March 23, 2009
The Aart of Analogy Revisited
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.
Peggy Aycinena - Contributing Editor


by Peggy Aycinena - Contributing Editor
Posted anew every four weeks or so, the EDA WEEKLY delivers to its readers information concerning the latest happenings in the EDA industry, covering vendors, products, finances and new developments. Frequently, feature articles on selected public or private EDA companies are presented. Brought to you by EDACafe.com. If we miss a story or subject that you feel deserves to be included, or you just want to suggest a future topic, please contact us! Questions? Feedback? Click here. Thank you!


There’s a loose analogy here. How can one woman have 9 babies in one month? It’s not possible. It’s one baby in 9 months with one woman, or 9 babies in 9 months with 9 women. Some applications are limited by nature in the amount of parallel processing that can occur.


Similarly for EDA software, like any other utilizer of compute power, we know some problems can be split up more easily than others. Three years ago, we started looking at the problem under the banner of Teracomputing. We said, let’s look at the state of the art of parallel programming. Now today, I can say that every one of our application tools uses multicore hardware.


It’s still true, however, that some problems are very, very tightly interconnected where it’s very difficult to split the application up into multiple pieces. But for things like a local DRC check, why not do these in parallel? That’s a very good use of multicore functionality.


*
Peggy – So, are you talking here about real progress in multicore, or just window dressing?


*
Aart – No, it’s not just window dressing. This goes at the very architecture of the program. But it is true, there’s quite a spread in terms of the results and what you can get. For the most inter-twined programs, you can get 2-to-5x faster. But for programs that can be highly partitioned – the things you would have in OPC or things related to very large amounts of data being checked locally – there, we can literally split the program into hundreds of simultaneous processes. And often, not just on one computer, but on many computers. The largest compute farms in the world today are being used for simulation and OPC.


*
Peggy – But is that the same problem? Partitioning programs for compute farms versus multithreading for multiple processors on a single chip?


*
Aart – Well, there’s link between those things. They have somewhat different characteristics, but simply put – it’s a balance between computation and shoving data around. If you have to go to a memory chip, it’s slower. If you have to go off to a completely different computer to process a thread, it’s really slowed.


My analogy for this is about doing the dishes. Every so often, my kids have 10 friends over to the house. I tell them it would be efficient for the 10 of them to do the dishes together, the process would go 10 times faster. But, with 10 kids in the kitchen, the reality is that you can forget about any form of efficiency. The dishes are just colliding. Nonetheless, it’s still better than just one person cleaning up for all 10.


You can see the issue – it’s one of too many little programs versus data space allocation.


*
Peggy – How do you customize your offerings for your customers, whether for their multicore compute platforms, or their compute farms, or whatever.


*
Aart – Ah. This is a good question! Increasingly over the last few years, we have been making quite an effort to help our customers customize in a way that makes sense. Back 5-to-6 years ago, when AMD came out with their first multicore processor, initially people jumped on them because they actually made our products run faster, which was a driver for us for selling those solutions.


But at the same time, people were saying to us, you also need to make the software work on our slower compute platforms. So, the interactions we began to have with the different people who provide the hardware for our customers was very important. But also, the good news – we’re now providing the tools to those people for designing their multicore processor products.


So, we’ve had good insights into all aspects of the problem. It’s really an ecosystem where we all have to dance together. If somebody is going to be wearing shoes that are too big, they’ll end up stepping on everyone else’s toes.


*
Peggy – So, how are your tools helping to promote multicore technology?


*
Aart – Actually, I was personally involved in that, because I got worried that our teams had so many different requests from our customers about all of this: “Help me re-architect your products, so it can match our multicore hardware.”


Clearly, we needed to coordinate our internal efforts. It turns out that by putting together a task force on terascale computing, we created a center of gravity at Synopsys on best-in-class tools. We could look at a [range of options] and say, that worked well, or that didn’t work for the following reasons.


I was able to just stand on the sidelines and receive [regular updates] on how things progressed. The task force allowed us to formulate our thinking around the general topic of multicore, and has moved our entire team forward.


*
Peggy – Who’s on your task force?


*
Aart – It’s just an internal group, although the first thing we did was to interact explicitly, and very informally, with all of the leading developers of multicore technology – our customers – plus a number of different universities. Mostly, we have incredibly deep technology knowledge at Synopsys, so that was our principle source of inspiration.


*
Peggy – Speaking of Synopsys, how is it the company has maintained its stock valuation over time? It’s been quite interesting to watch over the last year or two.


« Previous Page 1 | 2 | 3 | 4 | 5 | 6  Next Page »


You can find the full EDACafe event calendar here.


To read more news, click here.



-- Peggy Aycinena, EDACafe.com Contributing Editor.


Rating:


Review Article Be the first to review this article
Aldec

Featured Video
Jobs
Embedded Firmware Engineer for FlexRadio at Austin, Texas
Electronics Component Engineer for Lockheed Martin at Huntsville, Alabama
Senior RF Design Engineer for Lockheed Martin at Chelmsford, Massachusetts
Hardware Design Engineer for FlexRadio at Austin, Texas
Upcoming Events
Embedded Systems Conference (ESC) Silicon Valley at San Jose Convention Center San Jose CA - Aug 27 - 29, 2019
MEMS & Imaging Sensors Summit at World Trade Center Grenoble France - Sep 25 - 27, 2019
Embedded Systems Week (ESWEEK) at New York City NY - Oct 13 - 18, 2019



Internet Business Systems © 2019 Internet Business Systems, Inc.
25 North 14th Steet, Suite 710, San Jose, CA 95112
+1 (408) 882-6554 — Contact Us, or visit our other sites:
AECCafe - Architectural Design and Engineering TechJobsCafe - Technical Jobs and Resumes GISCafe - Geographical Information Services  MCADCafe - Mechanical Design and Engineering ShareCG - Share Computer Graphic (CG) Animation, 3D Art and 3D Models
  Privacy PolicyAdvertise