I've merged these changes in master. I've also added another change that results in yet another large perf boost:
- direct invocation of known fns instead of going through .call It's now possible to write CLJS that has the exact same performance as handwritten JS - I compared the handwritten JS spectral norm from Alioth to one written in CLJS running under Node.js - the CLJS is actually a tad faster on my machine. This probably due to GClosure. I also tried zebrao from core.logic. On my machine I consistently see times just under 30ms in Google Chrome. David On Sun, Apr 8, 2012 at 8:59 PM, David Nolen <dnolen.li...@gmail.com> wrote: > The multi-arity fn optimization was causing some code size explosion. I've > committed a refactor that eliminates the issue. Please try! > > On Sun, Apr 8, 2012 at 6:26 PM, David Nolen <dnolen.li...@gmail.com>wrote: > >> I've spend the past weekend doing some comprehensive optimizations to >> ClojureScript, these can be seen here: >> https://github.com/clojure/clojurescript/compare/all-optimizations >> >> With these optimizations operations are anywhere from 1.5X-5X faster >> depending on the operation and the JavaScript environment. >> >> These optimizations include: >> >> * respect boolean type hints on fns >> * remove CLJS truth tests if we can infer that an if's test expression is >> boolean - this works for common combinations of and/or >> - this removes nearly all truth tests from PersistentVector, big wins >> here and for other persistent data structures to come >> * multi-arity fns that are not used higher-order are now dispatched >> directly to the correct arity instead of using JS arguments >> - multi-arity protocol fns also benefit from this optimization >> * remove all truth tests from protocol dispatch fns >> >> Feedback & improvements appreciated. Please try this branch out and let >> us know if it works for you - especially interested if things are faster, >> slower or if the changes adversely affect your file sizes under advanced >> optimization. >> >> David >> > > -- 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
