Practice Meeting, Jan. 7th

[Lyle Kopnicky]

When: Jan. 7th, 6:30-8:30pm

Where: Collective Agency Downtown, Suite 1108, 511 SW 10th Ave, Portland, OR (opposite side of the floor from the elevators)

We'll continue our discussion of Chapter 4, "Dataflow Parallelism: The Par Monad", of Parallel and Concurrent Programming in Haskell. We'll pick up with discussing the conference timetable example on page 73. At the meeting anyone will be able to present and contribute to the discussion.

Meetup link to RSVP:

https://www.meetup.com/Portland-Functional-Programming-Study-Group/events/rtfghqyzcbkb/

Hope to see you then!

[Lyle Kopnicky]

We had a great meeting, where we finished Chapter 4.

It turned out that the 40x memory usage issue I had seen with my rate-limited version of the pipeline parallel code was illusory. I ran the original version and the rate-limited version again, and total memory usage was very close. Peak memory usage of the rate-limited version was less than twice the original, and that was expected because it's sometimes calculating two chunks in parallel.

We finished the discussion of the parallel timetable solver, and discussed the parallel type inferencer. I was a little disappointed because it really just does parallelism at the level of separate functions, no parallelism within different parts of an expression, which I thought would be difficult and interesting. Possibly it's too fine-grained to be worth it.