---

The computer you buy today is much faster than the computers of five years ago; the computers you buy in ten years will be up to ten times faster than today's state of the art. Chips are getting faster and faster, how are software and hardware companies using all this speed? Spencer Michels investigates.

---

SPENCER MICHELS: Buried inside some of these computers are powerful devices--chips that can make them do amazing things. Power is what makes generation of images like this possible, or the combining of images on this screen, or the 3-D effect in this medical program on the human body. These increasingly sophisticated uses of the computer depend upon the speed of the chip, a field pioneered by Gordon Moore, a founder of the giant semiconductor maker Intel. Today, at 68, he's still involved in the business of making ever more powerful chips and ever smaller components on those chips for electricity to flow through.

GORDON MOORE, Founder, Intel Group: A line like this distributes power to all the transistors.

SPENCER MICHELS: At the dawn of the computer age, back in 1968, Moore made a wild prediction that the number of transistors or switches on a chip, the brains of a computer, would double every eighteen months or two years, making the chip faster and more powerful and cheaper.

GORDON MOORE: It was amazingly correct. We followed that line really quite precisely over that decade. And that was the origin of Moore's law. And it's been extended now. Moore's law kind of applies to anything that grows geometrically in the industry, and I'm happy to take credit for all of it.

SPENCER MICHELS: The increased power of the chip, as predicted in Moore's law, has sparked continual expansion of the computer industry. What Moore's law means is that technology is constantly changing; that new chips inspire new computers; and they, in turn, invite new software, a cycle of innovation that is driving an expanding economy and a hectic industry. Sometimes the software is so hurriedly developed it comes on the market with bugs or flaws. And always the changing technology has those who designed for it, toiling night and day, while trying to anticipate the future. For example, Steve Dauterman oversees video game production at LucasArts in Northern California.

STEVE DAUTERMAN, LucasArts Entertainment: It's scary because you can't predict what is going to happen next, so, you know, you're doing a little bit of fortune telling every time you're figuring it out. Well, we're going to release a game two years from now. What's it going to be like two years from now?

SPENCER MICHELS: It's all based on the notion that everyone needs more computer speed and power. That's a message high-tech promoters send out regularly.

ANNOUNCER: Your PC's performance may be up against a barrier you don't even see, a performance barrier that seems invisible. Graphics Blaster Extreme can deliver much higher frame rates than standard 2-D and 3-D accelerators.

SPENCER MICHELS: Whether it's a graphics blaster or a 3-D accelerator or a Pentium 2 chip, the promise of more computer speed is alluring.

MAN: A 486 machine is a good, okay machine, but it's not going to be fast enough to run all of the programs my kids want.

SPENCER MICHELS: Speed for a computer is the rate at which the chip can process the instructions that go into it, and that, according to Intel microprocessor expert Steve Smith, translates into more realistic images on your screen.

STEVE SMITH, President, Intel: What you see on your screen can go faster. A current example would be running a video clip on your PC. And if you're running it on a slower PC, you might see the frame get updated five or ten or fifteen times a second.

SPENCER MICHELS: Sort of jerky?

STEVE SMITH: And it looks jerky. And if you're running it on the latest generation processor, such as one of these Pentium 2 processors, what you can see is a full frame, smooth motion, and it's--it gives you an image that's as good as you're used to seeing on television.

SPENCER MICHELS: The speed is made possible by all those tiny on-off switches, or transistors, on the chip or microprocessor.

STEVE SMITH: The first microprocessors that were just over 25 years ago had 2,300 transistors. And the latest processors have about 7 million transistors. And in the next 10 years we're going to approach 100 million transistors on a chip that will be running in a PC that you can have in your home or office.

SPENCER MICHELS: Speed is essential for the designers and the players of video games. This game, Dark Forces, was designed at LucasArts four years ago for the then hot 486 chip made by Intel. It sold a million and a half copies, 35 to 50 dollars each.

STEVE DAUTERMAN: You've got the gun here. When you can switch weapons, break a show, you know, going to another weapon there--this was really the first generated chips that we could create these immersive 3-D worlds.

SPENCER MICHELS: But game designers and their customers are never satisfied. They want more realism on the screen. And to get it, they need faster chips. A just-released Jedi Knight plays on faster, newer computers. The main character is much more versatile.

STEVE DAUTERMAN: You can look anywhere, compared to sort of just your straight up and down before. Now, if it goes up to the edge here, you can look down and see characters actually walking around. You've got a sense of depth. We can actually go to an external shot of the character. This is the first time we're able to do this, actually go outside of it.

SPENCER MICHELS: What about the person whose parents bought a computer five or six years ago, can they run a game like this?

STEVE DAUTERMAN: It's getting harder and harder to run games like this, and, you know, what we've found is the computer game market, the people that play these games primarily really like to, you know, push the edge.

SPENCER MICHELS: But the edge is really being pushed by companies after a profit, according to Jon Forrest, a computer systems manager at the University of California at Berkeley.

