From: D. M. M. <mic...@ro...> - 2010-01-14 00:04:25
|
On Wednesday 13 January 2010, Abrolag wrote: > There is one thing I've noticed, but haven't mentioned up to now, and > that is processor usage. This is much higher than 'classic'. Is this > due to some debug code? Maybe to some extent, but a lot of this is just that we have a lot of brand new code in very important places, and have not yet had time to do much of anything to streamline it. (Remember that not only is our own code brand new in many important places, but it also depends on new Qt code too. Qt has had a lot longer to shake these problems out, but I've seen enough variation in behavior going through different Qt versions along the way that I'm not at all convinced the library code we're using is stable and mature.) Just to establish one benchmark, I have a gigantic MIDI file I like to test with. Open the thing in 1.7.3, open all the segments in a 17-staff, 854- measure notation view, and it takes about six seconds. I haven't tried this test again since Chris made one performance tweak, but the last time I did (a few days ago) it ran for five minutes before X11 finally died. Even if his performance tweak cut that in half and avoided killing X11, it's still 150 seconds vs. 6 seconds. Classic wins clearly on this test. We're not going to able to tweak all of this stuff up overnight, and that's not even the focus of the February release. We'll get there though. -- D. Michael McIntyre |