Hi @remysucre, there are no plans to introduce parallelizations of community detection algorithms, at least not from my side. Although the performance may increase, as you indicate, there are a number of algorithms that do finish (even if it is over night). I believe that the Leiden algorithm should be one of the fastest among those (disclaimer: I am one of the authors of that algorithm).
What “quality” was not good enough?
The Leiden algorithm supports also a different objective function (CPM, instead of Modularity), and you may also tune the resolution parameter. If your problem is that you need to explore many different resolution parameters, that is something that is actually trivial to parallelize. This will also be a much more efficient gain compared to trying to parallelize the algorithms themselves.
I am not sure what the goal is of your analysis? Of course, one dataset is not better than another dataset just because a community detection algorithm works “better” somehow.