So the significant improvement on performance is showed not only in quick startup and less resource requirement but in a long runtime. Is it more related to the enhancement in VM? Besides the difference in VM, maybe we can spot out some bottlenecks in classlib and add them to our todo list. Aside from the stability and compatibility, performance is another important factor to influence our potential customers.
On 12/14/06, Stefano Mazzocchi <[EMAIL PROTECTED]> wrote:
Alexei Fedotov wrote: > Hello, > JFYI, Sun shipped the next major Java release: > http://java.sun.com/javase/6/. yes, and I got an amazing 40% speed improvement on winxp emulated inside parallels on macosx over java5 on the application I'm currently working on (some in-memory RDF manipulation library).... so much that it's faster to run java6 on winxp emulated than it is to run it on macosx's java5 (haven't tested harmony there yet, I'm waiting for the blessed snapshots). This seems to be consistent with my previous benchmarks on linux: Sun's java6 is perceptually faster than Sun's java5. -- Stefano.
-- Leo Li China Software Development Lab, IBM
