Chris Harwood came to Science House on Friday to deliver his second workshop on memory techniques. Chris is a memory expert who has been placed in the top 5 in the USA Memory Championships. Chris shared some of the techniques used in competitive memory competitions, such as the story method list systems and the “Roman room” method. See Chris’ notes here.

Last Friday former neurosurgeon turned entrepreneur Katrina Firlik came to Science House to give a talk titled “How to Think Like a Brain Surgeon”;. Katrina provided us with some fascinating insights into this gruelling and sometimes gory profession and talked about some of the surprises, tragedies and success stories from her career. All attendees received a signed copy of Katrina’s book “Another Day in the Frontal Lobe: A Brain Surgeon Exposes Life on the Inside”.

Katrina demonstrating how to access tumors in different parts of the brain.

US, China and UK editions of Katrina’s book.

Katrina talks about the mechanical approach to brain surgery.

More photos from the event >

A couple of weeks back we held a talk at Science House on Chip Design, led by guest speaker Lawrence Chernin. Lawrence gave a fascinating talk on the subject, starting with the basics – semiconductor materials and transistors – and building up to the remarkably complex field of chip design, where many thousands, millions or even billions of transistors are arranged into optimized layouts resulting in the chips found in our iPhones, laptops, and digital cameras.

The talk wouldn’t have been so interesting if Lawrence didn’t have such unique and exciting stories to tell about his own experiences in the field. A former astrophysicist with a PhD from Harvard University, Lawrence spent his early career researching star formation and supersonic flows. After a three year stint at Berkeley, Lawrence was drawn into the booming high tech field in nearby Silicon Valley. Below, Lawrence writes about his adventures in the field.

Lawrence explains how transistors are built

Chip Design Adventures in Silicon Valley

by Lawrence Chernin

My first job in Silicon Valley was at Hitachi Semiconductor (now part of Renesas), and I considered myself quite lucky to have been given this opportunity since almost all of my studies and academic work before that were in astronomy… As you can imagine – I’ve been asked over a hundred times, “well what does chip design have to do with astronomy?” Nothing really. By fluke I had met someone at a friend’s barbeque and he invited me to come in and talk to his boss the following week. At that time I had a position at Berkeley which ended in three months. I did have another postdoc lined up, but I was ready for a bigger change – so I thought “what the heck, let’s find out more about chips!” The next morning I did the sixty mile drive down the I-80/880 to Hitachi in San Jose.

The Hitachi interview was like nothing I ever expected. I spoke with several interviewers who asked me questions about the kind of astronomy research I had done, and then one guy recommended a few of the standard textbooks on chip design (e.g. Weste and Eshragian, see below). After two hours, the group manager came in and asked if I would come back tomorrow and sign their offer letter! Wow, that was nothing like I’d ever experienced before. Most academic job prospects had a hundred applications for each position. Also I had been exploring alternative careers for several months and even had a first round interview at McKinsey – which I badly flunked, as well as a couple of interviews with financial quant firms in which I fared no better.

So after glancing at the pretty picture in chip design textbook in one of the Silicon Valley tech bookstores the next day, I drove back down to San Jose and signed the offer letter and gave my notice at Berkeley.

The first project I had involved transistor modeling, but I was given plenty of time to read up on the subject and bring myself up to speed. There were several good textbooks on the subject, as well as internal publications. Transistor models are current – voltage equations that account for the behavior of a transistor. These equations are then inserted into large scale computer simulations which may model a specific chip functional unit like a memory, or perhaps even the entire chip. A program called “Spice” and it’s numerous variations were used for this purpose. The usual tradeoff becomes accuracy vs runtime vs circuit size, and this is the task of the engineer to determine. My task was to find the highest accuracy of the model given the data and measured parameter bin size, and also to consider the boundary conditions and the continuity of the derivatives.

My boss also paid a private consultant to come in for a week and to lecture a few of us on the latest developments. Once we were up to speed this turned into a very exciting project as we were able to improve on some of the Hitachi transistor models. Then I turned my attention to the simulation software package assessment, and became what is known in the field as a CAD (Computer Aided Design) engineer.

After the chip circuit project were completed I was moved into the rapidly growing area of digital chip design automation. This means creating lithography and etching mask patterns for the transistors and wire interconnects that cover the entire chip. A typical chip has three to six mask layers used for the creation of the transistor structure inside the silicon material and then a six or more layers of metals such as copper or aluminum wires on top of it. Mask layers are used to guide etching and deposition process on a silicon wafer.

The transistors of the chip are arrayed in basic circuit gates such as inverters, nands, latches, etc. The circuit gates need to be decomposed out of a software like description of a chip function.

The general name given to this design process is called “place and route”, because it involves placing basic circuit gates in optimal locations on the chip’s silicon surface, in order to minimize the wire connections or “routing” for the circuit. The less the wires, the less the power, the less the heat, and the faster the chip.

So the design companies would purchase large banks of servers to run the design simulations. The computational tasks are broken down to run in parallel using a techniques of threading and multiprocessing as much as possible. one large software design companies that I worked at had 5000 servers that were running simulations 24×7. Most of the time the servers were running quality and regression tests. For a run example, a large CPU chip such as one in a digital camera (typically 40 million transistors or more) could be turned from a software description to a set of mask layers to be fabricated in less than twelve hours on a 4-cpu server.

