ICCC 2023 Keynote by Zhiwei Xu on Computing & Communication

Computing & Communication：A New Era and Two Opportunities
by Zhiwei Xu, Institute of Computing Technology, Chinese Academy of Sciences;
Great BayUniversity
[email protected]

With the convergence of the computing and the communication fields, we are entering a new human-cyber-physical ternary computing era, which presents a research opportunity of building a planet-scale cyberinfrastructure and an educational opportunity for coded abstractions.

1. Computing & Communication
A New Era and Two Opportunities
ZhiweiXu
Institute of Computing Technology, Chinese Academy of Sciences
Great Bay University
[email protected]
http://novel.ict.ac.cn/zxu/
Xu ZW, Li ZY, Yu ZS et al. Information Superbahn: Towards a Planet-Scale, Low-Entropy and High-Goodput Computing Utility.
JCST Special Issue in Honor of Prof. Kai Hwang’s 80
th
Birthday, 38(1): 103−114 Jan. 2023. DOI:10.1007/s11390-022-2898-7.
ICCC 2023
Chengdu, 2023.12.09

2. Outline
•A new era: Ternary Computing
–The human-cyber-physical computer
–Open cyberinfrastructure
–Ternary abstractions
•Research opportunity
–Planet-scale cyberinfrastructure
•Education opportunity
–Coded abstractions

3. 1. The new era
•What computing is good for, according to Butler Lampson
Simulation 1950-
today
nuclear weapons, protein
folding, payroll,
games, virtual reality
Communication
(storage)
1980-
today
email, airline tickets,
books, movies, Google,
Terraserver
Embodiment
(physical world)
2010-
...
factories, cars,
robots, smart dust
The Uses of Computers:
The Best is Yet to Come
Butler Lampson’s
presentation at the Nobel
Forum, Beijing, November,
2008

4. Towards the ternary computing era
•Phase 1: 2000 BCE-today
–Manual computing
–Human-machine binary computing
Abacus Pascal Adder Leibniz Multiplier

5. Towards the ternary computing era
•Phase 1: 2000 BCE-today
–Manual computing
–Human-machine binary computing

6. Towards the ternary computing era
•Phase 1: 2000 BCE-today
–Manual computing
–Human-machine binary computing (手工二元计算 )
•Phase 2: 1945-today
–Automatic computing
–Machine unary computing (当代自动一元计算 )

7. Towards the ternary computing era
•Phase 1: 2000 BCE-today
–Manual computing
–Human-machine
•Phase 2: 1945-today
–Automatic computing
–Machine unary computing
单台计算机的计算速度随时间增长趋势
Nordhaus W D. Two Centuries of Productivity Growth in Computing.
Journal of Economic History, 2007, 67(1):128-159.
William Nordhaus
威廉•诺德豪斯
耶鲁大学教授
2018年诺贝尔经
济学奖获得者

8. Towards the ternary computing era
•Phase 1: 2000 BCE-today
–Manual computing
–Human-machine binary computing
•Phase 2: 1945-today
–Automatic computing
–Machine unary computing
•Phase 3: 1970-today
–Man-machine symbiosis
–Modern human-machine binary computing
–当代人机共生二元计算

9. Towards the ternary computing era
•Phase 1: 2000 BCE-today
–Manual computing
–Human-machine binary computing
•Phase 2: 1945-today
–Automatic computing
–Machine unary computing
•Phase 3: 1970-today
–Man-machine symbiosis
–Modern human-machine binary computing
Benefitting
billions of users
•Interactive computing
•Personal computing
•Human computation
•Crowdsourcing
•WeChat, Tiktok

10. Towards the ternary computing era
•Phase 1: 2000 BCE-today
–Manual computing
–Human-machine binary computing
•Phase 2: 1945-today
–Automatic computing
–Machine unary computing
•Phase 3: 1970-today
–Man-machine symbiosis
–Modern human-machine binary computing
•Phase 4: 2010-
–Human-cyber-physical ternarycomputing

11. Towards the ternary computing era
•The 1
st
Era: 2000 BCE-today 人机手工 二元计算
–Manual computing
–Human-machine binary computing
•The 2
nd
: 1945-today 计算机一元计算
–Automatic computing
–Machine unary computing
•The 3
rd
Era: 1970-today 人机共生 二元计算
–Man-machine symbiosis
–Modern human-machine binary computing
•The 4
th
Era: 2010- 人机物三元计算
–Human-cyber-physical ternary computing
–人类正在进入人机物融合的智能万物互联时代

12. 1.1 The universe is the computer
•Or, the planet is the computer
•Computational processes execute
on the human-cyber-physical ternary
computer
人
物机
人机物
三元计算机

13. The universe is the computer
•Human society, cyberspace, physical
world all can be both the subject and the
object of computational processes
•This kind of ternary computer needs new
types of hardware and software
•A person, an AI agent, and a physical
thing can all be the components of this
kind of ternary computer
人
物机
人机物
三元计算机

