I have started running R jobs on a high performance compute cluster, but inevitably find that I am just performing array jobs which are ultimately not taking advantage of true parallelism, although they do speed up the experiment.
I would like to start writing some parallel code in R where parallelized functions are available and wondered what people's experiences of this are and what packages you would recommend ?
Thanks everyone for the answers. In the end I'm going for Revolution-R as the best answer mainly because it is a big step forward for learning and programming in parallel with R. It includes at install the domC and multicore packages that Chris mentioned and achieves some pretty impressive speed ups on processes like matrix multiplication. Watch out for it hammering the CPUs though.