Online Education : Coursera.

org
EE2323:Microprocessors & • www.coursera.org
Computer Architecture • 123 Course offered by 60 Top Universities
• 2 Million online students
Spring 2013

Microprocessor Architectures
Number Systems
Lecture 3
February 25, 2013

2/25/2013 2

Types of Computer Architectures Embedded Processors

• Computer Architectures can be divided into • An Embedded Processors is simply a RISC
many categories µProcessor that has been embedded into a
• Reduced Instruction Set Computers (RISC) device
• Complex Instruction Set Computers (CISC) • Embedded Processors are programmable,
operate at low clock frequencies, consume less
• Type of Architectures power and have limited memories
– Von Neumann Architecture
• Embedded Processors perform both control
– Harvard Architecture
tasks and computationally intensive tasks but
• Type of Processors their performance capabilities are never better
– General Purpose Processors (GPPs) then general purpose processors (GPPs)
– Embedded Processors • An embedded Processor may exist as :
» Microcontrollers
– Microcontroller
» Digital Signal processors (DSPs)
– Digital Signal Processor (DSP)
2/25/2013 3 2/25/2013 4
Microcontrollers ( µC ) Digital Signal Processor (DSP)
• Microcontrollers integrate all the necessary • Digital Signal processors are a special class of
components (control, memory, I/O) of a computer embedded processors for DSP applications such
system into one integrated circuit as audio, video codecs,
• Microcontrollers are single chip solutions for systems • DSP processors are based on Harvard
requiring low to moderate processing power architecture with separate program and data
• e.g. Intel 8051, PIC µC, ATMEGA32 etc memory and buses
• DSP processors have large register files &
special addressing mode and instructions
Microcontroller
• Devices such as mobile phones often have a
microcontroller and a DSP integrated on a single
board
• TI and Analog Devices are two main vendors of
DSPs
2/25/2013 5 2/25/2013 6

Microprocessor Performance : MIPS Fueling the Innovation : Moor’s Law

• Microprocessor performance is often measured in
100000000
million instructions per second (MIPS)
• In 1965 , Gordon Moore
,Intel cofounder said :
10000000
• No. of transistors on a
MIPS =
Processsor Clock Moore’s Law Pentium chip will become
Cycles per Instructio n (CPI) x 10 6 i80486
double every 2 years
1000000

i80386
• If every microprocessor instruction executes in one • Related trends
i80286
100000
clock cycle then – Processor
performance
• CPI = 1 10000
i8086 Twice as fast after
~18 months
• MIPS = Processor clock in MHz i8080
– Memory capacity
i4004
• MIPS figure is used to compare performance of Twice as much in
1000
<2 years
different processors for a given algorithm 1970 1975 1980 1985 1990 1995 2000

Year

2/25/2013 7 2/25/2013 8
Evolution of Intel Architectures Evolution of Intel Architecture
• A common way to categorize processors is the Intel 4004 (1971): First Intel Processor
number of bits their ALU can process • 4 bit microprocessor. 4004 was used in
• 4KB main memory , 45 instructions Calculators.
• An n-bit µP mean, its ALU , its register file and • 2300 transistors , Clock : 108 KHz
most of its instructions are designed to work Intel 8008: (1972)
with n-bits • 8-bit microprocessor. Intel 8008 was twice as
powerful as Intel 4004
• Intel microprocessors can be categorized as : • 16KB main memory , 48 instructions.
• Clock : 500-800 KHz
– 4-bit Processors Intel 8080: (1974)
Altair8800 Computer
– 8-bit Processors • 8-bit microprocessor. was based on Intel 8080
• 64KB main memory, 4500 Transistors
– 16-bit Processors • Clock : 2 MHz
– 32-bit Processors Intel 8086: (1978), Intel 8088 (1979)
Intel 8088 was selected in
• Evolution of Intel processors started in 1971 with • 16-bit microprocessors
power IBM PC in 1981
• 1MB main memory , Clock : 5 MHz
the 4-bit processor 4004
• 4/6-byte instruction cache.
2/25/2013 9 • Segment Registers and memory Segmentation 10

