Citing something that is not published will prevent reviewers from doing their job, so it's a big no-no if you want to improve your chances of being accepted. The best way to go is to be patient and submit to B next year, having had the chance to improve using the reviews from A.
If this is not at all possible, you may be able publish A right now as a technical report from your lab/department an cite it as such. You'll have to check the guidelines of both conferences, namely if A accepts material previously published as a TR (in CS at least this is very common) and if B accepts citing TRs (usually also true in CS as long as it is easily available online).
Most importantly, when citing from a non-refereed source like a TR, you have to be very prudent in the way you characterize the work. Remember that it was only accepted in your department as an interesting document, not properly validated using the scientific contribution standards of your community. If I read a claim that something was "proven", or "shown", or "demonstrated" by a tech report, I'll probably reject the paper.
In any case, do not just cite A unless it is tangential (and in that case, why cite it all?). If it's actually important, give it an overview in your B submission, sufficient for a reviewer to keep on reading.