I've actually tried with and without type hinting and end up with similar performance numbers. I haven't tried type hinting the second execution.
On Feb 18, 7:49 am, Ken Wesson <kwess...@gmail.com> wrote: > On Fri, Feb 18, 2011 at 7:03 AM, Daniel Werner > > > > <daniel.d.wer...@googlemail.com> wrote: > > On Feb 16, 6:20 am, Nick <npatric...@gmail.com> wrote: > >> (let [ newv1 (time (doall (map (fn [v u I] (+ ^java.lang.Double v (* > >> 0.5 (+ (* (+ (* 0.04 ^java.lang.Double v) 5) ^java.lang.Double v) 140 > >> (- ^java.lang.Double u) ^java.lang.Double I)))) v u I))) > >> newv (time (doall (map (fn [v u I] (+ v (* 0.5 (+ (* (+ (* 0.04 v) > >> 5) v) 140 (- u) I)))) newv1 u I)))] > > >> I've tried without type hints, with type hints, different ways of > >> doing the type hints, but I cannot explain the results that I'm > >> seeing. With all of the variations I've tried, I've always seen > >> timing like this: > > >> "Elapsed time: 49.876243 msecs" > >> "Elapsed time: 0.179701 msecs" > > >> What am I missing here? Is the compiler getting some advantage to > >> executing the same function twice that cannot be gained in the first > >> execution? > > > Since nobody has replied so far: One explanation I could imagine is > > that your type-hints in the first invocation force the compiler to use > > object types for all the nitty-gritty calculations, whereas the second > > invocation is able to use primitives for calculations, having to box > > them only when the final sequence is produced. > > Yes, that's exactly right. Some operators like + Clojure's compiler > knows to inline and then they don't have to box primitives. The type > hints force boxing in this case. -- 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
