Analysis: The estimated value of the lossy_pages is way higher than its actual value and reason is that the total_pages calculated by the "Mackert and Lohman formula" is not correct.
I think the problem might be that the total_pages includes cache effects and rescans. For bitmap entries we should use something like relation pages * selectivity.
Meanwhile, I ran TPC-H briefly with the v3 patch: scale factor 25, 2 workers, SSD storage.
It shows significant improvement on 4MB work_mem and no change on 2GB. Here are the results (execution time in seconds, average of 5 runs): work_mem 4MB 2GB Query master patch master patch 4 179 174 168 167 5 190 168 155 156 6 280 87 227 229 8 197 114 179 172 10 269 227 190 192 14 110 108 106 105 -- Alexander Kuzmenkov Postgres Professional: http://www.postgrespro.com The Russian Postgres Company -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
