News & Analysis
Interview with Chuck Thacker, PC pioneer
Rick Merritt
3/9/2010 4:19 PM EST
Charles Thacker won the 2009 A.M. Turing Award for his work in the early 1970's on the Xerox PARC Alto, a forerunner of the Apple Macintosh and IBM PC. Thacker, now a Microsoft researcher, is testing out parallel programming concepts on a multi-core FPGA development system he designed.
 |
| Chuck Thacker Technical Fellow, Microsoft |
Thacker was cited for his contributions to Ethernet as well as his work on an early multiprocessor workstation and the prototype for the tablet PC. The Turing Award is considered akin to a Nobel Prize in computing, and comes with a $250,000 prize, with financial support provided by Intel Corp. and Google Inc.
A fellow Microsoft researcher and Turing Award winner Butler Lampson described Thacker as "an engineer's engineer" in a nominating letter. "His skills span the full range, from analog-circuit and power-supply design through logic design, processor and network architecture, system software, languages, and applications as varied as CAD and electronic books, all the way to user-interface design," wrote Lampson.
Microsoft posted online a feature story on Thacker's career. We talked with him briefly from his office at Microsoft Research in Mountain View.
EE Times: What was it like working on the Alto at Xerox PARC?
Thacker: That period was the most intensely creative in my life. A lot of things came together--the entire system of the Alto computer, the Ethernet network, the laser printer, the file systems and more--all that was developed over a five-year period.
EET: Why wasn't Xerox able to commercialize the Alto?
Thacker: It was before its time. Xerox made a tremendous amount of money based on the work at PARC, primarily based on the laser printer which was worth several billion dollars to the company.
It took large scale integration for computing to take off, and we didn't have that, so the things we built were not economically viable. We had Altos in the Carter White House, but they were simply too expensive for ordinary office workers at about $12,000—which was a lot of money then.
At that time, DRAM had just fallen below a tenth of a cent per bit. Every pixel in the Alto had a tenth of a cent of memory behind it, and we had a lots of pixels. The rest was built out of several printed circuit boards. It took another decade before Intel produced a processor on a chip.
|
|
|
|
|
Mapou
3/9/2010 6:36 PM EST
Excellent interview, Rick. Thacker's response to your last question is what caught my attention:
-
"You have to have parallel computers before [you] can figure out ways to program them, but you have to have parallel programs before you can build systems that can run them well."
-
I doubt very much that this is a chicken and egg problem for the industry. It's really a case of last century's computing paradigms getting in the way of finding a solution. You don't need parallel programs to design a good parallel processor. You need a good parallel software model. The industry does not have one and, judging by what's being done at various labs around the world (including Microsoft Research), it's not going to have one anytime soon.
-
The industry's chosen approach to parallelism is multithreading, an ancient programming method that was not originally devised with true parallelism in mind. Multithreading is woefully inadequate and wrong-headed, as everybody in this business should have realized by now. If you don't believe me, go ask Prof. Edward Lee at UC Berkeley. He knows.
-
Neither BEE-3 nor Beehive is going to help solve the problem, sorry. What is really amazing in all of this is that the solution to the parallel computing crisis is not rocket science and has been pretty much staring everybody in the face for decades.
-
My advice to Mr. Thacker and to the folks at Microsoft Research is to Google "How to Solve the Parallel Programming Crisis". The solution is just a few mouse clicks away, folks. But you will need to get rid of your Turing Machine blinders in order to see it.
Sign in to Reply
Mapou
3/9/2010 6:45 PM EST
The [you] in my comment above is a mistake. I inadvertently discovered that putting brackets around a word turns it into a link. Sorry.
Sign in to Reply