On Thursday, Jan 9, 2003, at 15:12 US/Pacific, Fregas wrote: > As we all know, CFMX compiles to Java Bytecode and gets cached this > way. Therefore, CFMX should be considerably faster than CF5. > However, a friend of mine did a very simple perfornance test using > CFMX that had NO database interaction (just looping and finding prime > numbers) and didn't see much of a speed increase. Is there something > special you have to do in the new cf administrator to force it to > cache these files as Java classes/bytecode? This is CFMX Pro, not > andy of the J2EE versions.
As has been discussed, here and elsewhere, heavily numerical work tends to be slower in CFMX than in CF5. For "typical" CF web applications, CFMX is usually quicker than CF5 and often much quicker. Why is numerical work sometimes slower in CFMX than CF5? Numbers in CF are held as strings for the most part and converted as necessary to floating point numbers to perform arithmetic. CF5's interpreter - written in C++ - processed this sort of thing faster than Java's JVM. Not surprising: C++ is generally very fast at processing strings (heavily optimized string library) compared to Java. At least, that's why I think you're seeing this result. Of course, not many real-world CF applications spend time calculating prime numbers :) Sean A Corfield -- http://www.corfield.org/blog/ "If you're not annoying somebody, you're not really alive." -- Margaret Atwood ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~| Archives: http://www.houseoffusion.com/cf_lists/index.cfm?forumid=4 Subscription: http://www.houseoffusion.com/cf_lists/index.cfm?method=subscribe&forumid=4 FAQ: http://www.thenetprofits.co.uk/coldfusion/faq Get the mailserver that powers this list at http://www.coolfusion.com Unsubscribe: http://www.houseoffusion.com/cf_lists/unsubscribe.cfm?user=89.70.4