Evolution of Intel Architecture Evolution of Intel Processors

Intel 80286: (1982) Intel Pentium: (1993)
• 16-bit microprocessor • 32-bit microprocessor
• 16MB main memory, 1,34,000 transistors, Clock : 6 MHz • Super-scaler architecture
• 80286 supported multitasking • 4GB main memory, Clock : 60, 66, 90MHz
• 80286 modes : real mode , protected mode • 16KB L1 cache (split instruction/data Cache: 8KB each).
Intel 80386: (1985) • Introduction of MMX instructions
• 32-bit microprocessor. Intel Pentium I , II , III
• 4GB main memory, 2,75,000 transistors, Clock : 12-33MHz. Intel Pentium IV: (2000)
• Variations: DX, EX, SL, SLC (cache) and SX. • 32-bit microprocessor
Intel 80486: (1989) Intel 32-bit Processors are • Clock : 1.5 GHz, 42 million transistors
• 32-bit microprocessor collectively known as • Hyper-Threading technology
• 4GB main memory, Clock : 20-50MHz. IA-32 Architectures • 1MB/512KB/256KB L2 cache.
• Introduction of floating point math coprocessor and an 8K byte cache on • Introduction of Specialized streaming (SSE , SSE 2 instructions)
one package.
Itanium Processors: (2001)
• Variations: SX, DX2, DX4 (DX2 : Double clocked version)
• 64-bit microprocessor (IA-64 Architecture)
• Itanium processors are used in server machines
2/25/2013 11 2/25/2013 12
Evolution of Intel Architecture Evolution of Intel Processors

Transistor Clock Number of transistors Minimum transistor sizes (µm)

Data Address 7
Intel µP Year Count Speed
100,000,000
P III 8080
Bus Bus 10,000,000
Pentium
P4 6
1,000,000 80386 P II 5
100,000 8088 80486
4
80286
4004 1971 2 250 0.108 MHz 4-bit 12-bit 10,000
8080
3
8088
1,000
8080 1974 6 000 2 MHz 8-bit 16-bit 100
2 80386
Pentium
8085 1976 6 500 3-8 MHz 8-bit 16-bit 10 1 80286
80486
P II P III P 4
1 0
8086 1978 29 000 5 MHz 16-bit 20-bit 1974 1979 1982 1985 1989 1993 1997 1999 2000 1974 1979 1982 1985 1989 1993 1997 1999 2000

8088 1979 29 000 5 MHz 8-bit 20-bit

Clock frequencies (MHz) MIPS
80286 1982 134 000 8 MHz 16-bit 24-bit 10000 10000
P4
80386DX 1985 275 000 16 MHz 32-bit 32-bit 1000 1000 P III P4
Pentium
80486DX 1989 1.2 million 25 MHz 32-bit 32-bit Pentium
P II
P III 100 P II
100
Pentium 1993 3.1 million 60 MHz 32-bit 32-bit 80386 10 80386 80486
Pentium Pro1995 5.5 million 150 MHz 32-bit 36-bit 10
8080
8088 80486
1 8080 8088
80286
Pentium II 1997 7.5 million 233 MHz 32-bit -- 1
80286
0.1
Pentium III 1999 28.1 million 450 MHz 32-bit -- 1974 1979 1982 1985 1989 1993 1997 1999 2000 1974 1979 1982 1985 1989 1993 1997 1999 2000

2/25/2013 13 2/25/2013 14

Number Systems Decimal Number System

• Number Systems can be divided into : • Base (radix) = 10
–Decimal Number System – 10 digits { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9 }
Octal Number 2 1 0 -1 -2
System is not • Digit Position
–Binary Number System common in
Computer
– Integer & fraction 4 1 5 4 4
–Octal Number System Architectures • Digit Weight 100 10 1 0.1 0.01
–Hexadecimal Number System – Weight = (Base) Position