Most of my years in the industry were spent in the area of place and route software automation, but I also took on a few interesting projects like diagonal chip routing. All digital signal chips are produced in Manhattan routing, but diagonal routing provided another degree of freedom so as to enable further minimization of the total wiring length. This project was so high profile that it even received main stream media coverage, like from the Wall Street Journal. We pasted these clippings on the cubicle walls. However, in the end – after five years, this project was canceled. The main issue had to do with the complexity of manufacture and difficulty of adaptation to the new technology.

So looking back at my time in Silicon Valley, which included several very highly rated start ups, one in fact was more highly rated than Google in 2001, but unlike Google, my startup became just a short historical footnote. I rode the roller coaster of the ups and downs of the chip industry, and it was quite an adventure.

Inside an iPhone

—————-
:REFERENCE DATA:
–=————-

Large Semiconductor Chip Manufacturing Companies:
Intel, IBM, Samsung, TSMC, Amd, NEC, Toshiba, Renesas

Large Chip Design Companies, but not also manufacturers:
Qualcomm, Sun (Oracle), Apple

Large Companies in Chip Design Software:
Synopsys, Cadence, Mentor Graphics, Magma.

Philadelphia Semiconductor Index tracks 18 largest semiconductor companies: SOXX
http://finance.yahoo.com/echarts?s=^SOXX

Online Chip Design References:
http://www.anandtech.com/mac/showdoc.aspx?i=3026&p=3
http://en.wikipedia.org/wiki/Verilog
http://jas.eng.buffalo.edu/courses/ee549/cadence/Cadence_tutorial.pdf

Textbook that I learned from:
Principles of CMOS VLSI Design, by Weste and Eshragian.
http://search.barnesandnoble.com/Principles-of-CMOS-VLSI-Design/Neil-HE-Weste/e/9780201533767

Our first Sydney event on the 21st of December brought together four math PhDs, a handful of chemistry gurus and an MBA candidate. This was our first brainstorming session for the math Collaborator project that Dalibor is leading from the Sydney office. We’re very excited by the potential of this project to help thousands of mathematicians all over the world collaborate more effectively on challenging problems.

The event was organized by Dalibor Frtunik who is directing the Science House Sydney office. Dalibor will be leading events, workshops, and more in Sydney– all part of Science House’s mission to bring people together across the globe to promote and advance science.

Thanks Nickov, Laurie, Kate, Lorenn, Cliff, Karina, Julian and Dalibor for bringing your ideas to the table. We made some excellent breakthroughs from the evening and gathered a mountain of useful knowledge. We’ll be posting more on collaborator and related projects in future.

Below, our fantastic group of participants!

Books from Science House to say thanks

Over the last year, a lot of perplexed reporters have written about how genetic engineering has branched from industrial and academic labs into the garages and living rooms of amateurs. Some have highlighted its potential for education and innovation. Others have focused on its perils. Either way, it’s a growing movement whose promises and risks need to be addressed.

Science House and DIYbio will host an event about biosecurity on January 19th to open discussions between amateur biologists and law enforcement agencies about the risks and promises of moving bioengineering into the public. We are planning to have a mix of participants to offer different perspectives: from those in forensics and law enforcement, to politics, education and the sciences.

The event will begin with a short presentation from the do-it-yourself biologists and artists describing the movement, the potential of synthetic biology and the kinds of activities they do. FBI Special Agent Pat O’Brien from the Weapons of Mass Destruction Directorate will follow up with a presentation about law enforcement’s concerns as the science becomes mainstream.

We’ll end the evening with dinner and two tabletop exercises that ask all involved to define their responsibilities and responses in hypothetical situations related to the talk.

Our first invitational event, the discussion was attended by futurists, writers, computer scientists, physical scientists and entrepreneurs. More than half of those present had been attendees at the Singularity Summit held October 3rd and 4th in New York City. Run by the Singularity Institute, this event brought together hundreds of leading thinkers to discuss issues relating to the coming singularity – the moment when technological progress becomes so fast that predicting what the world will be like beyond that point becomes nearly impossible. Many Summit participants believe that it will be the development of generalized artificial intelligence that will be the key driver in the timing of the Singularity – potentially occurring in the next 20 to 40 years.

A huge thanks to all of you who took part in this event and made it a big success. It was a delight to meet those of you who were new to Science House. I will be posting information here with various links to information that was shared at the discussion. Feel free to contribute directly to this blog with your own links and recommended reading.

More photos from the event here >

What makes an interesting twitter feed? What are some of the most interesting and useful applications invented for Twitter? How do you build followers? Can twitter really be used to make useful connections with like-minded people? The Social Media Workshop at Science House began with a thought provoking talk by Paramendra Bhagat, a tech entrepreneur and social media consultant with more than 30,000 twitter followers, and gave way to a lively open discussion. At the end of the night many participants, including Science House, went on to try new ideas in their social media approach. A very worthwhile event!

From subatomic particles to 10 dimensional hyperspace, String Theory was perhaps the most exciting event at Science House yet. Speaker and PhD in String Theory, our speaker captivated the crowd with his explanation of this complex but fascinating topic. What is String theory? Could it really explain the workings of the entire universe? If all the elementary particles are composed of strings on the scale of 10-33 m, are string then composed of objects much smaller than themselves? Will it ever be possible to prove if String Theory is true? Or is it a purely mathematical construct, that will always remain so?

Attorney and Entrepreneur Dan Tedesco came to Science House to talk about legal issues facing startup companies. Dan discussed the selection of a business form (LLC? Sub S? Other?), basic corporate finance, vendor contract law, confidentiality agreements, basic intellectual property law, and how to balance legal issues and business issues.