Get email delivery of the Cadence blog featured here
Last night was the annual Kaufman Award dinner to present the award to this year's honoree, who is Andrzej Strojwas. I say "this year" but in fact this was the 2016 Kaufman Award. The dinner is usually held in October or November but this year it slipped out of 2016 into 2017. So this might be the year of two Kaufman Award dinners. Soon after the 2016 award was announced, I went to PDF Solutions and interviewed Andrzej. You can read that post here.
Just in case you don't know who he is, Andrzej actually has two full-time jobs: he is CTO of PDF Solutions, and he is a professor at Carnegie Mellon University (CMU) in Pittsburgh, Pennsylvania. In this second capacity, he has supervised 47 Ph.D. students, three tables of whom (30) were in the audience. One of whom was John Kibarian, one of the founders and the CEO of PDF Solutions. Through both jobs, and also working very closely for a period with KLA Tencor, Andrzej has revolutionized the way that semiconductor manufacturers ramp processes to volume. When TSMC show how they ramped 16FF from a standing start to 50,000 wafers per month in just 3 months, Andrzej (and colleagues) are a large part of the reason.
In the 1970s, there were really only semiconductor engineers, who both designed the ICs (with just a handful of transistors) and nursed them through the manufacturing process. Then, in the early 1980s, design and manufacturing completely split into two with the Mead & Conway revolution. Andrjez started to put them back together again, since it wasn't effective to keep them separate.
Andrzej has been responsible for many innovations. But one of the biggest ones (that those of us that are not specialists can understand) is this. Inspecting an entire wafer at the resolution necessary to find all the defects is just not possible. It would take too long to be useful. I've heard it likened to searching the state of California for all the golf balls. Andrzej's insight was that if you analyze the layout of the design carefully, you can find the weak spots where failure might occur, and you look only there. Maybe an analogy would be that if you want to find the golf balls in California, then get a map and find all the golf courses. As Ken Levy, the founder and (then) CEO of KLA Tencor told Andrzej, that would not work because "our customers [namely the manufacturers] don't have access to the layout." Instead, they just get the masks and manufacture the wafers. That would change, and it made finding a good proportion of the defects when ramping a process feasible. The next step was to deliberately put structures in unused parts of the reticle that stressed the process and which could then be inspected. This is known as Design for Inspection, or DFI.
Most Kaufman Dinners I've been to have been relatively staid. One or two people give a summary of the honoree's achievements, the honoree gives a speech. We all go home. Yesterday's was much more emotional, with both John Kibarian (CEO of PDF Solutions) almost in tears running over the history of Andrzej's career, and Andrzej himself overwhelmed emotionally even more.
John highlighted three personal qualities of Andrzej. First, his open-mindedness. Most engineering consists of dividing and conquering---but yield optimization is not like that, and requires reaching out across lots of disciplines (like layout and manufacturing, as I already talked about above). Next, his persistence (aka stubbornness). When he was trying to convince KT to use the critical layout approach was when Ken said it wouldn't work because they didn't see the layout. But he persisted, and now the entire industry uses the layout approach all the time. Third, his generous spirit. John's relationship with Andrzej started 30 years ago when he was an undergraduate, then a graduate student; then, of course, he founded PDF Solutions, which commercialized many of Andrzej's ideas, and where Andrzej was (and is) the CTO.
Andrzej took his 8 minutes (that's a joke, as you know if you were there) to point out how much of his career has depended on fate (and luck), and on decisions he never took himself. As you might guess from his name, he was born and raised in Poland. He wanted nothing more than to go sailing and to see the world, so he enrolled himself into a high school that was on a ship, I think. But his father found out and made him instead go to a high school where the instruction was in English. As a result, he spoke fluent English in an era (this is still USSR era) when few people in Poland spoke any. He also met his wife in high school. As he put it, "after a short courtship of 7 years we finally got married."
The next stroke of fate was that he was one of a handful of people selected to study statistical modeling of the IC industry. Of course, the IC industry in Poland at that time was a bit of a joke, so this could have been a dead end. But they started to produce papers and some of them got noticed. Next stroke of fate: Steve Director (that's his last name, not his title) at CMU invited him to an internship at Harris in Florida. This was also unusual since suddenly you had a Polish citizen from behind the iron curtain doing research in the fab producing chips for the B1 bomber. Apparently, the state department guy was so speechless when he found out that he signed off on it.
His original plan was to go back to Poland after the internship, but then martial law was imposed in 1997; this was the next stroke of fate. The obvious thing to do was to stay. Two centers of excellence were created in CAD by the US government—at CMU and Berkeley. Steve Director wanted him at CMU but let him interview at Berkeley and Stanford. According to Andrzej, he then called each school before he interviewed to say they were wasting their time since he'd already accepted at CMU. So he accepted his fate and went to CMU, where he has been ever since.
One person in his classes was John Kibarian, who stood out because he read the Wall Street Journal during lectures. By then, his group was supporting a lot of semiconductor companies with their modeling technology, and it was completely unsustainable. John saved him by being one of the founders of PDF Solutions. As Andrzej said, "There is no way I would be here today without PDF Solutions."
He wrapped up by talking about his wife. She was a skilled researcher in mathematics, although she abandoned her career when they started their family. But since they have the same last name, he regularly gets calls from people asking him complex mathematical stuff. In fact he thinks that, due to this confusion, his legacy will be "he did some pretty good stuff in DFM, whatever that is, but he was a great mathematician."
After that the evening wrapped up with a magician, Robert Strong. I can tell you roughly how one of the tricks was done. I arrived an hour early to make my regular What's for Breakfast video there. I watched Robert film his own video that he used in a trick where later he essentially is carrying on a conversation with the on-screen version of himself, and I saw some of the other preparation too. But my lips are sealed.
By then it was 9.30pm so time to go home. Congratulations Andrzej!