from pg_stast_get_blocks_fetched i can see that both queries need almost the same number of disk fetches which is quite reasonable ( the index is unclustered).
But as you can see there is a great variation between query runtimes.Can someone explain this differnce?