JON FORREST, University of California, Berkeley: If companies aren't growing by selling new stuff, then the stockholders get upset. So there has to be this movement toward faster and faster things. I think, though, if you talk to the public, what they want is higher quality, not faster.

SPENCER MICHELS: At Adobe Systems in San Jose they play the speed game as well, and there's no question that's what sells. They design and build Photo Shop software used by millions of graphic artists to retouch photographs and combine images that appear in magazines, catalogues, and other ads. Mark Hamburg is principal scientist and architect of the program.

MARK HAMBURG, Adobe Systems: Photo Shop is incredibly widely used. It touches virtually every image that you see.

SPENCER MICHELS: Here, he shows the difference between an older, slower version of the program, where it takes several seconds to change the color and the background of this photo, and a newer version designed for faster chips.

MARK HAMBURG: This allows us to change the way Photo Shop behaves so that all you have to do is every time you pause we redraw the screen. So there's no clicking the preview button and waiting, deciding whether you liked it or not.

SPENCER MICHELS: That's important for customers like catalogue designers who work on hundreds of photos quickly. But is it of value to the society?

MARK HAMBURG: It's open to a lot of opportunities for people to do artwork that they probably wouldn't have been able to do previously, or it would have been incredibly tedious using actual paints.

SPENCER MICHELS: How much faster do you want it to go?

MARK HAMBURG: Well, so far, we haven't seen anything that's made us go, well, okay, this is fast enough, you can just stop here.

SPENCER MICHELS: For the average computer user speed may not be so important. Most use word processing programs and search the Internet, functions that don't need blinding chip speed, according to Jon Forrest at Berkeley.

JON FORREST: The folk myth is that speed is important. Twenty years ago when I started in this business, it was important, and I spent a lot of time waiting for computers. But I noticed over time I was spending less and less time waiting for computers. The myth that a computer needed to be faster and faster was starting to be less and less true.

SPENCER MICHELS: Forrest says for high-end uses like graphics and games speed counts but for others--

JON FORREST: There comes a time when things really can't go any faster because the computer, itself, is not the bottleneck.

SPENCER MICHELS: What is?

JON FORREST: Well, quite often it's people's brains or people's fingers, or it's some other aspect of a computer system, like a network, because most of what you notice as delay on the Internet is not due to the speed of the chip; it's due to the speed of the network or congestion, where an infinitely fast computer wouldn't help.

SPENCER MICHELS: But just down the hall computer graduate students, hungry for speed, dispute Forrest's assertions, saying even word processing programs are now designed for powerful computers.

STEVEN GRIBBLE, University of California, Berkeley: You need the power because you end up using the tools created by these software inventors to take advantage of faster chips to do less work.

SPENCER MICHELS: Even to write the novel?

STEVEN GRIBBLE: Even to write the novel because you're writing the novel on tools like Microsoft Word, which require this extra speed and extra processing power.

SPENCER MICHELS: The semiconductor and computer industries are banking on the increased demand for speed to create increased demand for their products. They forecast 17 percent growth in chip sales in 1998. Recently, IBM announced a technical breakthrough that will allow it to begin using copper, a better electrical conductor than aluminum, in manufacturing faster running chips. In Scotland, scientists are trying to improve the electrical connections between processors to increase speed. And Intel is bringing out a faster chip called the Merced processor for companies that need higher-end applications like special effects for movies. Even for those who don't actually need it, Intel is betting that speed will sell.

STEVE SMITH: The human being loses attention if you make the human wait for more than 1/2 second. So I know when I'm doing my work and I go click on an attachment like a word file or some sort of attachment, I'd like to see that information immediately.

SPENCER MICHELS: What computer users and engineers would like to know is whether there is a real limit on how fast chips can go. Gordon Moore has thought about that.

GORDON MOORE: Of course, we can't exceed the speed of light. That's fairly fundamental, and that is not likely, however, to be that much of a limitation. We get indirect limits, like as we make things smaller, the resistance of the little wires goes up. And we end up working against ourselves. The interconnections start being what limits the speed of the whole circuit. We're right about at that point now, but these are all things we can work on. The fundamental limits are the velocity of light and the atomic nature of matter.

SPENCER MICHELS: How does the giver of Moore's law look on the value of faster and faster chip speed?

GORDON MOORE: You can ask: Is there a need for electronics? And mankind got along for hundreds of thousands of years without electronics, but modern society certainly depends on them. And to get the maximum advantage and to get the maximum participation in modern society, I think there's a real need for continuing to improve the performance, make faster and faster and cheaper and cheaper computers, for example. Okay. Great. Sorry. No time left. Thank you all very much. I appreciate it.

REGIONS TOPICS RECENT PROGRAMS ABOUT US FEEDBACK SUBSCRIPTIONS / FEEDS:

Support the kind of journalism done by the NewsHour...Become a member of your local PBS station.
PBS Online Privacy Policy Copyright ©1996- MacNeil/Lehrer Productions. All Rights Reserved.