dont forget parallel is much larger than cuda, gpu's just happen to be largely parallelizable chips. If your really basic, play around with java and get a feeling for threads and synchronous, maybe write a map/reduce program or 2 and run it on aws.