There was a thread in early December that had some information I found useful as someone unfamiliar with HTCondor. Here is a particularly useful response from Gary Jackson, specific to dedicated clusters (as you have):
He points out that if your jobs are typically tightly-coupled parallel jobs, then PBS may be better. Whereas if your jobs are mostly "embarrassingly parallel" or high-throughput jobs, then HTCondor may be superior.Â
Of particular interest to you, he also said: "There's no reason you can't use both HTCondor and a purpose-builtÂparallel batch scheduler at the same time." So, it's not trivial, but you don't have to eliminate PBS to try out HTCondor.
--Steve