Most of the math papers are not comparable to Perelman's proof in importance( that should be obvious) but most of them are mathematically correct and interesting to mathematicians. People will often see something on the arXiv and look at it. On the other hand, as I mentioned, people are also more likely to look at a preprint if they know it is actually accepted in a reputable journal.
Recently, I gave some reasons for SIAI to begin publishing in mainstream journals, and outlined how it could be done.
I've recently been made aware of some pretty good reasons for SIAI to not publish in mainstream journals, so here they are: