On Wed, Aug 12, 2009 at 3:03 PM, Andy Fingerhut<andy_finger...@alum.wustl.edu> wrote: > > My apologies for the noise if this is well known in the Clojure > community, but I'll ask anyway. > > One of the tweaks to my Clojure benchmarks that people have suggested > for improving performance, and that does help, is changing some > function definitions to macros. This is in effect inlining those > functions at the source level, so the Clojure compiler has a shot at > it. > > Is there some reason that the Java JIT is not doing this, with the > original code using defn, as fast as it works when using defmacro?
I think inlining via Clojure macro or :inline has the most benefit when it allows you to avoid boxing arguments and return values. That is, if you have primitive locals in the calling function and primitive locals in the called function, the Java method signatures created by the Clojure compiler will still be Objects and require boxing and unboxing for each invocation. I don't know for sure, but it appears HotSpot doesn't (usually? ever?) remove that un/boxing when inlining. Using a Clojure macro or :inline metadata allows the Clojure compiler to use the same local primitives with no boxing or unboxing. --Chouser --~--~---------~--~----~------------~-------~--~----~ 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 -~----------~----~----~----~------~----~------~--~---