• Every number system has different Base (Radix) – Weight = (10) Position
• Generalized representation of Number Systems • Magnitude 400 10 5 0.4 0.04

– Sum of “Digit x Weight” 2 1 0 -1

d2*B +d1*B +d0*B +d-1*B +d-2*B
-2

d2*B2+d1*B1+d0*B0 + d-1*B-1+d-2*B-2
Integer Part Fractional Part
• Formal Notation (415.44)10

2/25/2013 15 2/25/2013 16
Binary Number System Hexadecimal Number System
• Base (radix) = 2 • Base (radix) = 16
– 2 digits { 0, 1 }, also called “bits” – 16 digits { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, A, B, C, D, E, F }
• Weights 4 2 1 1/2 1/4 • Weights
Position 256 16 1 1/16 1/256
– Weight = (Base) 1 0 1 0 1 – Weight = (Base) Position
– Weight = (2) Position – Weight = (16) Position 1 E 5 7 A
2 1 0 -1 -2
• Magnitude • Magnitude 2 1 0 -1 -2
1 *2 +0 *2 +1 *2 +0 *2 +1 *2-2
2 1 0 -1

– Sum of “Bit x Weight” – Sum of “Digit x Weight” 1 *162+14 *161+5 *160+7 *16-1+10 *16-2
=(5.25)10
• Formal Notation =(485.4765625)10
(101.01)2
• Groups of bits • Formal Notation (1E5.7A)16
– 4 bits = Nibble {1 0 1 1}
– 8 bits = Byte {1 1 0 0 0 1 0 1}
– 16 bits = Word {1 1 0 0 0 1 0 1 1 1 0 1 0 1 0 0}
2/25/2013 17 2/25/2013 18

Octal Number System Number Systems

• Base (radix) = 8 • Decimal Numbers
Decimal Binary Octal Hex
– 8 digits { 0, 1, 2, 3, 4, 5, 6, 7 } – 10 Digits 00 0000 00 0
• Binary Numbers 01 0001 01 1
• Weights 64 8 1 1/8 1/64 02 0010 02 2
Position
– 2 Digits
– Weight = (Base) 03 0011 03 3
5 1 2 7 4 • Octal Numbers 04 0100 04 4
– Weight = (8) Position 2 1 0 -1 -2 – 8 Digits 05 0101 05 5
• Hexadecimal Numbers 06 0110 06 6
• Magnitude
5 *82+1 *81+2 *80+7 *8-1+4 *8-2 07 0111 07 7
– Sum of “Digit x Weight” – 16 Digits 08 1000 10 8
=(330.9375) 10 09 1001 11 9
10 1010 12 A
• Formal Notation (512.74)8 11 1011 13 B
12 1100 14 C
13 1101 15 D
14 1110 16 E
15 1111 17 F
2/25/2013 19 2/25/2013 20
 Number Base Conversion
Number Base Conversions
(Decimal to Binary, Octal, Hexadecimal)
Sum of • For Integer part
(Digit x Weight)
– Divide the number by the ‘Base’
Octal
(Base 8) – Take the remainder as a Coefficient
– Take the quotient and Repeat division till
Sum of
quotient is ZERO or less than base
(Digit x Weight)
?
Decimal Binary
(Base 10) (Base 2) • For Fractional part
– Multiply the number by ‘Base’
– Take integer part of result as coefficient
– Repeat the process with fractional part of result
Hexadecimal
till fraction is ZERO
(Base 16)
Sum of
2/25/2013 21 2/25/2013 22
(Digit x Weight)

Decimal to Binary Conversion : Decimal to Octal Conversion :

Example Example
Example (Integer) : (13)10 Example (Integer) : (175)10
Quotient Remainder Coefficient Quotient Remainder Coefficient

