EEM 486 Computer Architecture How Do the Pieces Fit Together?

Application
Operating
System
Chapter 1 Introduction
Compiler Firmware
Instruction Set
Architecture
Memory Instr. Set Proc. I/O system
system

Datapath & Control

Digital Design
Circuit Design

Coordination of many levels of abstraction

Under a rapidly changing set of forces

Design, measurement, and evaluation

Where is the Market?

Embedded
1200 1122 Desktop
Servers By the architecture of a system, I mean the complete and
Millions of Computers

1000 892 detailed specification of the user interface. … As

862 Blaauw has said, “Where architecture tells what
800 happens, implementation tells how it is made to
happen.”
600 488
The Mythical Man-Month, Brooks, pg 45
400 290

200 93 114 135 129 131

3 3 4 4 5
0
1998 1999 2000 2001 2002
Instruction Set Architecture (ISA) ISA Type Sales

ISA: An abstract interface between the hardware and Other
the lowest level software of a machine that encompasses SPARC
all the information necessary to write a machine 1400
Hitachi SH
language program that will run correctly, including 1200 PowerPC
instructions, registers, memory access, I/O, and so on.

Millions of Processor
Motorola 68K
1000
“... the attributes of a [computing] system as seen by the MIPS
programmer, i.e., the conceptual structure and functional 800 IA-32
behavior, as distinct from the organization of the data flows and
controls, the logic design, and the physical implementation.” ARM
600
– Amdahl, Blaauw, and Brooks, 1964
 Enables implementations of varying cost and performance to run 400
identical software
200

ABI (application binary interface): The user portion of the 0
instruction set plus the operating system interfaces used 1998 1999 2000 2001 2002
by application programmers.

Moore’s Law Processor Performance Increase

In 1965, Gordon Moore predicted that the number of
transistors that can be integrated on a die would double 10000
every 18 to 24 months (i.e., grow exponentially with Intel Pentium 4/3000
DEC Alpha 21264A/667
Performance (SPEC Int)

time). DEC Alpha 21264/600 Intel Xeon/2000

1000
DEC Alpha 5/500
DEC Alpha 4/266 DEC Alpha 5/300

Amazingly visionary – million transistor/chip barrier was 100 DEC AXP/500 IBM POWER 100
crossed in the 1980’s.
 2300 transistors, 1 MHz clock (Intel 4004) - 1971 HP 9000/750
10
 16 Million transistors (Ultra Sparc III) IBM RS6000
MIPS M2000
 42 Million transistors, 2 GHz clock (Intel Xeon) – 2001 SUN-4/260 MIPS M/120
 55 Million transistors, 3 GHz, 130nm technology, 250mm2 die 1
1987 1989 1991 1993 1995 1997 1999 2001 2003
(Intel Pentium 4) - 2004
Year
 140 Million transistor (HP PA-8500)
DRAM Capacity Growth Impacts of Advancing Technology

1000000 512M
Processor

256M  logic capacity: increases about 30% per year
128M
100000 64M  performance: 2x every 1.5 years
16M ClockCycle = 1/ClockRate
Kbit capacity

10000 4M 500 MHz ClockRate = 2 nsec ClockCycle

1M 1 GHz ClockRate = 1 nsec ClockCycle
1000 4 GHz ClockRate = 250 psec ClockCycle
256K
64K
Memory
100
16K
 DRAM capacity: 4x every 3 years, now 2x every 2 years
 memory speed: 1.5x every 10 years
10
1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002  cost per bit: decreases about 25% per year
Year of introduction

Disk
 capacity: increases about 60% per year

Example Machine Organization The Big Picture 1

Workstation design target The five classic components of a computer are Input,
 25% of cost on processor Output, Memory, Datapath, and Control. The last two
 25% of cost on memory (minimum memory size) sometimes combined and called the processor.
 Rest on I/O devices, power supplies, box

Computer
CPU Memory Devices

Control Input

Datapath Output
Five Components

That is, any computer, no matter how primitive or
advance, can be divided into five parts:
 The input devices bring the data from the outside world into the
computer.
 These data are kept in the computer’s memory.
 The datapath requests and processes them.
 The operation of the datapath is controlled by the computer’s
controller.
 Getting the data back to the outside world is the job of the output
devices.

The most COMMON way to connect these 5 components

together is to use a network of busses.