Communications of the ACM
Last byte
Q&A: From Single Core to Multicore
Microsoft Technical Fellow and 2009 ACM A.M. Turing Award winner Charles P. Thacker in front of the Charles Babbage Difference Engine No. 2 at the Computer History Museum, Mountain View, CA.
Credit: Richard Morgenstein
Charles P. Thacker, a Technical Fellow at Microsoft, is the winner of the 2009 ACM A.M. Turing Award for his pioneering contributions to computer architecture and networks, as well as his current work on multicore computing. (A profile of Thacker, "Committed to Success," is on p. 22.) We spoke with him about the technological highlights of his career, beginning with his work at Xerox Palo Alto Research Center (PARC) in 1970.
Let's talk about the development of the Alto, the first computer to incorporate a bitmap display and a graphical user interface.
The Alto was actually the second machine we built at PARCthe first one was a time-sharing machine. We wanted a PDP-10 because that was the standard machine that the ARPA [Advanced Research Projects Agency] research community used, but it would have been unseemly for us to buy one because Xerox had just bought a computer company that made a competing machine.
So you decided to build one instead.
Bob Taylor had continuously told us, "Computers are for people. They're personal devices." That, coupled with the fact we were in this company that handled documents, made us think it would be a good idea to pursue the question of how to build what came to be called the paperless office.
"The Alto had the property that anything you could represent on paper, you could put on the screen. We knew that was going to be a big deal."
You're talking about Robert Taylor, who managed the Computer Systems Laboratory at PARC.
Taylor was not a technologisthe was a psychologist by training. But he was an extremely effective leader. The other thing was, he knew everyone in computing because he had run ARPA's Information Processing Techniques Office. So when he was hired to staff the lab at PARC, he knew where to go.
Did you have a sense of how revolutionary Alto was as you worked on it?
Oh, yes, we knew it was revolutionary. We built it with the very first semiconductor dynamic RAM, the Intel 1103, which was the first memory you could buy that was less than a tenth of a cent a bit. As a result, we realized we could build a display that was qualitatively better than what we had at the time. We had character generator terminals, and some of them were quite nice. But they were limited in various ways, whereas the Alto had the property that anything you could represent on paper, you could put on the screen. We knew that was going to be a big deal.
You were also involved in the invention of the Ethernet.
The Ethernet grew out of the realization I had of how to provide a network for the Alto. We had been studying the ALOHA network, a radio network that was used to connect the various Hawaiian Islands. The limitation was that when a transmitter started to transmit, it could no longer receive anything. One night I was lying in bed thinking about the problem when I had this sudden realization that if you used a more limited media, say, the coaxial cables used in cable television, the transmitter could not only hear what it transmitted, it could also tell whether what it thought it put on the wire was the same as what actually got put on the wire.
So if another transmitter was interfering, it could drop back and retransmit later.
That idea was refined by Bob Metcalfe and Dave Boggs into what we knew as the Ethernet. Of course, the Ethernet in those days was quite different than it is today.
You joined Microsoft in 1997 to help establish the company's research lab in Cambridge, England, and were later involved in the development of the tablet PC, a subject that's much in the news of late.
The line of thinking about tablets actually started at DEC [Digital Equipment Corporation]. We built a tablet called Lectrice back in the early 1990s, primarily as an electronic book reader. When I returned to the U.S. from my two-year assignment in Cambridge, I was working with a group in Redmond that was trying to build an electronic book reader. That didn't work out too well, but it evolved into the idea of building a tablet PC. Of course the view there was it would be great to have a device that didn't require a keyboard.
What do you make of the persistence of the keyboard in spite of the alternatives that now exist?
Typing is so much faster than virtually any other way of entering information into a computer, so I don't expect that to change. There's only one thing that can be better, and that's to use a different set of musclesthe tablet allows you to do that. You're holding a stylus and writing or drawing with it, and the interaction can be faster.
More recently, you've been working on multicore systems.
I've been using field-programmable gate arrays (FPGAs) to explore multicore architectures. For a long time, it was impossible for academic researchersor even people working in industrial labsto design their own chips. It's now possible to build a nontrivial multicore computer, with something on the order of 15 cores, on a single FPGA on a board that's available for $750.
What does the future hold for you in terms of research? Are you tempted to go back and continue working on the tablet PC?
If I have a good idea I might go back to it. But right now I'm quite happy with what I'm doing, and there's a considerable amount of work to do in this area. So I think I'm set for the next few years.
©2010 ACM 0001-0782/10/0700 $10.00
Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.
The Digital Library is published by the Association for Computing Machinery. Copyright © 2010 ACM, Inc.
No entries found