I think I have found the problem: the qaccessible interface... While my program was kind of hanging during those long seconds, I have broken several times in the debugger, and saw that some accessibility code was being executed. Then I did this quick Qt change: in qaccessible.cpp QAccessible::queryAccessibleInterface(...) I insert "return 0;" at the start... et voilà, back to normal speed! (actually, a 20% reduction compared to 4.8, but this is the day and night compared to before). This code change is a dirty hack of course, but it simply make Qt 5 usable again for me...
There is definitly something wrong on the Windows accessibility code... Philippe On Mon, 15 Apr 2013 07:45:55 +0200 "Michael Klank" <michael.kl...@amk-antriebe.de> wrote: > We have a similar problem here. I did compare a small (bigxml) program from > http://qt-project.org/ which reads an xml file in a QTreeWidget and expands > all > nodes. The test xml file is 2MB. Qt5.0.2 MSVC2012 64Bit/32Bit is up to 20x > slower than Qt4.8.0 MSVC2010 32Bit, in release mode!!! That's strange. And I > didn't say anything from debug mode! > > > -----Ursprüngliche Nachricht----- > Von: interest-bounces+michael.klank=amk-antriebe...@qt-project.org > [mailto:interest-bounces+michael.klank=amk-antriebe...@qt-project.org] Im > Auftrag von ???? ?????????? > Gesendet: Sonntag, 14. April 2013 20:14 > An: Philippe > Cc: interest@qt-project.org > Betreff: Re: [Interest] Qt 5.0.2 speed and widgets > > Yes, we are having same problem here. Same setup, AFAIK. > > Don't know what's causing problem. > > Also, Mac OS has no noticeable speed regressions, it seems Windows-specific > issue. > > ???? ?????????? > > 14.04.2013, ? 21:55, Philippe <philw...@gmail.com> ???????(?): > > > Compared to Qt 4.8.x, I experiment a very sensible reduction of speed > > in 5.0.2 when widgets are created or complex windows activated. > > In Release mode this is "sort of acceptable",... maybe half speed. > > But in debug mode, this is dramatically slower, up to 6x slower than > > in 4.8.3. > > > > This is not a painting issue as the speed inside a window has been > > measured to be sensibly the same to 4.8.x (using ANGLE). > > > > Anyone else sees the same? Any known reason for this? > > > > Setup: Windows 7, Visual 2010, Qt 32 bit, *very* fast computer. > > > > Philippe > > > _______________________________________________ > Interest mailing list > Interest@qt-project.org > http://lists.qt-project.org/mailman/listinfo/interest _______________________________________________ Interest mailing list Interest@qt-project.org http://lists.qt-project.org/mailman/listinfo/interest
