February 23, 2009
Bee in the Bonnet, Bug on the Loose
Please note that contributed articles, blog entries, and comments posted on EDACafe.com are the views and opinion of the author and do not necessarily represent the views and opinions of the management and staff of Internet Business Systems and its subsidiary web-sites.
Teddy B.E.A.R.S. picnic …
U.C. Berkeley’s EECS Annual Research Symposium (B.E.A.R.S.) happens every February and is actually just a big advertisement for the Engineering School. It’s an opportunity for 10 or 12 faculty members to stand up in front of an audience in the Bechtel Engineering Center on campus and brag on their recent work, their brilliant successes, and various blue-sky ideas that might pan out in the future. The symposium is a not-so-subtle plea for those in the audience – corporate types with access to R&D dollars – to ante up some money to support the Department, the professors, and their students, and not necessarily in that order.
This is the fifth year in a row I’ve attended B.E.A.R.S. and, once again, I came away with some new ideas and some old prejudices. It turns out, as an aging taxpayer living here in financially troubled California, I’ve grown a bit tired of the hubris and arrogance of the academics on the stage at my own alma mater.
These guys stand up there (and trust me, there are virtually never any women professors presenting at this dog & pony show) and crow about their projects, their labs, their startups, and the prodigy they’ve mentored who now drive/own innovation in industry. And, yes those bragging rights are well deserved.
But, as much as these profs want to drive innovation, they also don’t want to live without their State-funded safety net. They’re happy to step out of academia for a year or two, to act as CTO, or chief scientist, or founder of some zippy start-up, but as soon as the clock strikes midnight on those 2 years – they’re right back in the arms of the university, teaching a few classes here and there, and looking forward to retiring on their State-funded pensions and healthcare.
Oh, well. If you’re somebody who wasn’t smart enough to sign up for lifetime benefits courtesy of the civil service, stop whining. It’s too late now. Go out and create something innovative instead. Start a company. Prevail. Win. It’s so much more impressive when it’s done without a safety net.
Meanwhile, you should really plan to attend B.E.A.R.S. every year going forward. Lots of cool engineering, free breakfast and free lunch. Doesn’t get much better than that. Just ask
Ted Vucurevich. Even he was at B.E.A.R.S. on the Cal campus on February 12th. Nice to see you, Ted!
When I was little, I knew kids who liked to fry bugs under a magnifying glass. I was reminded of same when I saw the B.E.A.R.S. presentation by
Prof. Kristofer Pister with footage of a fly-by-wire moth that his students rigged up using a chip-in-bug-brain device that allowed them to tell the moth to turn right or left in flight.
Ironically, two days before, researchers from the
University of Arizona demo’d the exact same type of work to an SRO crowd at
ISSCC in San Francisco. The UofA configuration also consisted of an invasive device, but theirs was attached via harness to the belly of a big ol’ beetle, allowing them also to make their bug-bot turn right or left in flight. Meanwhile,
IEEE Spectrum is currently bragging on similar work from DARPA.
Query: If there are Three Laws of Robotics, are there also Three Laws of Bug-botics?
Lest Stanford feel slighted …
For those among us who receive the Annual Report for
Stanford’s School of Engineering, I would recommend turning to page 21. There, traveling incognito in a photo-sans-caption, is Stanford’s Favorite [EDA] Son,
Dr. Walden Rhines.
Probably just capricious that listed in the column to the left of the photo, under the category of “Other” Endowed Faculty Funds, you'll find a mention of the
Cadence Design Systems Directorship in the Computer Systems Laboratory. And here I thought Cadence was Blue & Gold through and through. (Dr. S-V, what’s that all about?)
Meanwhile, it’s important to note the
Dr. Rhines was named a
Fellow of the International Engineering Consortium (IEC) at
DesignCon earlier this month. Following his award, Rhines gave
a rousing keynote, now available online and worth the half-hour spent viewing. Rhines reviews the history of EDA, and waxes poetic about reasons for hope and enthusiasm within the industry.
Professor Parallel …
Dr. Dave Patterson has been teaching at U.C. Berkeley for over 30 years, after doing his undergrad and graduate work at UCLA. Patterson’s technical accomplishments are immense. He coined the term RISC, authored 5 books (2 of them with his fellow RISC innovator, Stanford President
John Hennesey), was President of the ACM from 2004 to 2006, is an ACM, IEEE and AAAS Fellow, a member of both the National Academy of Engineering and Science, and a walking encyclopedia w.r.t. RAID, NOW, RAD, and RAMP. Patterson also raises money for MS research by riding his bike hither and yon.
Most importantly, he’s a certified expert on everything parallel on all things multicore. On February 11th, Dr. Patterson and I spoke for an hour by phone. It was a great interview. I asked questions. He talked. I typed.
Q: What is multicore?
Dave Patterson: Anything with multiple processors on a single chip.
Q: What is multithread?
Dave Patterson: That’s more complicated. It’s about trying to get more performance out of a single core by running multiple threads on a single core.
Q: What’s parallel processing?
Dave Patterson: In general, that means running multiple programs in parallel on multiple cores. From a programmer’s perspective, you would write the program as if you thought it was running on multiple cores. The hardware is just going to run those threads on the processors assigned, but the hardware’s not smart enough to decide which thread should run on which processor. Typically, it’s best to think about these questions on two different axes. One is multicore and one is multithread.
Q: What’s the state of the art today?
Dave Patterson: Today, we have desktop microprocessors that have 2-to-8 cores on the chip. It’s speculated that going into the future, the number will double every five years, going to 16, then 32, and so on. In terms of multithreading, today you have multicore chips with 1, or 2, or 4, or 8 threads running. To understand the full capacity of a system, you multiply the two numbers together. If there are 8 cores with 8 threads, it would appear as if there were 64 threads running in parallel.
Q: Why is it so important to do this?
Dave Patterson: It makes the hardware more effective, which means you get more value out of the hardware. From a programmer’s point of view, however, I’ve got to find a way to make programs [in such a way as to capitalize on the available hardware].
Q: I went to a tutorial at ESC in Boston in 2007 where the instructor said “allegedly-multicore” systems are being shipped where all but one core is effectively turned off. Multicore is really just a lot of marketing hype. So, if you buy a laptop with an 8-core chip, the programs are only using a single core out of the 8 .
You can find the full EDACafe event calendar here.
To read more news, click here.
-- Peggy Aycinena, EDACafe.com Contributing Editor.
Be the first to review this article