Beautiful! the results are much clearer this way, and it's good to know why. I noticed (googled) that you have had this figured out long ago :) This was a great learning excercise, and I'm glad I was on the right track.
br, jukka > -----Original Message----- > From: Eli Barzilay [mailto:e...@barzilay.org] > Sent: 03 October 2011 22:57 > To: jukka.tuomi...@finndesign.fi > Cc: Sam Tobin-Hochstadt; Racket mailing list; Robby Findler; Ivanyi > Peter > Subject: Re: [racket] Concurrent execution > > > Four hours ago, jukka.tuomi...@finndesign.fi wrote: > > > > Wow, that's great Sam! > > > > I tried to create a simple test around it (below), but it's > hard to find a > > pattern in the results. A few notes, though: > > > > - The results vary a great deal from run to run > > - 'Plain map' seems to do things in parallel itself!? > > (by looking at the System Monitor during the test run) > > This may paralyze secondary parallelizing :) > > - Plain map does usually better with short lists. No wonder. > > - The amount of distribution should be related to the available > > number of HW processors; not the more the merrier, obviously. > > Using factorial to measure speed is not a good idea -- since with > bigger numbers much of the time is spent on GC work for the bignums. > Also, it looks like you were timing the printouts too, and with big > numbers that can take a considerable amount of time too. > > Here's a short test using a factorial -- note that pmap takes a bit > more overall cpu time, but the real time is much shorter. > > -> (define (pmap f xs) > (map touch (map (λ (x) (future (lambda () (f x)))) xs))) > -> (define (fib n) (if (<= n 1) n (+ (fib (- n 1)) (fib (- n 2))))) > -> (time (map fib '(35 36 37 38))) > cpu time: 5088 real time: 5098 gc time: 0 > '(9227465 14930352 24157817 39088169) > -> (time (pmap fib '(35 36 37 38))) > cpu time: 5106 real time: 2292 gc time: 0 > '(9227465 14930352 24157817 39088169) > > -- > ((lambda (x) (x x)) (lambda (x) (x x))) Eli Barzilay: > http://barzilay.org/ Maze is Life! > _________________________________________________ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/users
