# The benchmark cases of the optimizations: one row per case, its SQL in cases/ID.sql. `bench/bench cases` times each against PostgreSQL and the other engines; # id: the case (bench/cases/ID.sql, the layout of bench/sql/store/queries: `--@ pg` the statement of every lane, `--@ sift` / `--@ duck` replace it, `--@ setup` DDL after the index) # changelog: a phrase of the CHANGELOG line the case measures (the gate matches it; it must stay in CHANGELOG.md) # dataset: orders (the store table of bench/sql/store/gen.sql with two derived columns, segment_id with 6,313 values and order_date); docs (the text table of bench/sql/docs/gen.sql, a BM25 index built in the case's setup, the PostgreSQL lane is full text search: tsvector and GIN) # index: the columns of the sift_col index (`default`: every column of the store questions plus the derived ones; `default+title` adds the columns named after the +) and its WITH options; `-` for a case that makes its own tables # plan: a regular expression the EXPLAIN of the sift lane must match (a load case: the output of its `--@ check`); a case that falls back to the normal plan fails # baselines: engines besides sift and native (PostgreSQL with sift.enable_agg = off) that can answer it: pg_duckdb (the others: bench/engines.tsv, steps/store_engine.py) # variants: name:SET a = b;SET c = d, further sift lanes with settings (the optimization turned off, or a lane without a pre-aggregate) # metric: the number the README table shows: ms (warm median), wal_bytes (per row) or, for a case that churns an index, index_mb (its size at the end) # block: the generated block of README.md / bench/out/BENCHMARKS.md the case feeds () # headline: yes puts the case in the README's table of application-shaped cases # peers: `-`, or the engines that answer the case in another results file, as FILE:ARM (generic:clickhouse is the arm clickhouse of bench/results/generic.tsv); `cases coverage` wants a measured row of each topn_page_totals a page of rows with the total of its set orders default Top-N: created_at DESC.*Pushed Aggregates: count$\*$, sum native,pg_duckdb - ms faster33 yes - topn_coalesce `ORDER BY coalesce( orders default Top-N: coalesce$customer_id, 0$ DESC native,pg_duckdb - ms faster33 yes - topn_lower `lower()` and `upper()` orders default Top-N: lower$category$ native,pg_duckdb - ms faster33 no - key_presence_distinct `key_columns` (`sift_col` reloption orders default WITH (key_columns = 'segment_id') count$DISTINCT segment_id$.*[1-9][0-9]* from pre-aggregates native,pg_duckdb scan:sift.agg_preagg = off ms faster33 yes - cube_minmax_groupby `cube_minmax` and `superunit` are on by default orders default [1-9][0-9]* from pre-aggregates native,pg_duckdb scan:sift.agg_preagg = off ms faster33 no - calendar_month `calendar_columns` of a `sift_col` index defaults to every date orders default date_trunc$'month'.*[1-9][0-9]* from pre-aggregates native,pg_duckdb scan:sift.agg_preagg = off ms faster33 no - numeric_group2_bare - orders default Custom Scan \(SiftAgg$ native - ms,minstr faster33 no - numeric_group2_scaled - orders default Custom Scan $SiftAgg$ native - ms,minstr faster33 no - dict_distinct_first_statement - orders default Index Scan using orders_col|Custom Scan $SiftAgg$ native no_ring:sift.col_dict_shared_cache_kb = 0,warm_backend:sift.enable_agg = on ms,minstr faster33 no - table_load_copy - orders - ^columns$ native - ms,cpu_ms,wal_bytes,minstr faster33 no - table_load_indexes - orders - ^[0-9]+$ native - ms,cpu_ms,wal_bytes,minstr faster33 no - table_load_vacuum - orders - ^[0-9]+$ native - ms,cpu_ms,wal_bytes,minstr faster33 no - table_load_analyze - orders - ^[0-9]+$ native - ms,cpu_ms,wal_bytes,minstr faster33 no - bulk_copy_wal `sift.table_bulk_groups` (on by default since 3.3 orders - ^columns$ native bulk_off:sift.table_bulk_groups = off wal_bytes,ms,cpu_ms faster33 yes - cluster_visibility_map the planner charges the heap check of the rows on pages that are not all-visible orders default Custom Scan $SiftAgg$ native,pg_duckdb - ms faster33 no - topn_text_key `ORDER BY` a text, varchar, uuid, bytea or inet column `LIMIT n` is a top-N of the column scan orders default+title Top-N: title native,pg_duckdb - ms faster33 no - topn_uuid_key reads a uuid or bytea sort key as the codes of the dictionary the segment tenants orders default Top-N: ref.*stored dictionary native,pg_duckdb - ms faster33 no - topn_text_code compares the rows of a unit with the bar by the order code of the dictionary of their segment orders default Top-N: sku native,pg_duckdb - ms faster33 no - topn_sorted_runs `order_columns` (`sift_col` reloption, default none orders default WITH (order_columns = 'customer_id') [1-9][0-9]* units read by sorted run native,pg_duckdb runs_off:sift.topn_runs = off ms faster33 yes - topn_sorted_runs_btree (a btree needs 0.35) orders default WITH (order_columns = 'customer_id') fallback:Index Only Scan using orders_customer native forced:enable_indexscan = off;enable_indexonlyscan = off;enable_bitmapscan = off ms faster33 no - topn_zone_keys bounds the units by the zones of the sort keys after the first too orders default Top-N: status, customer_id native,pg_duckdb zone_off:sift.topn_zone_keys = off ms faster33 no - topn_big_offset with `n + m` past 100,000 orders default Top-N: amount native,pg_duckdb - ms faster33 no - topn_many_keys up to 8 sort columns in the heap of the top-N scan orders default Top-N: customer_id, country, category, status, quantity native,pg_duckdb - ms faster33 no - count_distinct_hist `count(DISTINCT col)` of a 32-bit or text column the cube tenants a histogram of orders default Custom Scan $SiftAgg$ native,pg_duckdb scan:sift.agg_preagg = off ms faster33 no - group_numeric_key a group key that is an `int8` or a `numeric(p,s)` column orders default Custom Scan $SiftAgg$ native,pg_duckdb scan:sift.agg_preagg = off ms faster33 no - rollup_dims `rollup_dims` (`sift_col` reloption, default off orders default WITH (rollup_columns = 'country, category: amount') Custom Scan $SiftAgg$ native,pg_duckdb scan:sift.agg_preagg = off ms faster33 no - dict_equality finds its dictionary codes without testing every string orders default+title Custom Scan $SiftAgg$ native,pg_duckdb - ms faster33 no - dict_prefix a byte-order permutation of the dictionary orders default+title Custom Scan $SiftAgg$ native,pg_duckdb - ms faster33 no - partitions_scan the planner prices the opening of the partitions of a partitioned table, which grows faster than their number orders default Custom Scan $SiftAgg$ native - ms faster33 no - partitions_many 3,200 takes it too orders default fallback:Append|Seq Scan native - ms faster33 no - topn_join_filter a top-N over an inner join to a table that only filters events - Top-N native - ms faster33 no - spine_array_key takes each element of an integer, date or text type that equals a column of the joined table as a hash key events - Custom Scan $SiftAgg$ native - ms faster33 no - lateral_first_row a `LIMIT 1` child with an `ORDER BY` events - Custom Scan $SiftAgg$ native - ms faster33 yes - lateral_dim_test a `LIMIT 1` child that tests its row besides the unique key events - Custom Scan $SiftAgg$ native - ms faster33 no - cte_two_readers a common table expression that two aggregates read is planned inline events - Custom Scan $SiftAgg$ native - ms faster33 yes - filter_few_rows when the filters of the source keep under a 32nd of a unit's rows events - Custom Scan $SiftAgg$ native - ms faster33 no - list_elements the backend keeps the decoded list lengths of a unit of an immutable segment events - Custom Scan $SiftAgg$ native - ms faster33 yes - topn_sorted_runs_filter a top-N scan with `order_columns` reads the sorted run of a unit also when a `WHERE` leaves few of its rows orders default WITH (order_columns = 'customer_id') [1-9][0-9]* units read by sorted run native,pg_duckdb