On Jan 2, 2011, at 9:13 AM, Marek Kubica wrote: >> But in theory it could be posible to collect run-time data in one run, >> then JIT code at startup, using that collected data and current >> procesor architecture. > > Something like this is already used in practice, it is called > Profile-Guided Optimization (PGO) and available in GCC (but that uses > AOT, which sounds to me like equivalent to "JIT at startup"). The > OpenJDK-JVM does not support this and I doubt that it will. Maybe GCJ > at some time, but as far as I saw, GCJ does not work with Clojure yet.
A while back I was talking with someone from Sun that had done a fair bit of work on the client compiler, and this topic came up. Apparently, its implementation (and perhaps presumably that of -server as well?) is such that optimized code references memory locations specific to the then-running VM. It would apparently be trivial to dump JITed code, but you'd never be able to load that code again. - Chas -- 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
