Lecture 01

Download as pdf or txt
Download as pdf or txt
You are on page 1of 18

EEE 425/591: Digital Systems and Circuits

L-1: Introduction

Fall 2014, ASU


Yu (Kevin) Cao, [email protected], GWC 336

Highlight
Course orientation
Objective, textbook, and grading

VLSI design evolution


History, today, and tomorrow
Integration and abstraction

Key design metrics:


Area, performance, power, reliability, and cost

Challenges and trend for nanoscale design


Reading: Chapter 1
Handout: the syllabus and tentative schedule
EEE425/591, ASU, Y. Cao

Lecture 01

-2-

Basic Information
Instructor: Y. Kevin Cao, GWC 336
Office hours: M/W, 1:30pm-2:30pm; E-mail: [email protected]

Teaching Assistants (Lab at GWC 273):


Office TA (GWC 303): David Ganger ([email protected]),

Tu: 4-5pm, Th/F: 1:30-2:30pm


Lab TAs (GWC 273): David Ganger ([email protected]), Sankalp Jain
([email protected]), Mengbing Liang ([email protected]), Curtis
Mackay ([email protected]), Gaurav Singla ([email protected]),
Mahraj Sivaraj ([email protected])

Textbook:
Digital Integrated Circuits: A Design Perspective, by Jan M. Rabaey, et al.
(http://bwrc.eecs.berkeley.edu/IcBook/)

Other references:
CMOS VLSI Design: A Circuits and Systems Perspective, by Neil H. E. Weste and
David Harris
Logical Effort: Designing Fast CMOS Circuits, by Ivan Sutherland, Robert F. Sproull,
David Harris

Lectures, notes etc. are available at http://my.asu.edu


EEE425/591, ASU, Y. Cao

Lecture 01

-3-

What will you learn


Digital IC design knowledge that is both fundamental and practical,
in the context of key design principles

Pre-requisite: basic understanding of circuits and logic (ECE 334)


Schedule:

Design metrics

Fundamental

Practical

CMOS, interconnect
logic, memory,
cost, performance,
clock, power
power, and reliability
Further study: Silicon technology, analog design, computer
architecture and CAD
EEE425/591, ASU, Y. Cao

Lecture 01

-4-

Assignment and Exams


Homework (12%): exercise your learning
Once every two weeks (6 totally)

Labs (20%): practice the design and tools


5 software labs, using 32nm PDK
Lab 4&5 are combined as a mini design project
Online students: remote login available

Examination: evaluate your knowledge


Two mid-terms (20% each): design fundamentals
Final (28%) on December 3rd
All closed book with one page of cheatsheet
EEE425/591, ASU, Y. Cao

Lecture 01

-5-

Grading Policy
Letter grade depends on the relative distribution, with
+ and (: average; : standard deviation)
EEE 425 (undergraduate)*

EEE 591 (graduate)

A+: top 10%

A+: top 10%

A: > +

A: > + 0.5

A-: > + 0.5

A-: >

B+: >

B+: > 0.5

B: > 0.5

B: >

B-: >

B-: > 1.5

C: > 1.5

C: > 2

D: > 2

*The baseline of 425 will be reviewed later in the semester


EEE425/591, ASU, Y. Cao

Lecture 01

-6-

Highlight
Course information
Textbook, materials, project, and grading

VLSI design evolution


History, today, and tomorrow
Integration and abstraction

Key design metrics:


Area, performance, power, reliability, and cost

Challenges and trend for nano-VLSI


Reading: Chapter 1
EEE425/591, ASU, Y. Cao

Lecture 01

-7-

The First Computing System


Abacus
(3000 B.C. 300 A. D.)
by Chinese
and Mesoamerican
Size: > 10cm
Speed: your finger-run
Power: no sweating
Cost: home made
EEE425/591, ASU, Y. Cao

Lecture 01

-8-

The First Computer


The Babbage Difference Engine
(1832)
Designed by Charles Babbage
Size: 11 feet long, 7 feet tall,
25,000 parts
Speed: mechanical
Power: mechanical
Cost: 17,470
EEE425/591, ASU, Y. Cao

Lecture 01

-9-

The First Electronic Computer


ENIAC
(1946)
by U.S.A.
Size: 1800 ft2
Speed: 40 div./sec.
Power: 160kW
Cost: $486, 804.22

EEE425/591, ASU, Y. Cao

Lecture 01

- 10 -

The First Transistor: A Revolution


Transistor
(1948)
by Bell Labs
Nobel Prize, 1951
Shockley, Bardeen,
and Brattain
Much better scalability and reliability
EEE425/591, ASU, Y. Cao