runs_off:sift.topn_runs = off ms faster33 no - superunit_total the merged sums and cube of every group of that many consecutive units of a segment are stored once more orders default Custom Scan $SiftAgg$ native,pg_duckdb scan:sift.agg_preagg = off ms faster33 no - numeric_sum a sum over a `numeric` column without a precision was priced a tenth of the parallel sequential scan orders default Custom Scan $SiftAgg$ native - ms faster33 no - topn_six_keys up to 8 sort columns in the heap of the top-N scan orders default Top-N: customer_id, country, category, status, quantity, amount native - ms faster33 no - partitions_tiny A query over 5,000 partitions of 6 rows each orders default fallback:Seq Scan native - ms faster33 no - dirty_share_few A table with a few percent of dirty documents keeps the scan events - Custom Scan $SiftAgg$ native - ms faster33 no - dirty_share_half one with a third or more keeps the normal plan, which is cheaper there events - fallback:Seq Scan native - ms faster33 no - cte_series_arms Three `UNION ALL` arms over the elements of a jsonb array, counted `DISTINCT` above a spine, 100,000 rows events - Custom Scan \(SiftAgg native - ms faster33 yes - keys_in_wide_50k is tested by a bit table instead of a hash table keys - fallback:Bitmap Heap Scan on events_w native - ms faster33 no - keys_in_wide_500k the zone test of a set travels to the column layer up to twice the units of the scan keys - Custom Scan \(SiftAgg native - ms faster33 no - keys_in_narrow_500k the selectivity of the list is the one the planner kept for the condition keys - Custom Scan \(SiftAgg native - ms faster33 no - keys_subq_500k the rows a lookup join loads (`IN (SELECT ..)`, `EXISTS`, a join) are read by the heap scan keys - Custom Scan \(SiftAgg native - ms faster33 no - keys_exists_500k when an equality of the joined table keeps a third of it or more keys - Custom Scan \(SiftAgg native - ms faster33 no - keys_join_500k as an `IN` list of the table already was keys - Custom Scan \(SiftAgg native - ms faster33 no - keys_semi_load_first a table that a semi, anti or inner join by an int4 column answers from a byte table of its keys keys - Custom Scan \(SiftAgg native first:sift.agg_lookup_cache_kb = 0;sift.agg_lookup_shared_kb = 0 ms faster33 no - keys_in_few_256 a list of 256 keys over a range of a million values added about 2 ms to a scan that read two units keys - fallback:Bitmap Heap Scan on events_u native - ms faster33 no - count_distinct_spill a group table that spills writes each `count(DISTINCT)` set orders default Custom Scan \(SiftAgg native - ms faster33 no - group_spill_passes a group table of many times hash memory is read in passes of the groups, not spilled orders default Custom Scan \(SiftAgg native - ms faster33 no - text_key_filtered a near-unique text key under a filter made a table of ids as large as the rows read orders default+title Group Keys: title native - ms faster33 no - topn_numeric_wide takes the column as a sort key, as it took a numeric of up to 18 digits orders default Top-N: price native,pg_duckdb - ms faster33 no - topn_enum_key takes an enum column as a sort key, in the order of its type orders default Top-N: lvl native,pg_duckdb - ms faster33 no - topn_interval_key takes an interval column as a sort key orders default Top-N: lead native,pg_duckdb - ms faster33 no - count_distinct_wide_keys starts as a hash set, not a bitmap orders default Custom Scan \(SiftAgg native - ms faster33 no - grouping_sets_report grouping_sets_report orders default Pushed Aggregates native,pg_duckdb scan:sift.agg_preagg = off ms faster33 yes - ranked_top_per_group ranked_top_per_group orders default Pushed Aggregates native,pg_duckdb scan:sift.agg_preagg = off ms faster33 yes - window_json_report window_json_report orders default Pushed Aggregates native,pg_duckdb scan:sift.agg_preagg = off ms faster33 yes - two_level_spend two_level_spend orders default Pushed Aggregates native,pg_duckdb - ms faster33 yes - group_by_id_top group_by_id_top orders default+id Pushed Aggregates native,pg_duckdb - ms faster33 yes - count_distinct_unique count_distinct_unique orders default+id Pushed Aggregates native,pg_duckdb - ms faster33 yes - semi_exists_self semi_exists_self orders default Pushed Aggregates native,pg_duckdb - ms faster33 yes - anti_not_in_self anti_not_in_self orders default Pushed Aggregates native,pg_duckdb - ms faster33 yes - keyset_page keyset_page orders default+id fallback:Index Scan native forced:enable_indexscan = off;enable_indexonlyscan = off;enable_bitmapscan = off ms faster33 yes - point_agg_fk point_agg_fk orders default WITH (point_tid_columns = 'customer_id') Pushed Aggregates native,pg_duckdb - ms faster33 yes - point_agg_id point_agg_id orders default+id WITH (point_tid_columns = 'id') fallback:Index Scan using orders_pkey native forced:enable_indexscan = off;enable_indexonlyscan = off;enable_bitmapscan = off ms faster33 yes - keycut_customer_agg `key_cut` (`sift_col` reloption, default none orders default WITH (key_cut = 'customer_id') [1-9][0-9]* pruned native,pg_duckdb - ms faster33 yes - keycut_customer_btree `key_cut` (`sift_col` reloption, default none orders default WITH (key_cut = 'customer_id') fallback:Bitmap Heap Scan on orders native forced:enable_indexscan = off;enable_indexonlyscan = off;enable_bitmapscan = off ms faster33 no - keycut_customer_range `key_cut` (`sift_col` reloption, default none orders default WITH (key_cut = 'customer_id') [1-9][0-9]* pruned native - ms faster33 no - jsonb_each_member jsonb_each_member events - Pushed Aggregates native - ms faster33 yes - softdel_live_groupby `x IS NULL` and `x IS NOT NULL` are tests of the NULL count of a unit's zone orders default [1-9][0-9]* from pre-aggregates native,pg_duckdb scan:sift.agg_preagg = off ms faster33 yes - softdel_deleted_list `x IS NULL` and `x IS NOT NULL` are tests of the NULL count of a unit's zone orders default [1-9][0-9]* pruned native,pg_duckdb - ms faster33 no - point_in_100 an `IN` list of up to 65,536 keys is looked up in the point directory orders default orders_col native,pg_duckdb - ms faster33 yes - point_in_1000 an `IN` list of up to 65,536 keys is looked up in the point directory orders default orders_col native,pg_duckdb - ms faster33 no - point_exact_slots has `ceil(keys * 4 / 3)` slots instead of the next power of two orders default orders_col native,pg_duckdb - ms faster33 no - bigdict_point_probes dictionary the cache cannot hold is no longer parsed again for every test orders default orders_col native - ms faster33 no - topn_run_probes reads the sort key of a unit's rows only for the rows it keeps orders default WITH (order_columns = 'customer_id') [1-9][0-9]* units read by sorted run native - ms faster33 no - point_key_range tests the zones and summaries of the other predicates before it asks the point directory orders default WITH (point_tid_columns = 'country') Bitmap Heap Scan on orders|Index Scan using orders_col native - ms faster33 no - point_prefix_count priced by the planner as the lookups of the keys the dictionary names orders default ^Aggregate native - ms faster33 no - tbl_update_keys asks the table which of the page's versions are dead orders - tombstones native - ms faster33 no - point_keys_agg asks the point directory (`point_columns`, `point_tid_columns`) for the rows of a short `IN` list orders default Custom Scan $SiftAgg$ native - ms faster33 no - percentile_select finds the percentiles of a group by a quickselect over its histogram orders default Custom Scan $SiftAgg$ native - ms faster33 no - percentile_disc_int finds the percentile_disc of a decimal, timestamp or bigint on the 64-bit values of its histogram orders default Custom Scan $SiftAgg$ native - ms faster33 no - mode_histogram on the histogram of the values of a group, as it does `percentile_disc` orders default Custom Scan $SiftAgg$ native - ms faster33 no - group_topn are dropped before they become tuples: an `ORDER BY orders default [1-9][0-9]* dropped by top-N native - ms faster33 no - two_level_inner_max the largest value of a float key of each row averaged by group is one two-level scan orders default Custom Scan $SiftAgg$ native - ms faster33 no - distinct_rows_sets distinct rows of a grouping set over a preserved join orders default Pushed Aggregates: count, count, count, count$DISTINCT (o\.)