I have an application which is a waterbalance model that runs in 300 loops and shows the results in a series of labels on an interface. A button creates and starts the thread that is the model. The model is coded in c++ and I could port it directly to Qt/Mingw using Eclipse, from borland builder c++ 6.0. So the model code is exactly the same.
The weird thing is that in Borland the model runs in 17 seconds, in Qt 4.6.1 it runs in 26.5 seconds.
There is one difference though, in Qt the Borland the model is directly called in Qt it is a Qthread, but also if I switch of the interface the runtime is nearly the same (a lot slower than BCB).
Is the Qthread making the model slower? or is it mingw? Or is this a mingw question? Most of the data is grid based, so a lot of times "for c = 0 to nrcols, for r = 0 to nrRows", could it be that borland does that faster? Both are compiled as debug without optimizations.
This is with a small database but a larger database may easily be 30 minutes so before I continue in Qt I would like to be sure it is not a lot slower.
Has anyone experienced this?
thanks
Bookmarks