Hi Lee,
I've done some more digging and seem to have found the root of the
problem,
it seems that java native methods are much slower when called in parallel.
The following code illustrates the problem:
(letfn [(time-native [f]
(let [c (class [])]
(time (dorun (f
(fn [_] (.isArray c))
(range 10000000))))))]
(println "Sequential Test:")
(time-native map)
(println "Parallel Test:")
(time-native pmap))
On a dual quad-core xeon box I get the following results:
Sequential Test:
"Elapsed time: 1054.807725 msecs"
Parallel Test:
"Elapsed time: 15302.605697 msecs"
ie. the code executed in parallel was 15 times slower than the sequential
version.
The same can be seen with isInstance and isArray members of java.lang.Class.
I'm not sure where to go from here, these functions are frequently used by
clojure.core
Perhaps someone with more JVM implementation knowledge can help?
Cameron.
--
You received this message because you are subscribed to the Google
Groups "Clojure" group.
To post to this group, send email to clojure@googlegroups.com
Note that posts from new members are moderated - please be patient with your
first post.
To unsubscribe from this group, send email to
clojure+unsubscr...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/clojure?hl=en
