Lecture 1
Lecture 1
of Digital Systems
Spring 2007
Instructor: Azadeh Davoodi
Lecture 1
January 23, 2007
Course Information
• Lectures: RT 9:30am-10:45pm, 2540 EH
• Web address:
http://www.ece.wisc.edu/~adavoodi/
• Office: 4621 Engineering Hall
• Office hours: TR 11:00am-12:00pm
• Information on academic dishonesty
1
Textbooks
• Main textbook
Sabih H. Gerez, “Algorithms for VLSI Design
Automation”, Wiley, 1999
– does not discuss all the topics in CAD tools and is
relatively old, however provides depth in many core
topics
• Optional:
S. Sait, H. Youssef, “VLSI Physical Design
Automation: Theory and Practice”, World Scientific,
1999
– can be used as supplementary reading for the
physical design lectures
3
Prerequisites
• Elementary knowledge of C or C++
programming
• Some familiarity with digital IC design
– suitable for CS students as well as EE and CE
students
• Elementary knowledge of algorithms
2
Grading
• Midterm (20%)
• Final (30%)
• Homework (25%)
– almost biweekly home works
– include paper home works, programming assignments,
and home works on using academic/commercial tools
– late homework policy: each day late 20% off
• Class project (25%)
– implement a CAD tool (details will be posted soon)
• Effective class participation (up to 5% bonus)
5
3
The VLSI Design Problem
• VLSI design problem: realize a given
specification on silicon with the following
requirements:
– optimize the following metrics
• area
• yield
• power dissipation
• speed
• design time
• testability
7
• etc.
Silicon Compilation
Analogy with compilers for programming languages:
from algorithm to machine instructions
In VLSI:
from algorithm to mask patterns
• Current practice:
– still far away from a single push-button operation
– Q: What is the difference between EDA (Electronic
Design Automation) and CAD (Computer-Aided Design)?
4
Design for Conflicting Metrics
• Area
• Yield (percentage of working chips)
– causes of failure: manufacturing defects
• Power dissipation (trades-off with area)
– important in battery-operated portable devices
– relates to heat dissipation
• Speed (trades-off with power and area)
• Design time
– need to have quick time-to-market or turn-around-time for
economical purpose
– Q: give examples of ways to reduce the design time?
• Testability (trades-off with area)
– need techniques to quickly find the defective chips after
9
fabrication
10
5
Hierarchy and Abstraction:
A
Example
A Level 1
B D
E F
M N B C D Level 2
G H O
J Level 3
I C E F G H I M N O
K J Level 4
L K L
decomposition tree
(a) (b)
A
A C E F I M N
B D I
J
C G H K L O
11
(c) (d) (e)
Design Domains
• Design domains:
– behavioral: black box view
• relation between input and output given without a reference to
the implementation of these relations
– e.g., describing behavior of several transistors using logical
expressions and truth tables
– Q: what is the behavioral description at transistor level?
– structural: interconnection of sub-blocks
• subcircuits used and their interconnections
• e.g., transistor-level circuit describing a NAND gate
– physical: location of structural components on the chip
• e.g., location of a NAND gate or rout of an interconnect on the
2D chip area
• Each design domain has its own hierarchy
12
6
Gajski’s Y-Chart and
Relation with Design Methodology
BEHAVIORAL DOMAIN STRUCTURAL DOMAIN
Systems
Algorithms Processors
Register transfers ALU’s, RAM, etc.
Logic Gates, flip-flops, etc.
Transfer functions Transistors
Transistor layout
Cell layout
Module layout
Floorplans
Physical partitions
PHYSICAL DOMAIN
13
A design methodology defined depending on the way the Y-chart is traversed
Transistor layout
Cell layout
Module layout
Floorplans
Physical partitions
PHYSICAL DOMAIN
14
7
Top-Down Structural Design /
Bottom-Up Layout Construction
BEHAVIORAL DOMAIN STRUCTURAL DOMAIN
Systems
Algorithms Processors
Register transfers ALU’s, RAM, etc.
Logic Gates, flip-flops, etc.
Transfer functions Transistors
Transistor layout
Cell layout
Module layout
Floorplans
Physical partitions
PHYSICAL DOMAIN 15
Design Actions
• Synthesis: increasing information about the
design by providing more detail
• Verification: checking whether a synthesis step
has left the specification intact
• Analysis: collecting information on the quality of
the design
• Optimization: increasing the quality of the design
by rearrangements in a given description
• Design Management: storage of design data,
cooperation between tools, design flow, etc.
16
8
Top-Down Synthesis
High-Level
Synthesis
Logic
Synthesis
Physical
Design (synthesis)
Fabrication and
Packaging
17
Logic Synthesis
Transistor-Level Synthesis
Physical
Design
18
9
High-Level & Logic Synthesis
a multi-stage process
Specification
Logic Extraction
module example(clk, a, b, c, d, f, g, h)
input clk, a, b, c, d, e, f;
Technology-Independent
aoutput g, h; reg g, h; Optimization
b a Technology-Dependent Mapping
h
ealways @(posedge
g = a | b;
clk) begin g1
0 G
bif (d) beging0
if (c) h = a&~h;
f
else h = b; h5 g
G
dc if (f) g = c; else a^b;
end else h3 g
b if (c) h = 1; else h ^b; H
end e d
fendmodule h1 H h
ae
c c clk
d
f
clk
19
© Prof. Ciesielski
Partitioning
Floorplanning
&
Placement
Routing
Compaction
Fabrication
20
10
Design Styles
• Can the designer control the shape of all
mask patterns?
– Yes, full-custom fabrication
– No, semi-custom fabrication, e.g. gate array,
sea of gates
• Can the designer specify the design up to
the level of individual transistors?
– Yes, full-custom design
– No, semi-custom design e.g. by means of
standard cells, module generators, etc.
21
Design Styles
Design Styles
Semi-Custom
22
11
Design Styles – Full Custom
24
12
Design Styles – Gate Arrays
25
26
13
Design Styles Trade-offs
27
Course Objectives
• Show design automation methodology
– discuss the approaches to solve the VLSI problem
• Show inside the VLSI design tools
– understand functionality and algorithms, challenges
and limitations of the tools
– understand what tasks can or can not be performed
by the CAD tools
– understand why some problems can only be
approximately solved after long computations while
others are exactly solved in short time
• Provide introduction for students interested to
specialize in development of these tools 28
14
Course Timeline
Preliminaries:
CMOS technology, 6 lectures, paper home works
Algorithmic graph theory,
Optimization methods
11 lectures, programming
Physical Synthesis assignment, paper home works
29
15