14. 1.2 Open cyberinfrastructure
•The ternary computer’s infrastructure technology should be open
KrsteAsanovićand David A. Patterson,
Instruction Sets Should Be Free: The Case For RISC-V,
EECS Department, University of California, Berkeley,
Technical Report No. UCB/EECS-2014-146, August 6, 2014
NietubiczThesis
All
infrastructure
technology
should be open
Charles Nietubicz,
2005

15. 1.3 Ternary abstractions
•Traditionally, we use different types of
abstractions to describe processes in
human society, cyberspace, and
physical world
人
物机
人机物
三元计算机

16. 1.3 Ternary abstractions
•Traditionally, we use different types of
abstractions to describe processes in
human society, cyberspace, and
physical world
•Need to synergize them via
computational thinking
–and to produce novel abstractions for the
ternary computer
•Recent advances in AI provide new
ideas for novel abstractions
人
物机
人机物
三元计算机

17. 2. Planet-scale cyberinfrastructure
•The human-cyber-physical
ternary computer also provides a
planet-scale cyberinfrastructure
•However, the vision of utility-like
cyberinfrastructure is yet to be
realized
Irving Vladawsky-Berger, IBM, 2000

18. McCarthy’s utility computing vision
•John McCarthy’s MIT centennial lecture: “Time-Sharing Computer Systems”
–Three essential concepts of computer systems
•Stored-Program Digital Computer
•Time-Sharing Computer Systems
•Computing as a Public Utility (CaaPU)
–Time-sharing (and computing utility) is as important a concept as the
stored-program concept
•The new applications that time sharing will make possible will be of as much additional benefit to science and
management as resulted from the introduction of the stored-program digital computer.
McCarthy J. Time-sharing computer systems. In Management and the Computer of the Future, Greenberger M (ed.), New York: MIT Press, 1962, pp. 220-248.
Stored-Program
Digital Computer
Time-Sharing
Computer Systems
Computing as a
Public Utility

19. McCarthy’s utility computing vision
Computing utility is a telephone-like public utility providing
computing services to subscribers.
–Pay-per-use services. Users subscribe to computing services of the utility
instead of buying computer devices. A user pays only for capacity actually
used.
–Large computer. A user has access to all “programming language
characteristics”, or “computer culture”, of a very large system.
•Computer culture = what a programmer can access (HW+SW)
–Private computer. A subscriber sees a private computer dedicated to his use
as long as needed.
•Performance isolation not yet efficiently achieved

20. Computer science is the
mechanization of abstraction
计算机科学是抽象的自动化
—Alfred Ahoand Jeffrey Ullman, 1992
At the heart of computational thinking
is abstraction
计算思维的核心是抽象
—Alfred Ahoand Jeffrey Ullman, 2022
Alfred Aho Jeffrey Ullman
Turing Lecture Article
Aho, A. and Ullman, J. Abstractions, their algorithms
and their compilers. CACM 65, 2 (Feb, 2022), 76-91.
3. Coded abstractions
Abstraction = Data Model + Its Programming Language
Stanford Textbook
Alfred Ahoand Jeffrey Ullman,
Foundations of Computer Science,
W. H. Freeman, 1992.

21. Cloud Clusters








Cloud computing enables SMEs
•Yunzhanghu.com (云账户)
–Serving 100s million free-lance workers
•Cleaning ladies, repairmen, video creators, delivery men
–Facilitating societal changes
•Income tax payers in China: 5%→10% of population
•Income tax payers in USA: 44% of population
Dr. YongqiangZou
Co-founder & CTO
2016.8 Started
2021 Revenue: ￥50B
2021 Tax: ￥3B
0 to 70 million users in 5 years

22. Cloud Clusters








Cloud computing enables SMEs
•Yunzhanghu.com (云账户)
–Serving 100s million free-lance workers
•Cleaning ladies, repairmen, video creators, delivery men
–Facilitating societal changes
•Income tax payers in China: 5%→10% of population
•Income tax payers in USA: 44% of population
< 200 engineers!
focusing on value
Dr. YongqiangZou
Co-founder & CTO
2016.8 Started
2021 Revenue: ￥50B
2021 Tax: ￥3B
0 to 70 million users in 5 years

23. Cloud Clusters








Cloud computing enables SMEs
•Yunzhanghu.com (云账户)
–Serving 100s million free-lance workers
•Cleaning ladies, repairmen, video creators, delivery men
–Facilitating societal changes
•Income tax payers in China: 5%→10% of population
•Income tax payers in USA: 44% of population
Dr. YongqiangZou
Co-founder & CTO
He studied grid computing at ICT
邹永强的博士论文是 网格计算

24. 麦卡锡-利斯科夫愿景
The McCarthy-LiskovVision
约翰·麦卡锡 John McCarthy 芭芭拉·利斯科夫 Barbara Liskov
1927-2011; 1971 Turing award 1939-; 2008 Turing award
Time-sharing Data abstraction
Lisp CLU, LSP
AI 拜占庭容错
McCarthy’s MIT Centennial Lecture (1961)
1. Stored program
2. Time sharing
3. Computing as a Public Utility (CaaPU)
Liskov’s 2009 Award lecture
•Internet Computer
•Storage and computation
•Semantics, reliability,
availability, security