13/ 2 = 6 1 a0 = 1 175 / 8 = 21 7 a0 = 7
6 /2= 3 0 a1 = 0 21 / 8 = 2 5 a1 = 5
3 /2= 1 1 a2 = 1 2 /8= 0 2 a2 = 2
1 /2= 0 1 a3 = 1
Answer: (175)10 = (a2 a1 a0)8 = (257)8
Answer: (13)10 = (a3 a2 a1 a0)2 = (1101)2 Example (Fractional) : (0.3125)10
Example (Fractional) : ( 0.125 )10 Integer Fraction Coefficient
0.3125 * 8 = 2 . 5 a-1 = 2
Answer: = (0.001)2 0.5 *8= 4 . 0 a-2 = 4
Answer: (0.3125)10 = (0.a-1 a-2 a-3)8 = (0.24)8
2/25/2013 23 2/25/2013 24
Binary to Octal Conversion Binary − Hexadecimal Conversion
• Each group of 3 binary bits • 16 = 24 Hex Binary
represent an octal digit Octal Binary 0 0000
• Each group of 4 bits represents
• Starting from fractional point 0 000 a hexadecimal digit
1 0001
2 0010
combine binary bits in group of 3 1 001 3 0011
2 010 Assume Zeros 4 0100
Assume Zeros Example:
Example: 5 0101
3 011 6 0110
( 1 0 1 1 0 . 0 1 )2 ( 1 0 1 1 0 . 0 1 )2 7 0111
4 100 8 1000
5 101 9 1001
A 1010
6 110 B 1011
( 2 6 . 2 )8 7 111
(1 6 . 4 )16 C 1100
D 1101
E 1110
Works both ways (Binary to Hex & F 1111
Works both ways (Binary to Octal & Octal to Binary) Hex to Binary)
2/25/2013 25 2/25/2013 26

Octal to Hexadecimal Conversion Practice Questions

• Convert to Binary as an intermediate step i.e. Convert • Convert the following from Decimals to Binary
from Octal to Binary to Hexadecimal » 259 , 245.50 , 89. 35
Example: ( 2 6 . 2 )8 • Convert the following from Decimals to Octal
» 125 , 36.25 , 89. 45
Assume Zeros Assume Zeros
• Convert the following from Binary to Hexadecimal
» 101100.01010101 , 1011011.0101
( 0 1 0 1 1 0 . 0 1 0 )2
• Convert the following from Octal to hexadecimal
» 357 , 254.53 , 707. 135

(1 6 . 4 )16
Works both ways (Octal to Hex & Hex to Octal)
2/25/2013 27 2/25/2013 28
1’s Complement 2’s Complements
• Complements are used by Computers for • 2’s Complement of a number is computed by :
simplification of Subtraction operation – Take 1’s complement and add 1 to LSB OR
• 1’s Complement of a number is obtained by – Toggle all bits to the left of the first ‘1’ from the
replacing right
– All ‘0’s by ‘1’s
Example:
– All ‘1’s by ‘0’s
Number:
Example : (10110000)2 10110000 10110000
1’s Comp.:
1’s Complement = (01001111)2 01001111
+ 1
• 1’s Complement is also used in logic operations
Complement of Complement of a Number is 01010000 01010000
original number
2/25/2013 29 2/25/2013 30

Signed Numbers :
Signed Numbers Representation 2’s Complement Representation
• Computers Represent Information in ‘0’s and ‘1’s • Positive numbers are represented in Binary with sign
bit as ‘0’
• For signed numbers sign information has to be
saved in 0’s and 1’s 0 Magnitude (Binary)
• In computers, signed Numbers are represented in
–Complement Form (2’s Complement) • Negative numbers are represented in 2’s Complement
form
1 Code (2’s Comp.)
• All signed number representations use the MSB to
represent the sign of number: Example
–‘0’
Positive Number (+3)10  (0 011)2
–‘1’
Negative Number (−3)10  (1 101)2

2/25/2013 31 2/25/2013 32