A Generation of Computation

CS 101 (Spring 1999)
Lecture 1
A Generation of Computation

Quick picks: [ Introduction ] [ Course Info ]

Property	DigiComp (Given to me by my grandmother when I was 7 years old)	Furby (Given to my daughter by her grandmother when my daughter was 7 years old)
Cost (today's dollars)	$60.00	$35.00
Input	3 Flip/Flops	Ears (audio) Eyes (light/dark) Touch (pressure sensors on front, back) Middle ear (Gravity switch for orientation)
Output	3 binary digits (0/1)	Eyes move Ears move Two spoken langauges (Furbish and English)
Brain -- storage	3 bits	128 Megabits
Brain -- speed	1 cycle/second (.000001 MIPS)	16 MIPS
Reliability	Breaks every 10 seconds	Breaks every 5 years

The transition from digicomp to Furby shows the movement of interest from programming computers to using them.

Suppose a car experienced the same improvements over the years. Today's $20,000 car would have the following properties.

Cost	$4,000
Speed	60,000 mph
Seating	10,000 people, with room to hold every molecule in the universe
Fuel efficiency	20,000 mpg
Reliability	Breaks every 70 years

Analyze/visualize stuff
- Predict the weather (the need for a supercomputer)
- Search for oil
- Explain the universe
- MRI, radiology
Control stuff
- Automated surgery
- Robots to clean up nuclear spills
- Human heart
Organize stuff
- Legal databases
- GenBank
- Web crawlers
Compute stuff
- Trig tables
- Drug dosing
- Launch trajectories

In the beginning: Each computer unto itself
Usenet: Computers call each other up occasionally to exchange mail
Internet: Computers constantly in contact with each other
The WWW: Documents, pictures, movies, sound available. No real attempt to organize or editorialize.
Electronic commerce: Buy, sell, trade on the WWW

The web has become a reliable place to exchange information.

The amount of information posted per day in 1993 is equivalent to 20,000 pages of text.

Bottom line:

The consumer needs a computer. The commodity nature of comptuers is driving the price ever lower.
The market needs more people to program computers. The need for advanced software technologies is driving salaries ever higher.

Theory	Practice
What can be computed? What is an approach (algorithm) for computing something? Is the answer exact or approximate? How long does an algorithm take to compute an answer? Is there a best algorithm for computing something? If not, how do the various algorithms compare? How do we know the answers to the above questions are correct? (proofs) What is a good model for understanding the behavior of a program or a computer?	How should an application be organized? How should an application be programmed? How do we test an application to see if it works? How do we design software so that it works well the first time? How do we design software so that it can be extended in the future? How do we design software so that it is easy to understand? What is a good instruction set for a computer? How should the computer's hardware be organized?
There are many open problems. Theoreticians find problems of interest and try to solve them. A famous example is the P=NP question.	There are many ways to solve a given problem. A practitioner examines these and chooses one that not only works well now, but can be extended in the future.

CS 101 is	CS 101 is not
A course in computer science Meant to be fun Meant to be challenging For majors and minors in CS For anybody interested in CS A course with collaboration elements	A course only about programming Meant to be easy Meant to be frustrating Requiring prior CS knowledge Passable without effort A course that tolerates cheating

Rhythm of this course Mondays 2 PM: Lab designs due Wednesdays: Quiz in class Fridays 2 PM: Lab due 2PM	Information about this course Attendance policy for lectures and labs Grading Late policy Collaboration
The course calendar provides the details.	The web pages provide the details.