?customer_id$ native - ms faster33 no - placeholder_leaf_args the cast of a map value is the float leaf orders default Pushed Aggregates: count$\*$, count$\*$, count$\*$, avg$metrics_a\.\*\.value$ native - ms faster33 no - sets_lookup_keys resolves its rows in a loop of the column's kind orders default Group Keys: category: l\.label, country: r\.region.*grouping set native - ms faster33 no - lateral_group_decor `x JOIN LATERAL (SELECT g, sum(f.a) FROM f WHERE f.k = x.k GROUP BY g [HAVING ..]) ON true` orders default Custom Scan $SiftAgg$ native - ms faster33 no - lateral_comma_agg `FROM x, LATERAL (SELECT sum(f.a) FROM f WHERE f.k = x.k)` (the comma form orders default Custom Scan $SiftAgg$ native - ms faster33 no - exists_decor `EXISTS (SELECT .. FROM f WHERE f.k = x.k AND ..)` and `NOT EXISTS` where PostgreSQL leaves a sub-plan orders default Custom Scan $SiftAgg$ native - ms faster33 no - index_pick a table with several `sift_col` indexes is planned through the cheapest index orders default Index: orders_kc native - ms faster33 no - window_first_rows the first N rows of each group asked with a window function orders default+id fallback:Bitmap Heap Scan on orders native - ms faster33 no - soft_delete_exceptions `WHERE stamp IS NULL` is answered from the pre-aggregates orders default Filters: removed_at IS NULL native - ms faster33 no - approx_distinct_sketch `hll_columns` (default none; orders default+title WITH (hll_columns = 'title, customer_id') approx_count_distinct.*\b0 without the sketch native,pg_duckdb scan:sift.agg_preagg = off ms faster33 no - count_distinct_highcard a hundred thousand groups of a few rows orders default Custom Scan $SiftAgg native - ms faster33 no - seq_cost_heap_check is never priced above the random fetch of the same pages orders default Custom Scan \(SiftAgg$ native - ms faster33 no - lookup_table_whole is priced with the pages PostgreSQL's own sequential scan of it reads orders default fallback:Seq Scan on lookup native - ms faster33 no - lookup_columns is loaded from the leaves of a `sift_col` index of the table orders default Custom Scan $SiftAgg$ native - ms faster33 no - topn_from_units takes the rows of the result from the units that gave them orders default Global Order: true native,pg_duckdb - ms faster33 no - cte_lookup_par a scan whose lookup joins read a CTE, a sub-select or a function orders default Pushed Aggregates native,pg_duckdb one_process:max_parallel_workers_per_gather = 0 ms faster33 no - point_dense_keys only while the keys are expected to have fewer rows than 4,096 orders default Custom Scan $SiftAgg$ native,pg_duckdb - ms faster33 no - unit_head_scan reads the first 4 kB of a unit (its row directory and chunk table) instead of 24 kB orders default Custom Scan $SiftAgg$ native,pg_duckdb - ms faster33 no - tpch_pricing_summary tpch_q1 tpch - Pushed Aggregates native - ms faster33 no - tpch_forecast_revenue tpch_q6 tpch - Pushed Aggregates native - ms faster33 no - tpch_shipping_priority tpch_q3 tpch - Custom Scan $SiftAgg$ native - ms faster33 no - tpch_returned_items tpch_q10 tpch - Custom Scan $SiftAgg$ native - ms faster33 no - tpch_shipping_modes tpch_q12 tpch - Custom Scan $SiftAgg$ native - ms faster33 no - hits_count hits_count hits - Pushed Aggregates native - ms faster33 no - hits_filter_count hits_filter_count hits - Pushed Aggregates native - ms faster33 no - hits_sum_avg hits_sum_avg hits - Pushed Aggregates native - ms faster33 no - hits_distinct_users hits_distinct_users hits - Pushed Aggregates native - ms faster33 no - hits_distinct_phrases hits_distinct_phrases hits - Pushed Aggregates native - ms faster33 no - hits_region_users hits_region_users hits - Pushed Aggregates native - ms faster33 no - hits_top_phrases hits_top_phrases hits - Pushed Aggregates native - ms faster33 no - hits_like_count hits_like_count hits - Pushed Aggregates native - ms faster33 no - hits_like_group hits_like_group hits - Pushed Aggregates native - ms faster33 no - hits_like_not `NOT LIKE '%part%'` on a `text` or `varchar` column hits - Pushed Aggregates native - ms faster33 no - hits_like_order hits_like_order hits - Top-N: event_time native - ms faster33 no - hits_ilike_count `ILIKE '%part%'` of an ASCII part hits - Pushed Aggregates native - ms faster33 no - hits_regex_prefilter are no longer run on every string of a unit's dictionary hits - Pushed Aggregates native - minstr faster33 no - hits_order_time hits_order_time hits - Incremental Sort.*Top-N: event_time native - ms faster33 no - ts_range_by_day ts_range_by_day tseries - Pushed Aggregates native - ms faster33 no - ts_hour_metric ts_hour_metric tseries - Pushed Aggregates native - ms faster33 no - ts_top_devices ts_top_devices tseries - Pushed Aggregates native - ms faster33 no - ts_latest_device ts_latest_device tseries - fallback:Index Scan native - ms faster33 no - oltp_insert_index oltp_insert_index orders - ^[0-9]+$ native - ms,cpu_ms faster33 no - oltp_update_index oltp_update_index orders - ^[0-9]+$ native - ms,cpu_ms faster33 no - run_window a range on a leaf with a sorted run (`order_columns`), or an equality on it orders default [1-9][0-9]* filtered by sorted run native,pg_duckdb no_run:search_path = run_window_plain ms faster33 no - range_wide_bytes plain 3, 5, 6 or 7-byte values orders default Custom Scan $SiftAgg$ native,pg_duckdb - ms faster33 no - two_level_cell_conditions the conditions of the outer aggregates of a two-level scan read the value of each cell once orders default Pushed Aggregates native - ms faster33 no - dict_small_request the ids of a dictionary's entries (the key of a group by a text column orders default Group Keys: tag native - ms faster33 no - dict_first_scan made when a tuple first reads the entry orders default+id+title fallback:Index Scan using orders_pkey native,pg_duckdb warm_forced:sift.agg_force = on ms faster33 no - verdict_shared_report the choice between a shared CTE and its inline plan (several planner runs of the statement) is kept in the backend events - Custom Scan $SiftAgg$ native no_shared:sift.verdict_cache_kb = 0,warm_backend:sift.enable_agg = on ms faster33 no - set_pruned_request are checked in one loop for each width of the list orders default Pushed Aggregates native - ms faster33 no - grouping_sets_many_keys prepares the key ids of a chunk once for all the sets that read its rows orders default Custom Scan $SiftAgg$ native,pg_duckdb - ms faster33 no - topn_walk_order_kept keeps the order of that walk in the backend orders default+id Top-N: amount DESC, id native warm_backend:sift.enable_agg = on ms faster33 no - lateral_func_avg is read by the scan as the grouping of the elements by the key of `x` events - Custom Scan $SiftAgg$ native - ms faster33 no - count_distinct_lookup_pairs the slots of the (group, value) pairs of a joined `count(DISTINCT)` are fetched ahead orders default Pushed Aggregates: count$DISTINCT d\.cat native - ms faster33 no - point_tid_in_covered is planned on the addresses and the heap, as it is on a query it does not cover orders default Index native - ms faster33 no - lateral_flat_plan plans the flat query first and, when it reads every table that has a column index in a scan, stops there events - Custom Scan \(SiftAgg$ native warm_backend:sift.enable_agg = on ms faster33 no - hits_latest_time a descending scan of a column that follows the table hits - Top-N: event_time DESC native - ms faster33 no - ts_top_value over a `float4` or `float8` key hands the rows of a unit to the heap tseries - Top-N: value DESC native - ms faster33 no - order_array the reloption `order_array_columns` of a `sift_col` index orders default Global Order: true native no_array:search_path = order_array_plain ms faster33 no - hits_order_trail is a top-N of the first keys under PostgreSQL's `Incremental Sort` hits - Incremental Sort.