Hi,
We recently observed ~15% performance regression with dbt2 from PG 9.3. We narrowed down on testing master between 9.2 cut and 9.3 cut. It seems that 0ac5ad5134f2769ccbaefec73844f8504c4d6182 is the culprit commit. We did several runs and perf profiling comparing it against its parent (f925c79b9f36c54b67053ade5ad225a75b8dc803). Also tested a 12/16 commit on the master (3b97e6823b949624afdc3ce4c92b29a80429715f) once, it performed similarly as 0ac.. Regards, Dong Results: f92: 53k-56k'ish notpm 0ac: 47k-48k'ish notpm Server SUT: HP ML350 G6 Two Xeon E5520 (4c/p, 8 cores total, hyper-threading disabled) 12GB DRAM HP P410i RAID controller (256MB battery-backed cache) - three 10k-rpm SAS: / - three 10k-rpm SAS: /pgdata - one 15k-rpm SAS: /pgxlog - ext4 (rw,relatime,data=ordered) on all mounts. Fedora 19 (3.11.10-200.fc19.x86_64) max_connections=100 shared_buffers=8192MB effective_cache_size=10GB temp_buffers=8186kB work_mem=4093kB maintenance_work_mem=399MB wal_buffers=-1 checkpoint_segments=300 checkpoint_completion_target=0.9 logging_collector=on log_timezone=UTC datestyle='iso, mdy' lc_messages=C lc_monetary=C lc_numeric=C lc_time=C default_text_search_config='pg_catalog.english' listen_addresses='*' log_destination=csvlog log_directory=pg_log log_filename='pg-%a' log_rotation_age=1440 log_truncate_on_rotation=on Client and workload: Dell 390. Two core. Direct connect with the Server SUT. dbt2 (ToT) 40 warehouse 8 terminals, 8 connections zero think/key time 12-min run Flat perf profiles of two such runs look like: f92: Samples: 608K of event 'cycles', Event count (approx.): 6679607097416 + 4.04% postgres postgres [.] heap_hot_search_buffer + 3.63% postgres postgres [.] AllocSetAlloc + 3.37% postgres postgres [.] hash_search_with_hash_value + 2.85% postgres postgres [.] _bt_compare + 2.67% postgres postgres [.] SearchCatCache + 2.46% postgres postgres [.] LWLockAcquire + 2.16% postgres postgres [.] XLogInsert + 2.08% postgres postgres [.] PinBuffer + 1.32% postgres postgres [.] ExecInitExpr + 1.31% postgres libc-2.17.so [.] _int_malloc + 1.29% swapper [kernel.kallsyms] [k] intel_idle + 1.23% postgres postgres [.] MemoryContextAllocZeroAligned + 1.13% postgres postgres [.] heap_page_prune_opt + 1.06% postgres libc-2.17.so [.] __memcpy_ssse3_back + 1.02% postgres postgres [.] LWLockRelease + 0.94% postgres postgres [.] copyObject + 0.89% postgres postgres [.] fmgr_info_cxt_security + 0.82% postgres postgres [.] _bt_checkkeys + 0.81% postgres postgres [.] hash_any + 0.73% postgres postgres [.] FunctionCall2Coll + 0.69% postgres libc-2.17.so [.] __strncpy_sse2_unaligned + 0.67% postgres postgres [.] HeapTupleSatisfiesMVCC + 0.66% postgres postgres [.] MemoryContextAlloc + 0.65% postgres postgres [.] expression_tree_walker + 0.59% postgres postgres [.] check_stack_depth + 0.57% postgres libc-2.17.so [.] __printf_fp + 0.56% postgres libc-2.17.so [.] _int_free + 0.52% postgres postgres [.] base_yyparse 0ac: Samples: 706K of event 'cycles', Event count (approx.): 6690377376522 + 3.82% postgres postgres [.] GetMultiXactIdMembers + 3.43% postgres postgres [.] LWLockAcquire + 3.31% postgres postgres [.] hash_search_with_hash_value + 3.09% postgres postgres [.] heap_hot_search_buffer + 3.00% postgres postgres [.] AllocSetAlloc + 2.56% postgres postgres [.] _bt_compare + 2.19% postgres postgres [.] PinBuffer + 2.13% postgres postgres [.] SearchCatCache + 1.99% postgres postgres [.] XLogInsert + 1.53% postgres postgres [.] LWLockRelease + 1.32% postgres postgres [.] HeapTupleSatisfiesMVCC + 1.31% swapper [kernel.kallsyms] [k] intel_idle + 1.25% postgres postgres [.] ExecInitExpr + 1.21% postgres postgres [.] heap_page_prune_opt + 1.03% postgres libc-2.17.so [.] _int_malloc + 0.99% postgres postgres [.] MemoryContextAllocZeroAligned + 0.97% postgres libc-2.17.so [.] __memcpy_ssse3_back + 0.89% postgres postgres [.] heapgetpage + 0.76% postgres postgres [.] hash_any + 0.69% postgres postgres [.] fmgr_info_cxt_security + 0.69% postgres postgres [.] _bt_checkkeys + 0.65% postgres postgres [.] copyObject + 0.63% postgres postgres [.] FunctionCall2Coll + 0.58% postgres postgres [.] MemoryContextAlloc + 0.56% postgres libc-2.17.so [.] __strncpy_sse2_unaligned + 0.53% postgres postgres [.] XidInMVCCSnapshot + 0.52% postgres [kernel.kallsyms] [k] copy_user_generic_string + 0.51% postgres libc-2.17.so [.] __printf_fp + 0.50% postgres libc-2.17.so [.] vfprintf P.S. The regression was found when testing vPostgres performance. During the narrow-down, we found the commit 005f583ba4e6d4d19b62959ef8e70a3da4d188a5 (on REL9_2_STABLE) improves performance (versus its parent commit) by 5-10%. Perhaps serve an extra data point for the patch.
