I think the previous answers are now outdated in one respect. My understanding is that in Peter Singer's 2014 book The Point of View of the Universe (coauthored with Katarzyna de Lazari-Radek), Singer changed his position from preference utilitarianism to hedonistic utilitarianism. Although I haven't read the book, I have looked at a review of it in Notre Dame Philosophical Reviews by Bart Schultz. Here is a relevant passage of the review (Note--the first part of the quoted sentence is not directly relevant to the question, but just gives context):
Restoring Sidgwick to his rightful place of philosophical honor and cogently defending his central positions are obviously no small tasks, but the authors are remarkably successful in pulling them off, in a defense that, in the case of Singer at least, means candidly acknowledging that previous defenses of [R.M.] Hare's universal prescriptivism and of a desire or preference satisfaction theory of the good were not in the end advances on the hedonistic utilitarianism set out by Sidgwick.
I believe that the phrase "previous defenses of Hare's universal prescriptivism and of a desire or preference satisfaction theory" refers to Singer's previous defenses of those views (which he now rejects).
As the previous answers noted, 'act' in 'act utilitarianism' and 'preference' in 'preference utilitarianism' are not parameters along the same dimension. The OP appears to have been confused on this point. Versions of utilitarianism differ along a few dimensions.
Along one dimension there is debate about what human (or animal) well-being consists in. Positions along this dimension include but are not limited to (I'll stop saying this) hedonistic utilitarianism, preference-satisfaction (or desire-satisfaction) utilitarianism, and eudaimonian (I believe this is similar to so-called 'objective list theories') utilitarianism. Of course, there are still further debates within these positions.
Along a second dimension there a debate about the structure of utilitarian theories. Positions along this dimension include act-utilitarianism, rule-utilitarianism, as well as Donald Regan's so-called cooperative utilitarianism.
Perhaps a third dimension is the scalar versus binary dimension of the moral status of an action (although I know less about this). I think scalar utilitarianism says that some acts are better and worse than others (perhaps better or worse to certain degrees) but that there is no cuttoff between a right act and a wrong act or between a permissible act and an impermissible--indeed this view rejects the notions of right, wrong, permissible and impermissible. By contrast, binary utilitarianism finds these distinctions to be important in some respect.
A fourth dimension arises in settings where the population size is not fixed. Positions here include total utilitarianism, average utilitarianism, etc.
(There are probably other dimensions too but I'm getting away from the question now. Also some of these dimensions apply also to non-utilitarian ethical theories.)