hackbod wrote: [...] > Without knowing anything about the hardware or software running on > them, I would be very hesitant to make many assumptions from that.
Indeed. [...] > That said, you certainly don't want to have a UI with 100s of views, > because it will be slow. But this is as much because you are trying > to run on a mobile device and not a desktop computer -- in general > mobile CPUs are much slower than their actual MHz may make you think. Well, the last place I worked produced a JVM that at one stage had a mobile phone GUI in Java that was fast and responsive on a 13MHz ARM. No, that's not a typo. So while I do know exactly what you mean, I'll also point out that it's still possible to get a surprising amount of work out of them if you're careful! [...] > It is not that simple. There are both advantages and disadvantages to > using a JIT. One of the disadvantages is more memory use, and memory > use is actually about our #1 performance issue on actual hardware. [...] > At any rate, there definitely won't be a JIT in 1.0, it's just not > that necessary. Fair enough, especially after I read your post on -developers that described the 16MB hardware limit for process memory usage; I now have all kind of kinds of curious questions about what kind of weird MMU goldfish have. But you have the hardware, which means you get to actually measure the real performance, which we can't do. Also, you're doing all the work, which means you get the only vote worth having... but once the source for 1.0 comes out, I'd be extremely interested to poke around the Dalvik VM and see what might be done with it. -- ┌─── ｄｇ＠ｃｏｗｌａｒｋ．ｃｏｍ ───── http://www.cowlark.com ───── │ "I have always wished for my computer to be as easy to use as my │ telephone; my wish has come true because I can no longer figure out │ how to use my telephone." --- Bjarne Stroustrup
signature.asc
Description: OpenPGP digital signature