*Top-N: event_time DESC native - ms faster33 no - order_array_range the scan searches each segment's array for the first entry of the range orders default Global Order: true native no_array:search_path = order_array_range_plain ms faster33 no - order_array_text_range the scan finds the rank of the string in each segment by the collation orders default Global Order: true native no_array:search_path = order_array_text_range_plain ms faster33 no - topn_run_range starts at the first entry of the range, found by two binary searches of the run orders default Global Order: true native no_run:search_path = topn_run_range_plain ms faster33 no - order_array_keyset the scan takes the whole groups of rows that share a value of the first key orders default Global Order: true native no_array:search_path = order_array_keyset_plain ms faster33 no - order_array_group the reloption `order_array_group` of a `sift_col` index orders default Global Order: true native no_group:search_path = order_array_group_plain ms faster33 no - topn_offset_exact an OFFSET reads the sorted array of the key and does not confirm the entries before it in the heap orders default Global Order: true native btree:search_path = topn_offset_btree ms faster33 no - topn_numeric_offset an ORDER BY numeric(p, s) LIMIT reads the sorted array of the column, whose key is the number scaled by 10^s orders default Global Order: true native btree:search_path = topn_numeric_btree ms faster33 no - topn_offset_deep an OFFSET past the 100,000 rows a top-N heap holds is walked in the array orders default Global Order: true native btree:search_path = topn_offset_deep_btree ms faster33 no - order_layout_offset the reloption `order_layout` of a `sift_col` index orders default Global Order: true native whole:search_path = c_order_layout_offset__whole,btree:search_path = c_order_layout_offset__btree ms faster33 no - order_layout_range keeps a chunk of an order array compact where that is at least a fifth smaller orders default Global Order: true native whole:search_path = c_order_layout_range__whole,btree:search_path = c_order_layout_range__btree ms faster33 no - group_order_window emits its groups in that order orders default $[1-9][0-9]* order keys$ native - ms faster33 no - group_order_coalesce_key a group key that is itself a `COALESCE` orders default $[1-9][0-9]* order keys$ native - ms faster33 no - group_order_sorted_agg a parent that groups the groups again with an aggregate that sorts orders default $[1-9][0-9]* order keys$ native - ms faster33 no - tpch_shipping_priority_cold tpch_q3_cold tpch - Custom Scan $SiftAgg$ native - ms faster33 no - tpch_returned_items_cold tpch_q10_cold tpch - Custom Scan $SiftAgg$ native - ms faster33 no - tpch_shipping_modes_cold tpch_q12_cold tpch - Custom Scan $SiftAgg$ native - ms faster33 no - dict_blocks_first_statement are decompressed one at a time orders default fallback:Index Scan using orders_pkey native warm_forced:sift.agg_force = on ms faster33 no - point_tid_key_range a key with a range on another column hands the heap only the rows of the range orders default Index Scan|Bitmap Heap Scan native - ms faster33 no - group_many_groups a group by of one integer column with millions of groups takes fewer instructions a row orders default Group Keys: grp native - ms faster33 no - lookup_keyed_probe is loaded by a btree probe for each of those keys while the descents and the pages of the rows they find cost less orders default [1-9][0-9]* index scans native lookup_off:sift.agg_lookup_pushdown = off ms faster33 no - two_level_lookup_arms the arms of a two-level aggregate that read the columns of a lateral `LIMIT 1` lookup (`rm.derived_cat AS category`, no `CASE` around it) are one scan events - Custom Scan $SiftAgg native - ms faster33 no - two_level_join_types the arms of a two-level scan may differ in the type of the join of a lateral `LIMIT 1` child events - Custom Scan \(SiftAgg native - ms faster33 no - cte_shared_series the shared plan of a CTE that several aggregates read is made with `enable_memoize` off events - Custom Scan \(SiftAgg native - ms faster33 no - cte_hoisted_in no longer repeats an aggregating `IN` sub-select of the CTE body in every arm events - sift_hoisted native - ms faster33 no - topn_lateral_per_entity a top-N scan for each key an enclosing query sets tseries - Top-N: ts DESC native - ms faster33 no - elem_member_null are the null test and the count of the typed leaf of the member events - Custom Scan \(SiftAgg$ native - ms faster33 no - topn_scattered_time answers a superblock (and a unit) from the head of its sorted run orders default Custom Scan $SiftAgg$ native - ms faster33 yes - calendar_session_zone `date_trunc`, `extract` and `date()` (`::date`) of a `timestamptz` in the session's time zone are grouped by the step of the zone's clock orders default $per day$ native - ms faster33 no - numeric_image_sum built from the digits of the stored image, not from its text through `numeric_in` orders default Custom Scan $SiftAgg$ native - ms faster33 no - keycut_auto cut its units by a key (`key_cut`, column format minor 17) by default orders default [1-9][0-9]* pruned native - ms faster33 no - sum_product_47 each pair is multiplied in 256 bits orders default Custom Scan $SiftAgg$ native - ms faster33 no - col_scan_default `sift.enable_col_scan` is on by default orders default Index (Only )?Scan using orders_sc_col|Bitmap Heap Scan on orders_sc native - ms faster33 no - cube_wide_dims the byte budget of the cube of a `sift_col` unit counts what the cube holds events - [1-9][0-9]* from pre-aggregates native scan:sift.agg_preagg = off ms faster33 no - few_keys_lookup keeps its btree probes at the price they are planned for events - Custom Scan $SiftAgg$|Index Scan using events_pkey native - ms faster33 no - keycut_or_ranges no longer starts the setup of a scan for each morsel orders default WITH (key_cut = 'customer_id') [1-9][0-9]* pruned native - ms faster33 no - topn_keyset_run reads a unit from the first entry of the range without running the unit's filters orders default Global Order: true native no_run:search_path = topn_keyset_run_plain ms faster33 no - topn_desc_window reads the superblocks of each window of the walk from the window's best end orders default Top-N: created_at DESC native - ms faster33 no - key_table_point a key table per segment (column format minor 22, the reloption `key_table` orders default WITH (key_columns = 'segment_id') Rows: \b0 scanned native,pg_duckdb units:sift.agg_super = off ms faster33 yes - null_count_zone reads the null counts of the zones of the units events - [1-9][0-9]* from pre-aggregates native scan:sift.agg_preagg = off ms faster33 no - calendar_day_filter a comparison of a calendar step of a `timestamp` or `timestamptz` column with a constant orders default Custom Scan $SiftAgg$ native,pg_duckdb scan:sift.agg_preagg = off ms faster33 no - group_float_key by a float key whose values have a short mantissa orders default Custom Scan $SiftAgg$ native,pg_duckdb - ms faster33 no - jsonb_contains_object is answered from the leaves of the paths it names orders default Custom Scan $SiftAgg$ native - ms faster33 no - rolling_window compared with a stable operand orders default Custom Scan $SiftAgg$ native,pg_duckdb scan:sift.agg_preagg = off ms faster33 no - image_equality are tests of the source on the stored image orders default Custom Scan $SiftAgg$ native,pg_duckdb - ms faster33 no - contain_key_cut cut the units by the start of a range column orders (id) Custom Scan $SiftAgg$ native no_cut:search_path = contain_key_cut_off,gist:search_path = contain_key_cut_gist ms faster33 no - contain_range are answered by a `sift_col` index from integer leaves orders (id) Custom Scan $SiftAgg$ native gist:search_path = contain_range_gist ms faster33 no - contain_inet `ip <<= net` 1.4 ms orders (id) Custom Scan $SiftAgg$ native gist:search_path = contain_inet_gist ms faster33 no - calendar_step_filter is evaluated once for each day, hour or minute the rows fall in orders default $per hour$ native,pg_duckdb - ms faster33 no - text_rest_sparse tests its other conditions on the rows that test left orders (id) Custom Scan $SiftAgg$ native - ms faster33 no - sum_int_expr the exact sum of the expression orders default Custom Scan $SiftAgg$ native,pg_duckdb - ms faster33 no - distinct_pairs_sort are sorted as int8 datums orders (id) Custom Scan $SiftAgg$ native - ms faster33 no - distinct_sparse_sets give way to a hash set orders (id) Custom Scan $SiftAgg$ native - ms faster33 no - distinct_pairs_runs sorted by radix as 8-byte keys orders (id) Custom Scan $SiftAgg$ native - ms faster33 no - calendar_periodic the rows of a unit that the prefilter of an equality of a periodic calendar step keeps orders default Custom Scan $SiftAgg$ native,pg_duckdb scan:sift.agg_preagg = off ms faster33 no - numeric_sum_records made from the lanes of each row orders default Custom Scan $SiftAgg$ native - ms faster33 no - contain_rows returns the rows of a containment of a range (`@>` orders (id) Index Scan using ranges_rows_col|Bitmap Index Scan on ranges_rows_col native gist:search_path = contain_rows_gist ms faster33 no - key_lookup_units skips the morsels of units that the zones clear orders (id) Custom Scan $SiftAgg$ native u4008:search_path = key_lookup_u4008,u500:search_path = key_lookup_u500 ms faster33 no - dict_shared_first_statement copied from shared memory orders default Custom Scan $SiftAgg$ native - ms faster33 no - dict_shared_find where it lies in the ring orders default Index Scan using orders_col native - ms faster33 no - key_cut_day_count a day count 25.4 M to 0.7 M orders (id) Custom Scan $SiftAgg$ native btree:search_path = key_cut_day_count_bt ms faster33 no - key_cut_week_count a week 25.6 M to 0.9 M orders (id) Custom Scan $SiftAgg$ native btree:search_path = key_cut_week_count_bt ms faster33 no - key_cut_group_day a group by day 88.7 M to 3.8 M orders (id) Custom Scan $SiftAgg$ native btree:search_path = key_cut_group_day_bt ms faster33 no - key_cut_range_small `@>` on an `int4range` of 500,000 rows orders (id) Custom Scan $SiftAgg$ native gist:search_path = key_cut_range_small_gist ms faster33 no - long_text_unwrap_cheap wrapped with LZ4 instead of zstd where the default formula says so orders default Custom Scan $SiftAgg$ native - minstr faster33 no - dict_network_equal_cheap no longer fits the cache of parsed dictionaries orders default Custom Scan $SiftAgg$ native - minstr faster33 no - int2_sum_const `sum` and `avg` of a smallint column plus, minus or times an integer constant orders default Pushed Aggregates: sum$\(orders\.quantity \+ 1$\).*[1-9][0-9]* from pre-aggregates native - ms faster33 no - hits_regex_count `LIKE '%suffix'`, `starts_with(col, 'abc')` hits - Pushed Aggregates native - ms faster33 no - contain_union a containment that holds in two ways (`&&`, `<<`, `-|-`) prunes the units orders (id) Custom Scan $SiftAgg$ native gist:search_path = contain_union_gist ms faster33 no - lookup_key_list_twice counts the restriction of one column once orders default fallback:Index Scan (using|on) orders_customer_created native - ms faster33 no - flat_lateral_key_lookup stays in the scan when the query also reads another table with a column index only through the key lookups of a CTE events - Custom Scan $SiftAgg$ native - ms faster33 no - keycut_second_window cut into units of 4,096 rows events - [1-9][0-9]* pruned native - ms faster33 no - cte_materialized_notes is planned inline as well as shared events - Custom Scan $SiftAgg$ native - ms faster33 no - groupset_window_order emits its groups in the order the parent sorts by events - Custom Scan $SiftAgg$ native - ms faster33 no - groupset_window_sort emits the order of the Sort above its window events - Custom Scan $SiftAgg$ native - ms faster33 no - unit_tab_first the chunk table of a key-cut unit comes before its directory orders default WITH (key_cut = 'customer_id') Custom Scan $SiftAgg$ native,pg_duckdb - ms faster33 no - point_packed is a packed table of entries by bucket orders default orders_col native,pg_duckdb - ms faster33 no - array_elements_no_stats size a table that was never vacuumed or analyzed from its pages events - Custom Scan $SiftAgg$ native - ms faster33 no - union_arms_window_order emit their groups in the order of the parent, which the parent merges events - Merge Append native - ms faster33 no - per_event_arms the cells of the arms of a two-level grouping events - Custom Scan $SiftAgg$ native - ms faster33 no - scope_sublink_count scope_sublink_count events - Custom Scan \(SiftAgg native - ms faster33 no - page_newest_lateral page_newest_lateral events - Custom Scan \(SiftAgg native - ms faster33 no - page_since_cursor page_since_cursor events - Top-N: updated_at DESC NULLS LAST, id native - ms faster33 no - page_deep_offset page_deep_offset events - Top-N: ts DESC NULLS LAST, id native - ms faster33 no - scope_exists_any_array scope_exists_any_array events - fallback:Seq Scan on dims native - ms faster33 no - search_scope_filter search_scope_filter events - Custom Scan \(SiftScan native - ms faster33 no - flags_or_null_range flags_or_null_range events - fallback:Bitmap Heap Scan native - ms faster33 no - time_event_buckets a `GROUP BY` over a `UNION ALL` of selects (a queue's depth over time events - Custom Scan \(SiftAgg native - ms faster33 no - live_counts_by_dim live_counts_by_dim events - fallback:Seq Scan native - ms faster33 no - upsert_returning_conflict upsert_returning_conflict events - ^[0-9]+$ native - ms,cpu_ms faster33 no - series_snapshot_windows series_snapshot_windows events - Custom Scan \(SiftAgg native - ms faster33 no - ledger_rollup_grain ledger_rollup_grain events - Custom Scan \(SiftAgg native - ms faster33 no - count_key_outside_index count_key_outside_index events - Custom Scan \(SiftAgg native - ms faster33 no - topn_list_page_ordered_walk a page of a list (`ORDER BY .. LIMIT n [OFFSET m]` over a table with a col orders default Top-N: created_at DESC native - ms faster33 no - topn_equality_filter_page a page of a list (`ORDER BY .. LIMIT n [OFFSET m]` over a table with a col orders default Custom Scan \(SiftAgg native - ms faster33 no - topn_soft_delete_global_order a page of a list (`ORDER BY .. LIMIT n [OFFSET m]` over a table with a col orders default Global Order: true native - ms faster33 no - topn_text_tiebreak_global_order followed by another key or an id tie-break orders default Global Order: true native - ms faster33 no - topn_text_locale_tiebreak_global_order a page of a list (`ORDER BY .. LIMIT n [OFFSET m]` over a table with a col orders default Global Order: true native - ms faster33 no - topn_text_c_buffer_5k a page of a list (`ORDER BY .. LIMIT n [OFFSET m]` over a table with a col orders default Global Order: true native - ms faster33 no - topn_text_locale_buffer_5k a page of a list (`ORDER BY .. LIMIT n [OFFSET m]` over a table with a col orders default Global Order: true native - ms faster33 no - topn_page_dirty_pages_5pct a page of a list (`ORDER BY .. LIMIT n [OFFSET m]` over a table with a col orders default \A(?!.*[0-9]{3,} vetted by heap).*Top-N: k, id native - ms faster33 no - topn_page_dirty_pages_20pct walks the array, where any condition outside the key sent the scan to the units orders default \A(?!.