> A "simple" jit would probably give a 5-10x+ speed up of the java
Yes, this is what I observed in my tests of Nokia 6680 and Sony Ericsson X1 versus Android G1 for the same computational benchmark. It is not good for battery life either when the CPU has to spend 5-10 times as many clock cycles for a given task. On Apr 1, 8:25 pm, dm1973 <david050...@gmail.com> wrote: > Correlation is not causation. Yes android is slower than it could be. > But jazelle isn't the only reason. I wouldn't be surprised if the j2me > variant isn't even using jazelle as jits tend to be faster in a lot of > applications. A "simple" jit would probably give a 5-10x+ speed up > of the java (not necessary the app since a lot of time is spent in > native code) and a really good one gets within 10% (and faster in some > cases) than native C code, a generational garbage collector would make > allocation fast and reduce some of the pauses, and so. There are > costs in memory usage and complexity for both of them. > > Sun has been working on their VM for over 15 years. Google is more > like 2 or 3. I would expect as the platforms get more ram and as > Google spends more time that we will see some pretty decent > performance improvements. Personally I would have preferred that > Google use java byte codes instead of come up with their own scheme > since there are a lot of languages out there that compile to byte > codes but they had their reasons but there is no reason why in 5 years > dalvik code isn't as fast as any other VM out there. --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Android Developers" group. To post to this group, send email to android-developers@googlegroups.com To unsubscribe from this group, send email to android-developers-unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/android-developers?hl=en -~----------~----~----~----~------~----~------~--~---