Lecture 01

- 11 -

The First Integrated Circuit


Integrated Circuit
(1958)
by TI
Nobel Prize, 2000
Kilby
1 transistor, 3 resistors, and 1 capacitor

EEE425/591, ASU, Y. Cao

Lecture 01

- 12 -

The First Microprocessor Chip


4004 CPU
(1971)
by Intel
Size: ~ 9mm2,
2.3 K transistors @10m
Speed: 1MHz
Design team: 3

EEE425/591, ASU, Y. Cao

Lecture 01

- 13 -

The Pentium 4 CPU


MOSFET

Interconnect

P4 CPU (2002)
Size: ~217mm2, 42M @ 0.18m
Speed: 2GHz
Design team: 1000
EEE425/591, ASU, Y. Cao

Lecture 01

- 14 -

The Cell Processor


Used MOSFET
in PS3 and other
applications
8 processor cores
Low-power and highspeed

Cell BE (2006)
Size: ~221mm2, 234M @ 90nm
Speed: 4GHz
Design team: STIR
EEE425/591, ASU, Y. Cao

Lecture 01

- 15 -

Moores Law
In 1965, Gordon Moore (Intel) noted that the number of
transistors on a chip doubled every 18 to 24 months.

Prediction: semiconductor technology will double its


16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0

Electronics,

EEE425/591, ASU, Y. Cao

1975

1974

1973

1972

1971

1970

1969

1968

1967

1966

1965

1964

1963

1962

1961

1960

April 19, 1965


1959

LOG2 OF THE NUMBER OF


COMPONENTS PER INTEGRATED FUNCTION

effectiveness every 18 months

Lecture 01

- 16 -

Moores Law in Microprocessors


Year
Transistors

1980

1990

2000

2010

Chip
Frequency
(Hz)

3.2G

1000M
100M

Itanium
10M

1.0G

Pentium

1M

80386

100K

8086
Technology
Node

200M
33M/ 2 years
Transistor counts: 2x

Channel length: 0.7 5M


/ 18 months

3.0m 1.0m 0.35m 180nm

50nm

EEE425/591, ASU, Y. Cao

Lecture 01

- 17 -

Lecture 01

- 18 -

Evolution in Complexity

EEE425/591, ASU, Y. Cao

Driving Forces
Technology scaling
Semiconductor device shrinks by 0.7x / generation

Circuit and design


Cleverness
Functions per chip doubles every generation;
chip cost does not increase significantly
Cost of a function decreases by 2x

On the other hand:


Design population does not double every two years
Productivity per designer decreases due to

complexities of design and team management


EEE425/591, ASU, Y. Cao

Lecture 01

- 19 -

10,000
10,000,000

100,000
100,000,000

Logic Tr./Chip
Tr./Staff Month.

1,000
1,000,000

Source: Sematech

10,000
10,000,000

100
100,000

Productivity
(K) Trans./Staff - Mo.

Complexity
Logic Transistor per Chip (M)

Productivity Trends

1,000
1,000,000
58%/Yr. compounded
Complexity growth rate

10
10,000

100
100,000

1,0001

10
10,000
x

0.1
100
xx

0.01
10

xx
x

1
1,000

21%/Yr. compound
Productivity growth rate

0.1
100
0.01
10

2009

2007

2005

2003

2001

1999

1997

1995

1993

1991

1989

1987

1985

1983

1981

0.001
1

Profound impact on the way VLSI is designed


Exploit different levels of abstraction
Automated design with CAD tools
EEE425/591, ASU, Y. Cao

Lecture 01

- 20 -

10

Design Abstraction Levels


SYSTEM

MODULE
+
GATE

CIRCUIT

DEVICE
G
D

S
n+

n+

EEE425/591, ASU, Y. Cao

Lecture 01

- 21 -

Top-Down Design Automation


systems
systems
application software
system software

Outp
ut

M2

k
T

h=
3;

k = h; h
>3;
M1

h<
3;

sapchg#

platform / architecture

circuits

LA TC H

s a b it
s a b it #

devices/interconnect
structures
structures
materials
materials
physics
physics
EEE425/591, ASU, Y. Cao

Tool capability is limited and exploited


by our physical understanding
Lecture 01

- 22 -

11

Highlight
Course orientation
Objective, textbook, project, and grading

VLSI design evolution


History, today, and tomorrow
Integration and abstraction

Key design metrics:


Area, performance, power, reliability, and cost

Challenges and trend for nano-VLSI


Reading: Chapter 1
EEE425/591, ASU, Y. Cao

Lecture 01

- 23 -

Design Adaptation
speed
area/cost

1970s

speed

speed/power
/reliability

speed/power
power

low power

1980s

1990s

EEE425/591, ASU, Y. Cao

reliable ultralow power


2000s

Lecture 01

- 24 -

12

Challenges in Future VLSI


Macroscopic Issues

Microscopic Problems

Time-to-Market
Millions of Gates
High-Level Abstractions
Reuse & IP: Portability
Predictability

Ultra-high speed design


Interconnect
Noise, Crosstalk
Reliability, Manufacturability
Power Dissipation
Clock distribution

Main roadblocks: power, cost, and reliability

EEE425/591, ASU, Y. Cao

Lecture 01

- 25 -

Power Dissipation
18KW
5KW
1.5KW

Power (Watts)

104

500W

103

Pentium proc

102

103
102

Power (W)

105

486
8086 386
8080

10

Switching

101
100
10-1
10-2
10-3

1 4004

Source: Intel

Leakage

10-4

10m

0.1
1971 1974 1978 1985 1992 2000 2004 2008

1m

100nm

10nm

Technology node

Year

Power delivery and dissipation will be prohibitive


EEE425/591, ASU, Y. Cao

Courtesy, Intel

Lecture 01

- 26 -

13

Power Density
Power/transistor switching goes down with
technology scaling, but:
Power Density (W/cm2)

10000

Rocket
Nozzle

1000

Nuclear
Reactor

100

Hot Plate
10

8086
4004
8008 8085

1
1970

P6
Pentium

386

286

486

8080

1980

1990

2000

2010
Source: Intel

Year
EEE425/591, ASU, Y. Cao

Lecture 01

- 27 -

Innovative Design Demanded


Module Heat Flux (w/cm2)

14
IBM ES9000

12

Bipolar

CMOS

10
8

Fujitsu VP2000
IBM 3090S
NTT

IBM RY5
IBM RY7

Fujitsu M-780

IBM3090
CDC Cyber 205
IBM4381
IBM3084
IBM370 Fujitsu
IBM360
IBM3033 M380

Pulsar
IBM RY6
IBM RY4
Apache
Pentium II

0
1950

1960

1970

1980

1990

2000

2010

Year of Announcement

Technology used to be the answer for this


trouble; but, no candidate around the corner
EEE425/591, ASU, Y. Cao

Lecture 01

- 28 -

14

Reliability
Process variations
Dynamic uncertainties:
Temperature
Power supply (Ldi/dt noise)

138 W/cm2

Crosstalk
Soft error

All degrade your yield


and waste resource
EEE425/591, ASU, Y. Cao

Lecture 01

- 29 -

Cost of Integrated Circuits


NRE (non-recurrent engineering), i.e., fixed costs
Design time and effort
Mask generation
Design verification

Recurrent costs, i.e., Proportional cost


Silicon processing (proportional to chip area)
Packaging (proportional to volume)
Test (proportional to volume)

EEE425/591, ASU

Lecture 01

- 30 -

15

NRE Cost

EEE425/591, ASU

Lecture 01

- 31 -

Recurrent Cost
Cost per Transistor
10-4
10-5

Dollar

10-6
10-7
10-8
10-9
1m

EEE425/591, ASU

100nm

10nm

Lecture 01

- 32 -

16

Design Trend I: Integration


Integrate various technologies
Integral design solutions
Small
Signal RF

Power
RF

Power
Management

Analog
Baseband
Digital Baseband
(DSP + MCU)

EEE425/591, ASU, Y. Cao

Lecture 01

- 33 -

Design Trend II: Parallelism


Concurrency for better
reliability, programmability,
and cost factor

Examples:
Multiple chip
Array structure

EEE425/591, ASU, Y. Cao

Lecture 01

- 34 -

17

Broader Applications
Digital IC have come a long way and still have
quite some potential left for the coming
decades:
Computation and Communications
Automobile (30-70 chips/car now)
Consumer electronics
Energy conservation
Security and intelligence
Biomedical
Much more with your innovation

EEE425/591, ASU, Y. Cao

Lecture 01

- 35 -

Summary
EEE 425/591: digital design fundamentals and
practices underlying VLSI systems

Design metrics: timing, power, and reliability


Challenges ahead: power and reliability
Clear understanding from a circuit perspective
Office hours: M/W 1:30pm-2:30pm, GWC 336
Contact: [email protected],
Web access: http://my.asu.edu

EEE425/591, ASU, Y. Cao

Lecture 01

- 36 -

18

You might also like