I've now been living in New York for three years. In the same apartment, no less. That means that after you subtract out the summers I was away from Boston, I've lived here for about the same amount of time that I lived at Random. Funny, it doesn't feel as long.
In the spirit of retrospection, I went over to Google Scholar to see how the JCilk papers. Aside from the usual incestuous citations, I was very excited to see that my thesis had been cited by people I'd never heard of. That meant that someone actually it! Well, okay, at least some of it, but still--somebody read my thesis! Then I searched through their paper for the reference. I quote:
The C implementation for a similar parallel system, called Cilk, introduce a similar overhead for this benchmark (3.63x slowdown ). Our system however, signiﬁcantly outperforms the Java version of Cilk (JCilk) which imposes a 27.5x slowdown for this benchmark ).
So they flipped through my thesis, found the biggest, worst number in there, and quoted it to make themselves look better by comparison. And in another paper, they used the analysis from my own Performance chapter to discredit our whole idea, ignoring my suggestions for improvement:
With the volunteer stack splitting mechanism, we avoid the synchronization overhead incurred by work-stealing, which can be signiﬁcant in a Java system .
The sad part is, I'm still kind of excited about it. Somebody read my thesis!
In making-me-feel-old news, the froshlings from my 6.001 class have now graduated. I looked up the graduating Course VI class on the alumni database, and was gratified to see that I didn't manage to scare them all off. I was especially happy that one of my students who seemed to be getting things by the end, but who still got a No-Record, tried it again and ended up graduating Course VI on time.
Of course, once one of my students starts working at my office, then I'll really feel old. It probably won't be long now..