25. 麦卡锡-利斯科夫愿景
The McCarthy-LiskovVision
约翰·麦卡锡 John McCarthy 芭芭拉·利斯科夫 Barbara Liskov
1927-2011; 1971 Turing award 1939-; 2008 Turing award
Time-sharing Data abstraction
Lisp CLU, LSP
AI 拜占庭容错
McCarthy’s MIT Centennial Lecture (1961)
1. Stored program
2. Time sharing
3. Computing as a Public Utility (CaaPU)
Liskov’s 2009 Award lecture
•Internet Computer
•Storage and computation
•Semantics, reliability,
availability, security
•The Power of Abstraction

26. Coded abstractions
•At the heart of computing is abstraction
•At the heart of communication is ？？？

27. Coded abstractions
•At the heart of computing is abstraction
•At the heart of communication is coding?
–as in coding theory 编码理论
•We need to research and teach abstractions for the
human-cyber-physical ternary computer
•Especially important are novel coded abstractions for
cyberinfrastructure

28. “Greater Bay Area” as defined
includes nine cities in
Guangdong Province, plus
Hong Kong and Macau
Great Bay University 大湾区大学 [email protected]
松山湖校区
滨海湾校区

29. 谢谢!
[email protected]://novel.ict.ac.cn/zxu/

30. social networks
computational finance
……
bioinformatics
deep potential
……
Five thrusts with computing core
communication
electronicsbig data for
social science
AI for science
computing
sensor integration
electromagnetic measurement
……
6G architecture
all optical switching network
……
systems
artificial intelligence
theoretical CS
……

31. One Worldwide Computing Utility
•Each user has one universal compute account (UCA) to
access all resources of the utility
–Instead of multiple accounts for silos of cloud vendors
CC BY 4.0
Related work: Cloud Federation, Multicloud, Berkeley Sky Computing,
W3C DID (Decentralized Identifier)
All tasks are issued to and
executed by one global queue,
instead of by separate slices
Alleviate stranding

32. Learn from Three-Phase Traffic Theory
(a) Traffic Jam
Kerner, Boris S. Introduction to Modern Traffic Flow Theory and Control: The Long Road to Three-Phase Traffic Theory. Springer 2009.
Speed Low (6 miles/h)
Throughput Low (1.2 car/s)
User
Experience
Not on time
Advantage
•Flexibility, allowed to make various moves
•speeding up, slowing down, changing
lanes, over-taking, and U-turn
Disadvantages
•At high load, chaotic flexibility leads to jam
•Low speed
•Not on time
•Low throughput
Illusions
•High “CPU” utilization

33. Learn from Three-Phase Traffic Theory
(a) Traffic Jam (b) Free Flow
Kerner, Boris S. Introduction to Modern Traffic Flow Theory and Control: The Long Road to Three-Phase Traffic Theory. Springer 2009.
Speed Low (6 miles/h) High (60 miles/h)
Throughput Low (1.2 car/s)Low (4 car/s)
User
Experience
Not on time On time
Advantages
•Flexibility
can make various moves
•High speed
•On time
Disadvantages
•Only work at light load
•Low utilization
•Low throughput

34. Learn from Three-Phase Traffic Theory
(a) Traffic Jam (b) Free Flow (c) Synchronous Flow
Kerner, Boris S. Introduction to Modern Traffic Flow Theory and Control: The Long Road to Three-Phase Traffic Theory. Springer 2009.
Speed Low (6 miles/h) High (60 miles/h)High (60 miles/h)
Throughput Low (1.2 car/s)Low (4 car/s) High (14.7 car/s)
User
Experience
Not on time On time On time
Advantages
•High speed
•On time
•High thruput
•High utilization
“Disadvantage”
•Disallow
arbitrary
moves

35. Beyond Three-Phase Traffic Theory
CarPool
(a) Traffic Jam (b) Free Flow (c) Synchronous Flow (d) Prioritized Synchronous Flow
Kerner, Boris S. Introduction to Modern Traffic Flow Theory and Control: The Long Road to Three-Phase Traffic Theory. Springer 2009.
Speed Low (6 m/h) High (60 m/h)High (60 m/h) High (70/60/50 miles/h)
Thruput Low (1.2 car/s)Low (4 car/s)High (14.7 car/s)High (15.2 car/s)
User
Experience
Not on time On time On time On time

36. A Runtime Architecture
Human-cyber-physical ternary tasks are initialized
by computers, humans, or physical things







Service assemblies are orderly
application programs, accessing
services and responding to task
requests
App in running becomes a grip,
a construct integrated with
runtime that can be monitored
and managed as one entity
Bare-metal & other resourcesare namedto
facilitate accessibility, controllability,
accountability and charging

37. Compared to the K8s system
ISB prototype increases yield and throughput
At high load,
yield increases 30~920%, and goodput increases 30~2320%
High Load

38. At high load, ISB prototype increases
good utilization 6~24X
High Load

39. ISB vs. K8s: Latency & Thruput
Latency
Throughput
Threshold of
User Experience
Current
Cloud
Systems
Research
Goal