I noticed that setting temp_store to DISK surprisingly improved the performance in some cases comparing to the same operation when it was set to MEMORY. Looking at the numbers I noticed that in case of MEMORY one operation that actually does select spread over a large table always led to big data read although in case of DISK it appeared that the table completely went to cache after the first iteration.
So do memory temp_store and general caching share the same memory pool? I prefer to use MEMORY temp_store for some reasons. If they share the memory, what is the strategy should be on my side to choose the right cache size if the amount of data that I plan to transfer through temp tables is unpredictable. Should I always DELETE or DROP temp table data asap in order to increase the probability that the disposed memory would be used for caching in the next SELECT? Max _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users