Live update of the plots can be turned off in LTspice. The simulation speed might come from the fact that QUCS is not based on SPICE, it has its own engine/solver, or that's what I was told.
Anyways, the simulation time for a given circuit, or even the success/fail to converge to a solution for either of these can vary widely by adjusting the max allowed steps/errors or other tuning parameter of the solver. At this aspect LTspice might be advantageous because it has a wider adoption, and solutions to problems are already known, so it's easy to find online how to get out of LTspice troubles.
QUCS still have some nice RF goodies, filters, microstrips, Smith charts, etc. but other things like placing a measuring probe are cumbersome in QUCS when compared with LTspice. That's what makes me use mostly LTspice. Only very rare I found myself using QUCS. QUCS has a lot of examples and a nice PDF with a few examples and how to use QUCS in general, some in the Help menu from QUCS, or online:
http://qucs.sourceforge.net/docs.htmlI have QUCS installed on Kubuntu, don't recall how I solved the qt4 requirement. I guess I've let it install all the needed dependencies, or maybe added qt4 manually, IDK.
LTspice only has a Windows installer, but works fine in Linux, too, from Wine. LTspice charts looks more appealing to me, and are much smarter than the charts in QUCS. For example, the charts know to put electrical units on the axes, you can add formulas from the GUI, the plot "knows" that U*I formula will measure Watts, has RMS, has useful options for Bode plots, like showing group delay or using linear instead of log, knows octaves and decades, etc.
For more plotting the data can be exported (I think both QUCS and LTspice can do that). For example, yesterday I've fool around plotting LTspice raw data using Python and matplotlib. There is even a Python module called "ltspice", module that can understand the data format of the LTspice raw results file. Doesn't have implemented all the features (i.e. the Python module doesn't identify automatically the data sections for multiple parametric runs when using the .step SPICE directive, but can be easily seen from the module's sources how to grab a certain step from the data file, or how to add new features to the Python module.
LTspice can be run in batch mode, from command line, and the .net or .asc schematics can be eventually generated or edited from Python, then all the runs and data files collected and post processed, all done programmatically from Python.
Because many other physics phenomena can be simulated with electrical circuits (e.g. a mechanical weight and a spring will behave like an LC tank, water ram in a pipe can be a reflection in a transmission line, a sound absorbent material can be seen like a filter, etc.) mixing Python with LTspice can be a very interesting combination.