Information Technology InfrastructureHardware

Hardware: The first building blockHardware: The first building block

“Hardware: The parts of a computer you can kick”-- Source unknown

“Computers in the future may weigh no more than 1.5 tons” --Popular Mechanics, forecasting the relentless march of science, 1949

“But what ... is it good for?” -- Engineer at the Advanced Computing Systems Division of IBM, 1968, commenting on the microchip.

“There is no reason anyone would want a computer in their home”-- Ken Olson, president, chairman and founder of Digital Equipment Corp., 1977

The First?: ENIAC

February 15, 1946: Major General Gideon Barnes pushes a button in Philadelphia and changes the world.

Electronic Numerical Integrator and Computer represents the dawn of the Information Age

Mauchly and Eckert at UPenn

ENIAC• Size: 30’ by 50’• Weight: 30 tons• 17,468 vacuum tubes• Needed six technicians

in each shift• 1000 instructions per

second• First bug was a live

one that got grilled• Cost: $486,800 ($10

million present value)

Fixing a problem

ENIAC

“A new epoch in the history of human thought began last night...”

• Philadelphia Enquirer, Feb. 16, 1946

Story ran deep in the paper next to “Judge Frees 5 in Liquor Graft”

"I have always taken the position that there is enough credit for everyone in the invention and development of the electronic computer" - John Vincent Atanasoff

The Atanasoff Controversy

Legacy of ENIAC: UNIVAC

First commercially available computer

Univac used in elections

• Manufactured by Remington Rand• Used to predict 1952 US Presidential

elections• Adlai Stevenson expected to win• UNIVAC posted 100:1 odds that

opponent would win• Cronkite did not report the results

because CBS didn’t believe them• Eisenhower won in a landslide

Computer Generations

Four major generations Each distinguished by different base

technology Each generation significantly improved

computational power while lowering costs Cost of 100,000 calculations

1950s: several dollars 1980s: $.025 1995: $.00004

Computer Generations

First Generation (1946-1956)Based on vacuum tube technologiesHuge tubes that burnt out quicklyMain memory 2000 bytesRotating drums used for hard disk and

punch cards used for external storageTypically used for limited scientific and

engineering work

Second Generation (1957-1963) Based on transistor technology Smaller than tubes, generated less

heat Main memory reached 32 KB Speeds of up to 300,000 instructions

per second Used for science, engineering and

some business tasks (payroll and billing)

Computer Generations

Third Generation (1964-1979) Based on integrated circuits technology Made by printing hundreds (later,

thousands) of transistors on a silicon chip Known as semiconductors RAM expanded to 2MB Speeds of upto 5 MIPs Introduced software that could be used

without extensive technical training

Computer Generations

Fourth Generation (1980-present) Based on VLSI (very large-scale integrated

circuits) technology Packs tens of millions of transistors on a

single circuit Memory, logic, and control on a single chip

– hence the term, microprocessor Allowed the development of smaller

machines

Computer Generations

Power, Cost and Moore’s Law 1965: Gordon Moore of Fairchild

Semiconductors predicted that the number of transistors would double every 24 months…

This has held for nearly 30 years Intel plans to unveil a one-billion transistor

chip capable of 100,000 MIPs in 2011 Check site below for more information

http://www.intel.com/labs/eml/index.htm

What is a Computer System?

Central Processing Unit(ALU + CU)

Primary Storage

Input Devices

Output Devices

SecondaryStorage

CommunicationDevices

buses

The CPU The Central Processing Unit (CPU)

Manipulation of numbers, letters, symbols

Controls other parts of the computer system

Consists of • Arithmetic Logic Unit (ALU)

• Logical and arithmetic operations• Control Unit (CU)

• Coordinates and controls other parts of system

• Reads programs and directs other parts to performs tasks requested by program (machine cycle)

CPU- How Does it Work? Control Unit – initiates fetch and execute

cycles Code cache – very fast memory on CPU

chip RAM copies instructions here for fast

retrieval Data cache fast access to small amounts of

data Instruction location counter – points to next

instruction Instruction decoder – analyzes what each

instruction means Integer and floating point unit – does math ALU – does logical comparisons

Primary Storage

Sometimes referred to as primary memory or main memory

Three functions stores all or part of the program being used by

the CPU stores the operating system programs that

manage the computer store the data needed for the program being run

Random Access Volatile

What is cache?

Moving data between RAM and CPU can take several clock cycles

To do it in a single cycle needs high speed memory (expensive)

Caches are small holding areas on the chip using high speed memory.

Chip designer add cache (called L1 cache) on the chip

Manufacturers sometimes add L2 cache which may or may not be on the chip.

Linking the CPU, Primary Storage, and other devices

Three kinds of busesData bus (moves data to and from RAM)Address bus (signals for locating a specific

address in RAM)Control bus (signals to specify read/write

operations for RAM and peripheral devices)

Processing Speed

Determined in part by Word length (number of bits than can processed at one time by

the machine) Cycle speed (measured in MHz – internal beat set by control

unit) Data bus width (number of bits that can be moved at the same

time) Computation on more bits at a time Cache memory Floating Point calculations can be performed on hardware Number of transistors

Pentium 4 has about 55 million on a single chip Parallel processing

Categories of Computers

Mainframes Largest of the

computer types Massive memory Rapid processing

power Business,

science, engineering applications

Demise greatly exaggerated

Categories of Computers

Minicomputers Mid-range Originally DEC

aimed at getting a slice of IBM’s mainframe market (1957)

By 1969, scaled down version referred to as minicomputers

Categories of Computers

Personal Computers Sometimes called

a microcomputer Local storage and

processing Workstations

Powerful math and graphics capabilities

Typical of engineering and design projects

                                           

Categories of Computers

Supercomputers Can perform billions

of calculations per second (GFLOP)

Based on parallel processing

Originally designed for military for weapon systems

Massively Parallel Computing(Fifth Generation)

Thousands of processors Work in concert Split the workload and process in

parallel

Cluster Computing

Link computers together for faster performance or more reliable use

Two typesHigh availability clustering

• Server A fails, Server B takes over without pause

Performance clustering• Servers A and B work together on single

problem• Finish more quickly than either one could

do alone

Categories of Computers

Network Computers Also called “thin clients” Minimal storage and

processing Download data and

software from central server or Internet

Eliminates need for secondary storage devices

Aimed at reducing Total Cost of Ownership

The Future of Hardware

Limited by physics and economics Physics

Transistors currently etched using ultraviolet optical lithography

Can go down to 90 nanometers (200 atoms) Below 100 nanometers – wavelengths of light too big – IBM

using X-rays; Intel using Xenon; Lucent using beams of electrons

Intel has begun shipping the 90 nm technology in 4th quarter of 2003

Economics As size decreases, cost of fabrication increases Currently, plants cost about $2.5b For <100 nanometers, typically shoots to $10b Need for “affordable scaling”

Software: The second building blockSoftware: The second building block

“My software never has bugs; it just develops random features”--Source unknown