Olaf,
>Well, it's no real surprise, VFP processes/threads can work with multicore cpu's. Calvian Hsia showed a while ago, that we even can do multithreading. This is all a job of the OS and as VFP runs on windows it can do that. Sorry, if I am not that enthusiastic about that. I wonder if rushmore can make use of multiple cores, it would be interesting to make a single SQL go faster, especially if the data is local or even cahed in memory in a cursor.

Just a hunch: those two cases would be the only cases having a chance to profit. And there are chances of thrashing / worse performance even then (esp. in case 1). OTOH I can imagine a real speedup in indexing on very complex key expressions (similar to the OP question) and in very complex SQL statements building different kinds of intermediate files - but only if working on different spindles for each intermediate branch. Not a scenario often seen in vfp, although I sometimes get such queries I could structure ths way. But for everyday use maximising one core and having all the other stuff running on core 2 is sufficient IMHO for the vfp-typical patterns.

Even for server tasks there are examples people think process threading is sufficient - see Postgres.

again: not hard won data, mostly hunches and twitches, but probably better than astrology.

regards

thomas