PostgreSQL 8.4 introduced an automatic parallel restore that lets you allocate multiple CPU cores on the server to their own dedicated loading processes. In addition to loading data into more than one table at once, running the parallel pg_restore will even usefully run multiple index builds in parallel. You just need to tell it how many jobs it can run at once, and once it has finished the basic table setup, it will try to keep that many things active at all times. You do have to use the custom database dump in just the right format for the parallel restore to use it.
COPY in PostgreSQL requires a pretty large amount of CPU to run, and it's actually quite easy to discover a single core is maxed out running it without keeping ...
