Thanks. I have seen some adhoc comparisons on-line. Mostly just matrix multiply. Having said that I wouldn't be surprised if I missed something. Based on personal experience I would expect about a 2-3 times speed hit between well written java and c/c++ code because of array overhead and language constraints. For pure arithmetic I have gotten nearly identical performance.
An accurate comparison could be done using this benchmark by writing a wrapper with JNI. The overhead from JNI wouldn't be measured because of how the benchmark is designed. I just grabbed the latest snapshot of commons-math-2.1a and will run it through the benchmark. - Peter On Fri, 2010-01-29 at 22:51 +0000, Weijian Fang wrote: > Hi, > > Great job! > > BTW, I am just curious if there is any performance comparison between > Java math lib and C/C++ math lib? After so many years, people are > still suspicious about Java's performance, particularly in the > numerical computing area. > > Cheers, > > Weijian > > > > On 29 January 2010 20:58, Ted Dunning <[email protected]> wrote: > > Already the performance is impressive. Kudos! > > > > On Fri, Jan 29, 2010 at 12:52 PM, Luc Maisonobe > > <[email protected]>wrote: > > > >> > I think that the developers of this library might find those results > >> > interesting. In addition, feedback on the benchmark itself and how I'm > >> > evaluating commons math would also be appreciated. > >> > >> Thanks for this work. Up to now we had almost no reliable comparison > >> tools. This will help us improve our library. > >> > > > > > > > > -- > > Ted Dunning, CTO > > DeepDyve > > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [email protected] > For additional commands, e-mail: [email protected] > --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
