First I want to say that the menu "search for similar files" is really really really useful. I enjoy a lot this feature. However, finding similar images by content (visual similarities) is really slow. I understand that it's not straightforward to compute it, but, since I changed my CPU, I noticed that this function is not multi-threaded.
So it means that for most of the people, this function is at least 2 times slower than it should be (and now quad-core cpus are not uncommon). I imagine that the algorithms is: extract features from images, then find similar pairs of features vectors, so it sounds quite easy to make a parallel version of that. I said "sound", but I'm not an expert, maybe it's not possible to do it
. But in the other case, it will be really useful to do it!Hope that my request make sense and has not been proposed