*[0-9]{3,} vetted by heap).*Top-N: k, id native - ms faster33 no - topn_pass_rate_one_in_100 a page of a list (`ORDER BY .. LIMIT n [OFFSET m]` over a table with a col orders default Custom Scan \(SiftAgg.*Top-N: k, id native walk:sift.test_topn_gbudget = 1000000,units:sift.test_topn_gorder = off ms faster33 no - topn_pass_rate_one_in_100_dirty_pages a page of a list (`ORDER BY .. LIMIT n [OFFSET m]` over a table with a col orders default Custom Scan \(SiftAgg.*Top-N: k, id native walk:sift.test_topn_gbudget = 1000000,units:sift.test_topn_gorder = off ms faster33 no - topn_pass_rate_one_in_10000 a page of a list (`ORDER BY .. LIMIT n [OFFSET m]` over a table with a col orders default Custom Scan \(SiftAgg.*Top-N: k, id native walk:sift.test_topn_gbudget = 1000000,units:sift.test_topn_gorder = off ms faster33 no - topn_pass_rate_one_in_10000_dirty_pages a page of a list (`ORDER BY .. LIMIT n [OFFSET m]` over a table with a col orders default Custom Scan \(SiftAgg.*Top-N: k, id native walk:sift.test_topn_gbudget = 1000000,units:sift.test_topn_gorder = off ms faster33 no - topn_list_page_not_exists a page of a list that excludes a scope with `NOT EXISTS (SELECT .. WHERE k = f.k)` orders default Top-N: created_at native btree:search_path = topn_not_exists_btree ms faster33 no - topn_deep_offset_heap_priced the scan prices the heap of a deep page (`OFFSET` past 1,024 orders default Top-N: created_at DESC native btree:search_path = topn_deep_offset_btree ms faster33 no - wbuf_narrowest_condition_first the write buffers are searched by their narrowest condition events - Index Only Scan using wb_narrow_col native - ms faster33 no - wbuf_batches_kept a scan keeps the rows of a write buffer as batches events - Custom Scan \(SiftAgg native - ms faster33 no - search_and_count - docs - Custom Scan \(SiftCount native - ms,minstr faster33 no - search_or_count - docs - Custom Scan \(SiftCount native - ms,minstr faster33 no - search_phrase_count - docs - Custom Scan \(SiftCount native - ms,minstr faster33 no - search_mixed_count - docs - Custom Scan \(SiftCount native - ms,minstr faster33 no - search_top_20 a block only on the pages where the terms present there can still beat docs - Order: score DESC native - ms,minstr faster33 no - search_filter_include a sift index can name `INCLUDE` columns (`CREATE INDEX ON do docs - Include Tests: [1-9] native - ms,minstr faster33 no - search_join_region the table of an `IN` test of an `INCLUDE` column docs - Include Tests: [1-9] native - ms,minstr faster33 no - search_language_german `stemmer` takes the name of a PostgreSQL text search configu docs - Custom Scan \(SiftCount native - ms,minstr faster33 no - search_build_index - docs - ^[1-9] native - ms,cpu_ms,wal_bytes,minstr faster33 no - col_maintain_cuts_free_end the rewrite of the segments of a `sift_col` index (the automatic one, `sift.col_maintain`) cuts the free end orders - ^[0-9]+$ native - index_mb,ms faster33 no - exceptions_few_groups the exceptions of a unit (the cube of the few rows where a soft-delete stamp is not NULL orders (id) [1-9][0-9]* net of exceptions native scan:sift.agg_preagg = off ms faster33 no - search_top_20_and a ranked search spends fewer instructions on the page index docs - Order: score DESC native - ms,minstr faster33 no - search_top_20_or_three a ranked `OR` builds the candidates of a block from its term docs - Order: score DESC native - ms,minstr faster33 no - search_list_terms_top_20 a term that is a list (fewer postings than `sift.grouped_min docs - Order: score DESC native - ms,minstr faster33 no - search_filter_recent a ranked search with a test of an `INCLUDE` column does not read a superblock docs - Include Tests: [1-9] native - ms,minstr faster33 no - search_vacuum_seals `VACUUM` seals the open write buffer of a BM25 index once it docs - Custom Scan \(SiftScan native - ms,minstr faster33 no - search_build_stemmer_en `CREATE INDEX` with `stemmer = 'en'` spends 30% fewer instru docs - ^[1-9] native - ms,cpu_ms,wal_bytes,minstr faster33 no - search_build_parallel the participants of a parallel `CREATE INDEX` or `REINDEX` of a `sift` (BM25) docs - ^same bytes$ native workers_0:max_parallel_maintenance_workers = 0,workers_2:max_parallel_maintenance_workers = 2,workers_4:max_parallel_maintenance_workers = 4 ms,wal_bytes faster33 no - search_build_parallel_include - docs - ^same bytes$ native workers_0:max_parallel_maintenance_workers = 0,workers_2:max_parallel_maintenance_workers = 2,workers_4:max_parallel_maintenance_workers = 4 ms,wal_bytes faster33 no - search_merge_parallel a merge of segments of a `sift` (BM25) index (`sift.merge` docs - ^segments 1$ native workers_0:max_parallel_maintenance_workers = 0,workers_2:max_parallel_maintenance_workers = 2,workers_4:max_parallel_maintenance_workers = 4 ms,wal_bytes faster33 no - search_merge_parallel_include - docs - ^segments 1$ native workers_0:max_parallel_maintenance_workers = 0,workers_2:max_parallel_maintenance_workers = 2,workers_4:max_parallel_maintenance_workers = 4 ms,wal_bytes faster33 no - par_many_groups_split_by_key a scan with workers of many groups (thousands, a key that is an integer, a date or a timestamp) orders default Groups Split By: grp native one_process:max_parallel_workers_per_gather = 0 ms faster33 no - spill_ordered_rollup_equal_widths the ordered spill of a scan (a rollup ordered by its time bucket past hash memory) orders default [1-9][0-9]* group (table spills|stream batches) native in_memory:work_mem = '4GB' ms faster33 no - spill_ordered_ranges_first_key a scan whose groups the `ORDER BY` wants in order (a rollup by `date_trunc('hour', ts)` orders default [1-9][0-9]* group (table spills|stream batches) native in_memory:work_mem = '4GB' ms faster33 no - col_scan_not_all_visible_lazy_offsets a scan of a `sift_col` index over a table that is not all-visible orders default Pushed Aggregates native vacuumed:search_path = col_scan_not_all_visible_vacuumed ms,minstr faster33 no - count_text_equality_groups_of_units a `count(*)` of one text equality (`WHERE label = 'x'`) on a `sift_col` index orders default Custom Scan $SiftAgg$ native units:sift.agg_super = off ms,minstr faster33 no - col_order_arrays_keep_ends the sorted arrays that the build chooses (`order_array_columns = 'auto: a, b'` orders default Top-N: created_at native whole:search_path = col_order_arrays_whole ms,minstr faster33 no - point_directory_skips_unique_btree `point_tid_columns = 'auto'` no longer picks a column that has a unique b-tree of its own orders - ^[0-9]+$ native listed_id:application_name = listed_id index_mb,ms faster33 no - group_order_rank_sort sorts the distinct values of an integer, date or timestamp key by the value and the groups by a radix sort of their packed ranks orders default Custom Scan \(SiftAgg native - ms,minstr faster33 no - search_insert_language_german a backend keeps the words its dictionaries answered between statements docs - ^[0-9]+$ native - ms,cpu_ms,minstr faster33 no - search_top_20_ties a ranked search whose `ORDER BY` has a key after the score docs - Order: score DESC native - ms,minstr faster33 no - group_one_table_start a scan of one table that groups is priced as starting at 0 events - Custom Scan \(SiftAgg native - ms faster33 no - in_list_and_subselect_planned_once a statement that holds a literal `IN` list and a sub-selec events - Custom Scan \(SiftAgg native - ms faster33 no - count_not_null_column_is_count_star `count(x)` of a column the table declares `NOT NULL` is `c events - Custom Scan \(SiftAgg native - ms faster33 no - date_bin_whole_seconds_step of whole seconds is a step of its own, like `date_trunc` events - per 900 seconds native - ms faster33 no - union_all_group_added_above a `GROUP BY` over a `UNION ALL` of selects (a queue's dept events - Custom Scan \(SiftAgg native - ms faster33 no - is_null_priced_as_comparison a test of `IS