In the default configuration it is triple buffered. Double buffering is already done by Qt for any widget and the 3rd buffer avoids replots in combination with overlays ( f.e the rubberband of the zoomer ). You can disable both buffers, but beside for live plots I don't see any good reason why.
I'm always surprised that almost every library dealing with graphics gets advertised as "high performant" while IMO such a statement is completely pointless without the specific usecase. Qwt offers several optimizations ( or workarounds for Qt bottlenecks ) and algorithms, but it's up to you to combine them to a performant plot application. ( F.e you can ruin any performance of a plot widget by using 10000 markers instead of a curve with 10000 points ).
If the buffering of QwtPlot makes an application more or less fluid depends on the type of plot you want to implement. But in data intensive applications its more about the question how to reduce what needs to be painted - the performance of painting itself is less important. How to do this depends on your requirements and the characteristics of your data - often you end with a combination of point weeding and polygon clipping. Check the archive of this forum for "levels of detail".
Uwe
PS: Note that Qwt from SVN trunk reintroduces several optimizations that were gone in Qwt 6.0.
Bookmarks