You can see benefits in two different ways. One is as you suggested is to put the partitions on different filegroups (which need to be on different physical drives for it to make sense), the other is to just present less rows for the optimizer to deal with when the where clause includes the partitioning column. Think about the savings in disk IO if you scan only the partition compared to the full table.

There's also value in partitioning as part of a load/archive strategy, using some of the advanced parts of partitioning to swap data in/out as needed very quickly.