In an ideal world, standard functions like map, sort, reduce, filter, etc. would know when to parallelize on their own, or even better, the compiler will do it for them.
Meanwhile, I tried playing with the parallel lib ( http://clojure.org/other_libraries ) which has functions like preduce, psort, and par:map functions, but I wasn't very successful. Anyone who knows what s/he's doing can shed some light on how to use it with this example? Mibu P.S. psort and sort's arguments are inconsistent (psort [coll comp] vs. sort [comp coll]). On Dec 30, 3:56 pm, Piotr 'Qertoip' Włodarek <qert...@gmail.com> wrote: > On Dec 29, 7:02 am, Chouser <chou...@gmail.com> wrote: > > > > (defn split-string-in-two [s] > > > (let [chunk-size (quot (count s) 2)] > > > [(subs s 0 chunk-size), (subs s chunk-size)])) > > > Might this cut a word in half and produce (slightly) incorrect > > results? > > True, I decided to let it be for the sake of simplicity. > > > > 1. Is there a better way to do it? Perhaps agents should share some > > > data structure? > > > I've got lots to learn in the realm of parallel program design, but > > the sequential summing step at the end stood out to me. Perhaps > > another agent that just does incremental summing along the way would > > reduce the running time. > > > Also, the entire file is read in by a single thread -- perhaps the use > > of mmap would allow the agents to start counting sooner and reduce > > total run time that way. You may want to look at clojure.contrib.mmap > > Thanks for suggestion. > > In fact, mmap/slurp is *much* faster than built-in slurp. > > Now, my program starts with: > > (ns com.wlodarek.examples > (:require [clojure.contrib.mmap]) > (:require [clojure.contrib.duck-streams])) > > (defn read-file [filepath] > (str (clojure.contrib.mmap/slurp filepath))) > > (defn write-file [filepath string] > (clojure.contrib.duck-streams/spit filepath string)) > > ; ... > > I hope Clojure will soon have fast, reasonably named read/write file > functions built in the core. > > Now I'm trying to parallelize also the formatting part, and to move > from 2 agents to n agents. > > Again, thanks for all tips! > > regards, > Piotrek --~--~---------~--~----~------------~-------~--~----~ 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 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 -~----------~----~----~----~------~----~------~--~---
