The Parallel Revolution Has Started: Are You Part of the Solution or Part of...

Google Tech Talks
December 18, 2008

ABSTRACT

This talk will explain
* Why the La-Z-Boy era of sequential programming is over
* The sorry record of prior commercial forays in parallelism
* The implications to the IT industry if the parallel revolution should fail
* The opportunities and pitfalls of this revolution
* What Berkeley is doing in general to be at the forefront of this revolution
* Roofline, A New Insightful Visual Performance Model for Multicore Computers

Speaker: David Patterson, Director of the Par Lab, U.C. Ber
David Andrew Patterson (born November 16, 1947) is an American computer pioneer and academic who has held the position of Professor of Computer Science at the University of California, Berkeley since 1977.

A native of Evergreen Park, Illinois, David Patterson attended UCLA, receiving his A.B. in 1969, M.S. in 1970 and Ph.D. (advised by David F.
Martin and Gerald Estrin) in 1976. He is one of the original innovators of the widely used Reduced Instruction Set Computer (RISC) (in collaboration with Carlo H. Sequin), Redundant Arrays of Independent Disks (RAID) (in collaboration with Randy Katz), and Network of Workstations (NOW) (in collaboration with Eric Brewer and David Culler).
Past chair of the Computer Science Department at U.C. Berkeley and the Computing Research Association, he served on the Information Technology Advisory Committee for the U.S. President (PITAC) during 200305 and was elected president of the Association for Computing Machinery (ACM) for 200406.

He co-authored five books, including two with John L. Hennessy on computer architecture: Computer Architecture: A Quantitative Approach (4 editions—latest is ISBN 0-12-370490-1) and Computer Organization and
Design: the Hardware/Software Interface (3 editions—latest is ISBN 1-55860-604-1). They have been widely used as textbooks for graduate and undergraduate courses since 1990.

His work has been recognized by about 30 awards for research, teaching, and service, including Fellow of the Association for Computing Machinery
(ACM) and the Institute of Electrical and Electronics Engineers (IEEE) as well as by election to the National Academy of Engineering and the Silicon Valley Engineering Hall of Fame. In 2005 he and Hennessy shared Japan's Computer & Communication award and, in 2006, was elected to the American Academy of Arts and Sciences and the National Academy of Sciences and received the Distinguished Service Award from the Computing Research Association. In 2007 he was named a Fellow of the Computer History Museum and a Fellow of the American Association for the Advancement of Science. In 2008, won the ACM Distinguished Service Award, the ACM-IEEE Eckert-Mauchly Award, and was recognized by the School of Engineering at UCLA for Alumni Achievement in Academia.

Since 2003 he has riden in the annual Waves to Wine charity event as part of Bike MS; he was the top fundraiser in 2006, 2007, and 2008.
David Patterson's projects in 200708 have been the RAD Lab: Reliable Adaptive Distributed systems, RAMP: Research Accelerator for Multiple Processors, and the Parallel Computing Laboratory.

Category:

People & Blogs

Tags:

• google
• techtalks
• techtalk
• engedu
• talk
• talks
• googletechtalks
• education

[caspin27]

did you even watch the lecture? The speaker posed a similar question in jest: why do I need 1000 cores to run MS Word? He answered it very well.

He also discussed the problem of consumers not buying new computers because they would not make their current apps run faster. Again, it was addressed

Watch the video you'll learn something, or just complain some more and get idiots like me to answer a question that you didn't really want answered.

[Kamikaze576]

Rate up if you are forced to watch this for ECE 222 at the University of Waterloo!

[Dhonigk]

Damn since morning i was trying to understand the roofline model but 5 minutes of listening to the section where Patterson discusses it, I understood the whole thing in a jiffy...The guy is a genius!

[Dhonigk]

Get lost noob

[gzacto]

Suggestions. Stick to your area of expertise. Starting of with "Global Warming and Evolution are controversial withing scientific circles" - ah NO, both are generally accepted within scientific circles, it is outside of science they are controversial. If you don't know that, how can I believe what follows?

Two. DEATH BY POWERPOINT. If iI have to read that much, it might as well be a PDF file.

[srhpenquin]

Yes. Turn the volume up and it should clear up any low audio problems. This is a tech solution but check your manual on volume adjustment. It is not too hard to operate.

IT Dept.

[yywill]

If we have the tech to build muti-cores cpu, why not split the cpu to many smaller cpus? Then we can connect the computers to compute coordinately. That must be faster than muti-cores CPU, because we don't need to develop new architecture for parallel. There are some matured architecture for network computer.

[yywill]

Muti-cores should not be the correct way of parallel computing. I bought a 2-cores laptop which is no faster than my 4-years-old PC (I just updated its memory ). Even my laptop is really 2 times faster than my old PC, I will never fell it. I don't think I can open firefox in 1ms by using a 1000...0-cores CPU.

OK. My point is CPU is fast enough in my daily usage, I don't want to pay for a faster CPU which I don't need.