGH-SPEED PROCESSING.OF VOICE, GRAPHIC, AND
DIGITIZED IMAGE INPUT AS WELL AS PERFORMANCE AS A PERSONAL-
INTERFACE MACHINE
NO EXISTING PERSONAL COMPUTER MEETS THESE SPECIFICATIONS
SOURCE: MANUEL, BYTE, 5/82
Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
 Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
JAPANESE
ULTRA-H I GH-SPEED. COMPUT I NG FACILITY
COVERED
IN THIS PROGRAM
9
1INPUT/OUTPU1L FRONT ENDL
DEVICES
PROCESSOR
?
?
--------------------- f ......................
HIGH SPEED
PARALLEL PROCESSING
SYSTEM
DISTRIBUTED HIGH SPEED
PARALLEL PROCESSING LARGE CAPACITY
SYSTEMS MEMORY SYSTEM
?
GRAPHICAL
INPUT/OUTPUT
SUB SYSTEM
HIGH SPEED
DISK FILE
COMPUTING Los Atanos
Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
 Approved For Release 2008/06/27: CIA-RDP85B01152R000700910068-9
z
0
Z
r)
m
m
Approved For Release 2008/06/27: CIA-RDP85B01152R000700910068-9
 Approved For Release 2008/06/27: CIA-RDP85B01152R000700910068-9
r
0
2
w w
Pr Pr
0
0
z
m
z
0
L
m
m
Approved For Release 2008/06/27: CIA-RDP85B01152R000700910068-9
 Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
SUMMARY
JAPAN HAS LAUNCHED A NATIONAL
PROGRAM TO BECOME A WORLD
LEADER IN SUPERCOMPUTERS.
ALL SIX MAJOR JAPANESE VENDORS
ARE PARTICIPATING.
?
THE PROJECT IS LIKELY TO PRODUCE
A COMPUTER AND EVEN PARTIAL
SUCCESS COULD HAVE FAR
REACHING CONSEQUENCES.
COMPUTING Lo*AJwnoo
Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
 Approved For Release 2008/06/27: CIA-RDP85B01152R000700910068-9
5TH GENERATION COMPUTER: SOME JAPANESE EXPECTATIONS
- BENEFITS TO BE GAINED
- RAISING PRODUCTIVITY IN LOW-PRODUCTIVITY FIELDS
- PRESERVATION OF INTERNATIONAL COMPETITIVE CAPABILITY BY DEVELOPMENT
OF NEW TECHNOLOGY
- CONSERVATION OF ENERGY AND RESOURCES
- PROMOTE UTILIZATION OF CAPACITY OF AGING CITIZENS
- INFORMATIONALIZATION OF SOCIETY
?
SOURCES: SID, FERNBACH, BRIEF ON THE
JAPANESE COMPUTER INDUSTRY, MAY, 1981
ZEN YAMADA, MEMORANDUM ON RECENT
PUBLISHED INFORMATION, MAY-IT,--1981
Approved For Release 2008/06/27: CIA-RDP85B01152R000700910068-9
 Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
5TH GENERATION COMPUTER: SOME JAPANESE EXPECTATIONS
0
FEATURES OF 5TH GENERATION COMPUTER
CONSIDERABLE DIVERSITY
EMPHASIS ON SPECIALIZATION
NON-VON NEUMANN ARCHITECTURE
COMPOSITE MICRO-ARCHITECTURE
INPUT/OUTPUT OF.DAILY LANGUAGES, CHARACTERS, GRAPHS WITH NO
MODIFICATION
SELF-RECOVERING FUNCTION (AUTOMATIC RECOVERY)
SELF-PROGRAMMABLE WITH SIMPLE INSTRUCTIONS ENO REQUIREMENT
FOR HUGE PROGRAMS)
FUTURE PROBLEM SOLVING BASED ON RECORDED DATA
SOURCES: SID FERNBACH, BRIEF ON THE
JAPANESE COMPUTER INDUSTRY, MAY, 1981
ZEN YAMADA, MEMORANDUM ON RECENT
PUBLISHED INFORMATION, MAY-1371981
Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
 Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
5TH GENERATION COMPUTER: SOME JAPANESE EXPECTATIONS
0
- MAJOR R&D THEMES
- DEVICE TECHNOLOGY
- ARCHITECTURE AND HIGH PERFORMANCE PROCESSES
DISTRIBUTED FUNCTIONAL SYSTEMS
- SOFTWARE ENGINEERING
- INTELLIGENT ROBOTS
HIGH RELIABILITY, SECRECY PROTECTION FUNCTION
SOURCES: SID FERNBACH, BRIEF ON THE ,
JAPANESE COMPUTER INDUSTRY, MAY, 1981
ZEN YAMADA, MEMORANDUM ON RECENT
PUBLISHED INFORMATION, MAY-7-T-1981
Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
 Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9
JAPANESE COMPUTER RESEARCH: SOFTWARE DEVELOPMENT
0
"THE ELECTRONIC COMPUTER BASIC TECHNOLOGY DEVELOPMENT ASSOCIATION"
- COOPERATIVE ASSOCIATION TO DEVELOP SOFTWARE
- MEMBERS
- HITACHI
- TOSHIBA
- FUJITSU
NEC
MITSUBISHI
OKI
MATSUSHITA
SHARP
NEC - TOSHIBA INFORMATION SYSTEMS
COMPUTER DEVELOPMENT LABORATORY
WILL SPEND $56 M (1981-1986)
MITI WILL PROVIDE HALF THE FUNDS
SOURCE: SID FERNBACH, BRIEF REPORT
ON THE JAPANESE COMPUTER
TNDUSTRY,~'~AY- ~98I-
Approved For Release 2008/06/27: CIA-RDP85BO1152R000700910068-9