I invited Paul McKenney from IBM to present at the 2011 Miniconf in Open Source, Multicore and Parallel Computing (part of LCA2011, conditional to the miniconf proposal to be accepted by the organisation).

But I want to share what Paul sent to me about possible topics to present (the selection would depend on the mix of proposals for that day, however all of them are interesting and available online)

From Paul…

“As to topics, there are a number of possibilities:

1.      I am always happy to hold forth on RCU.  Some historical topics may be found at http://www.rdrop.com/users/paulmck/RCU.  In addition, I could talk about:

a.      User-level RCU (joint work with Mathieu Desnoyers).

b.      Conceptual properties of RCU (joint work with Mathieu Desnoyers, Michel Dagenais, Alan Stern, and Jon Walpole).

c.      Recent Linux-kernel RCU software-engineering work. Because 3500+ uses of RCU is too many to keep track of manually.

2.      “Is parallel programming hard, and if so, why?”  Lots of people answer “yes” without really thinking this through.  An overview may be found at:

http://paulmck.livejournal.com/tag/is%20parallel%20programming%20hard

3.      I could do a comparison of transactional memory and locking. This builds on an earlier SOSP paper and presentation:
http://www.rdrop.com/users/paulmck/scalability/paper/TMeval.2007.09.24a.pdf
http://www.rdrop.com/users/paulmck/scalability/paper/TMevalSlides.2008.10.19a.pdf

This topic would be reasonably likely to induce some excitement in the audience.

4.      I could give an overview of remaining challenges facing transactional memory.  This would build on my blog series from last year:

http://paulmck.livejournal.com/tag/transactional%20memory%20everywhere

This topic would be -really- likely to induce some excitement in the audience.”

Advertisements