> An interesting founding is that it seems typecasting from short to int > and asetting to an int-array is 3-4 times faster than typecasting from > int to short and asetting to an short array.
Thought: Java primitive types, when manipulated individually, are all "secretly ints" — there is no real unsigned int, short, byte, etc. type in Java. The only time those declarations matter are: * For range checking/documentation * In arrays, where the JVM will pack values to save space. You could only know for sure what's happening by inspecting the bytecode and JITted assembly, but I suspect that ints are faster for one reason: when you're dealing with shifting free shorts between variables and arrays, you're dealing with a representation change. With an int, a 32-bit store is all that's needed for array setting. For shorts you need a fetch a segment of the array into a 32-bit register, use bitwise operations against that segment and your "short" (stored in 32 bits!) to set the appropriate part of those 32 bits, and a store back to the array. When you're treating that short as an int, your simply setting a 32- bit region of the array. --~--~---------~--~----~------------~-------~--~----~ 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 -~----------~----~----~----~------~----~------~--~---
