j
: Next unread message k
: Previous unread message j a
: Jump to all threads
j l
: Jump to MailingList overview
Bruce O'Neel wrote:
It now has a place of honor on my wall. BTW, I noticed the 26 benchFib line in there. If you don't mind, what does [26 benchFib] timeToRun return?
If the inlining is on (big and (supposed to be) fast interpreter), the result is around 7000 on faster versions of Zaurus. If the inlining is off, the result will be three times bigger than that. (I can't tell the actual result now.)
Note that if the executable is bigger than 128k, ZaurusOS starts paging for binary segment, and the non-inlined version of VM is just smaller than 128k. So, the difference of the results of micro benchmark doesn't reflect the performance of 'real' applications. They runs much faster on non-inlined VM.
OHSHIMA Yoshiki Dept. of Mathematical and Computing Sciences Tokyo Institute of Technology
squeak-dev@lists.squeakfoundation.org