NULL` or `IS NOT NULL` on a column of a `sif events - Custom Scan \(SiftAgg native - ms faster33 no - topn_offset_key_cut the sorted array of the key of a `key_cut` index orders default Top-N: created_at native btree:search_path = topn_offset_key_cut_btree ms faster33 no - order_layout_topn the reloption `order_layout` of a `sift_col` index events - Top-N: created_at native whole:search_path = c_order_layout_topn__whole,compact:search_path = c_order_layout_topn__compact,btree:search_path = c_order_layout_topn__btree ms faster33 no - order_layout_offset_early the reloption `order_layout` of a `sift_col` index events - Top-N: created_at \(limit native whole:search_path = c_order_layout_offset_early__whole,compact:search_path = c_order_layout_offset_early__compact,btree:search_path = c_order_layout_offset_early__btree ms faster33 no - order_layout_offset_mid the reloption `order_layout` of a `sift_col` index events - Top-N: created_at \(limit native whole:search_path = c_order_layout_offset_mid__whole,compact:search_path = c_order_layout_offset_mid__compact,btree:search_path = c_order_layout_offset_mid__btree ms faster33 no - order_layout_offset_deep the reloption `order_layout` of a `sift_col` index events - Top-N: created_at \(limit native whole:search_path = c_order_layout_offset_deep__whole,compact:search_path = c_order_layout_offset_deep__compact,btree:search_path = c_order_layout_offset_deep__btree ms faster33 no - order_layout_count the reloption `order_layout` of a `sift_col` index events - Pushed Aggregates: count native whole:search_path = c_order_layout_count__whole,compact:search_path = c_order_layout_count__compact,btree:search_path = c_order_layout_count__btree ms faster33 no - order_layout_ties_offset the reloption `order_layout` of a `sift_col` index events - Top-N: d \(limit native whole:search_path = c_order_layout_ties_offset__whole,compact:search_path = c_order_layout_ties_offset__compact,btree:search_path = c_order_layout_ties_offset__btree ms faster33 no - order_layout_ties_count the reloption `order_layout` of a `sift_col` index events - Pushed Aggregates: count native whole:search_path = c_order_layout_ties_count__whole,compact:search_path = c_order_layout_ties_count__compact,btree:search_path = c_order_layout_ties_count__btree ms faster33 no - wbuf_layout_topn the reloption `order_layout` of a `sift_col` index events - Top-N: created_at native whole:search_path = c_wbuf_layout_topn__whole,compact:search_path = c_wbuf_layout_topn__compact,btree:search_path = c_wbuf_layout_topn__btree ms faster33 no - wbuf_layout_offset gives the order of a key in ranges that are sorted when a read touches them events - Top-N: created_at \(limit native whole:search_path = c_wbuf_layout_offset__whole,compact:search_path = c_wbuf_layout_offset__compact,btree:search_path = c_wbuf_layout_offset__btree ms faster33 no - wbuf_layout_count the reloption `order_layout` of a `sift_col` index events - Pushed Aggregates: count native whole:search_path = c_wbuf_layout_count__whole,compact:search_path = c_wbuf_layout_count__compact,btree:search_path = c_wbuf_layout_count__btree ms faster33 no - wbuf_layout_text_topn the reloption `order_layout` of a `sift_col` index events - Top-N: txt native btree:search_path = c_wbuf_layout_text_topn__btree ms faster33 no - wbuf_layout_text_offset the reloption `order_layout` of a `sift_col` index events - Top-N: txt \(limit native btree:search_path = c_wbuf_layout_text_offset__btree ms faster33 no - wbuf_layout_text_count the reloption `order_layout` of a `sift_col` index events - Pushed Aggregates: count native btree:search_path = c_wbuf_layout_text_count__btree ms faster33 no - point_scattered_1 the reloption `order_layout` of a `sift_col` index events - from point directory native btree:search_path = c_point_scattered_1__btree ms faster33 no - point_scattered_10 the reloption `order_layout` of a `sift_col` index events - from point directory native btree:search_path = c_point_scattered_10__btree ms faster33 no - point_scattered_60 the reloption `order_layout` of a `sift_col` index events - Index Scan using ps_col native btree:search_path = c_point_scattered_60__btree,serial:search_path = c_point_scattered_60__serial ms faster33 no - point_scattered_500 the reloption `order_layout` of a `sift_col` index events - Index Scan using ps_col native btree:search_path = c_point_scattered_500__btree,serial:search_path = c_point_scattered_500__serial ms faster33 no - search_open_buffer a search reads the documents of the open write buffer of a BM25 index by their hashes docs - Custom Scan \(SiftScan native - ms,minstr faster33 no - col_build_default - orders - ^[0-9]+$ native workers_0:max_parallel_maintenance_workers = 0,workers_4:max_parallel_maintenance_workers = 4 ms,cpu_ms,wal_bytes,minstr faster33 no - col_build_date_key takes fewer passes over its entries orders - ^[0-9]+$ native workers_0:max_parallel_maintenance_workers = 0,workers_4:max_parallel_maintenance_workers = 4 ms,cpu_ms,wal_bytes,minstr faster33 no - col_build_cube_13cols the cube of a unit adds the integer measures of a dimension in one pass per group orders - ^[0-9]+$ native workers_0:max_parallel_maintenance_workers = 0,workers_4:max_parallel_maintenance_workers = 4,workers_8:max_parallel_maintenance_workers = 8 ms,cpu_ms,wal_bytes,minstr faster33 no - col_vacuum_13cols the zone of a chunk without a NULL and the exact sums of an integer leaf orders - ^[0-9]+$ native - ms,cpu_ms,wal_bytes,minstr faster33 no - col_vacuum_churn_sidecar the cube of a unit takes the minimum and maximum of a measure without a branch for the first value of each group orders - ^[0-9]+$ native - ms,cpu_ms,wal_bytes,minstr faster33 no - col_key_cut_decision the choice of the default `key_cut` of a `sift_col` index orders - ^[0-9]+$ native - ms,cpu_ms,minstr faster33 no - text_order_array_size the order array of a text column (`order_array_columns`) names the strings of its ranks by their dictionary codes orders - ^[0-9]+$ native - index_mb,ms,minstr faster33 no - order_array_global_bits the global orders of an index with order arrays and no point directory or ends orders - ^[0-9]+$ native - index_mb,ms faster33 no - col_merge_cuts_free_end the merge of a `sift_col` index (the maintenance worker's, `sift.col_merge`) cuts the free end of the file it leaves orders - ^[0-9]+$ native - index_mb,ms faster33 no - topn_offset_tiebreak a deep `OFFSET` over the sorted array of the first key with a tie-break key after it orders default Top-N: d native btree:search_path = c_topn_offset_tiebreak__btree ms faster33 no - topn_offset_after_vacuum an `OFFSET` over the sorted array of a key does not confirm in the heap the entries it throws away after a `VACUUM` orders default Top-N: created_at native btree:search_path = c_topn_offset_after_vacuum__btree ms faster33 no - search_ingest_fanin `merge_fanin` defaults to 4 (8 before) docs - ^[1-9] native - ms,wal_bytes,minstr faster33 no - point_scattered_subselect is an IN list of the scan, read when the scan starts events - Pushed Aggregates: count$\*$, sum$v$ native btree:search_path = c_point_scattered_subselect__btree ms faster33 no - wbuf_big_batches_kept the batches of a write buffer that an index allows to grow big events - Custom Scan \(SiftAgg native - ms faster33 no - count_churn_no_vacuum the check against the heap of the rows a scan selects on a page that is not all-visible orders default Custom Scan \(SiftAgg native - ms faster33 no - sum_dead_units reads its liveness bits 64 at a time orders default Custom Scan \(SiftAgg native - ms faster33 no - count_distinct_sparse_sets give way to hash sets when they pass 1 MB orders default Custom Scan \(SiftAgg native - ms faster33 no - dict_shared_order_array the hash table and the sorted order of a dictionary that the backends share orders default Custom Scan \(SiftAgg native - ms faster33 no - topn_walk_condition_offset the walk of the global order with a condition on another column orders default Top-N: k, id native btree:search_path = c_topn_walk_condition_offset__btree ms faster33 no - lateral_top1_text_filter is tested on the heap row of each entry of the order-array walk orders default Top-N native btree:search_path = c_lateral_top1_text_filter__btree ms faster33 no - lateral_top1_kept_source keeps its source of the index open from one outer row to the next orders default Top-N native btree:search_path = c_lateral_top1_kept_source__btree ms faster33 no - lateral_top1_wbuf_groups puts the rows of the write buffer in groups once for the statement orders default Top-N native btree:search_path = c_lateral_top1_wbuf_groups__btree ms faster33 no - distinct_on_latest_per_entity keep the first row of a tie in the scan orders default Custom Scan \(SiftAgg native btree:search_path = c_distinct_on_latest_per_entity__btree ms faster33 no - distinct_on_latest_all_entities is chosen among the rows of a chunk before the values of the row are copied orders default Custom Scan \(SiftAgg native btree:search_path = c_distinct_on_latest_all_entities__btree ms faster33 no - join_filter_no_key not for the tuples the scan keeps orders default fallback:Hash native - ms faster33 no - topn_offset_churn_int are kept by the backend for its next scan orders default Top-N: k native btree:search_path = c_topn_offset_churn_int__btree ms faster33 no - topn_offset_churn_text continues through the entries whose heap page is not all-visible orders default Top-N: t native btree:search_path = c_topn_offset_churn_text__btree ms faster33 no - topn_offset_churn_date - orders default Top-N: d native btree:search_path = c_topn_offset_churn_date__btree ms faster33 no - topn_offset_dates_auto weighs a date or timestamp column of fewer than 1,000 distinct values by the default formula orders default (?s)\A(?=.*Global Order)(?=.*Top-N: d)(?!.*[^0-9](?:10[1-9]|1[1-9][0-9]|[2-9][0-9]{2}|[0-9]{4,}) heap fetches) native btree:search_path = c_topn_offset_dates_auto__btree ms faster33 no - wbuf_copy_13cols is encoded and queued without a call of the StringInfo functions for each value orders - ^[0-9]+$ native - ms,cpu_ms,wal_bytes,minstr faster33 no - col_merge_forced_key_cut reads the range of an integer column from the zone map instead of scanning its rows again orders - ^[0-9]+$ native workers_0:max_parallel_maintenance_workers = 0,workers_2:max_parallel_maintenance_workers = 2,workers_4:max_parallel_maintenance_workers = 4 ms,cpu_ms,wal_bytes,minstr faster33 no - search_vacuum_dead_postings by reading the postings of a LIST term one by one against a flat table of the dead rows of each superblock docs - ^[0-9]+$ native - ms,cpu_ms,wal_bytes,minstr faster33 no - point_in_ordered_60 keeps the index and bitmap scans of the `sift_col` index beside the aggregate scan events - Index Scan using pi_col|Bitmap Heap Scan on pi|from point directory native btree:search_path = c_point_in_ordered_60__btree,scan:sift.agg_force = on ms faster33 no - topn_offset_presorted reads the 10,020 rows from the end of the one unit its key follows orders default Top-N: k native btree:search_path = c_topn_offset_presorted__btree ms faster33 no - point_tid_auto_60 by default also for an integer key of nearly unique values that events - Index Scan using pa_col|Bitmap Heap Scan on pa|from point directory native btree:search_path = c_point_tid_auto_60__btree ms faster33 no - key_order_offset `key_order = 'ts'` (or `'entity, ts'`) on a `sift_col` index events - Global Order: true native default:search_path = c_key_order_offset__default,btree:search_path = c_key_order_offset__btree ms faster33 no - key_cut_time_order where the default `key_cut` of `CREATE INDEX` and `REINDEX` of a `sift_col` index chooses a date or timestamp column events - Global Order: true native off:search_path = c_key_cut_time_order__off,btree:search_path = c_key_cut_time_order__btree ms faster33 no - key_order_entity_latest the default layout of a `sift_col` index with an entity column (a foreign key or `*_id`) and a date or timestamp column is `key_order = 'auto: entity, ts'` events - Global Order: true native off:search_path = c_key_order_entity_latest__off,btree:search_path = c_key_order_entity_latest__btree ms faster33 no - key_order_latest the latest row of each of 100 tenants events - Global Order: true native default:search_path = c_key_order_latest__default,btree:search_path = c_key_order_latest__btree ms faster33 no - lateral_top1_partitions reads the top-N of each partition and merges them orders default Top-N native btree:search_path = c_lateral_top1_partitions__btree ms faster33 no - lateral_top1_hash_partitions runs the top-N of the one partition that holds the value orders default Top-N native btree:search_path = c_lateral_top1_hash_partitions__btree ms faster33 no - join_eq_nl_probe starts over for each row of an outer query orders default Nested Loop native btree:search_path = c_join_eq_nl_probe__btree ms faster33 no - topn_offset_filter_int a page under a condition on another column orders default (?s)\A(?=.*Global Order)(?=.*Top-N: k)(?!.*[^0-9](?:10[1-9]|1[1-9][0-9]|[2-9][0-9]{2}|[0-9]{4,}) heap fetches) native btree:search_path = c_topn_offset_filter_int__btree ms faster33 no - topn_offset_filter_date - orders default (?s)\A(?=.*Global Order)(?=.*Top-N: d)(?!.*[^0-9](?:10[1-9]|1[1-9][0-9]|[2-9][0-9]{2}|[0-9]{4,}) heap fetches) native btree:search_path = c_topn_offset_filter_date__btree ms faster33 no - lateral_latest_row_default order_array_group = auto groups the sorted arrays of dates, timestamps and integers by an entity column orders default Top-N native btree:search_path = c_lateral_latest_row_default__btree ms,minstr faster33 no - ioss_whole_scan an index-only scan of a `sift_col` index whose returned columns are i orders default Seq Scan on iw|Index Only Scan using iw_col native ioss:sift.enable_agg = off;enable_seqscan = off;seq_page_cost = 1000;enable_indexscan = off;enable_bitmapscan = off;max_parallel_workers_per_gather = 0,btree:search_path = c_ioss_whole_scan__btree ms,minstr faster33 no - group_count_entity counts the group from the entries of its second order orders default Custom Scan $SiftAgg$ native btree:search_path = c_group_count_entity__btree,ungrouped:search_path = c_group_count_entity__ungrouped ms faster33 no - group_latest_entity the latest row of each of 200 groups orders default Custom Scan $SiftAgg$ native btree:search_path = c_group_latest_entity__btree,ungrouped:search_path = c_group_latest_entity__ungrouped ms faster33 no - tenant_latest_20 reads the units a filter leaves when the walk of the order array would fetch more rows orders default Top-N: event_date native btree:search_path = c_tenant_latest_20__btree ms faster33 no - topn_offset_dict_blocks keep the last 32 blocks read orders default Top-N: t native btree:search_path = c_topn_offset_dict_blocks__btree ms faster33 no - group_range_count counts the entries of the group between the two keys orders default from pre-aggregates native btree:search_path = c_group_range_count__btree ms faster33 no - tbl_page_tombstones_scan a scan of a `sift_table` with deleted or updated rows not yet vacuumed orders - tombstones native - ms faster33 no - point_one_row - orders default Index (Only )?Scan native - ms faster33 no - table_read_churned the background pass of a `sift_table` is asked for when the reads of the table since the last pass have cost as much as the pass orders default Custom Scan \(SiftAgg native - ms faster33 no - table_read_settled a VACUUM of a `sift_table` writes the pre-aggregates of the units that kept a dead row net of those rows orders default Custom Scan \(SiftAgg native - ms faster33 no - col_build_participants a parallel `CREATE INDEX` of a `sift_col` index launches a worker for each further 150,000 rows of the table orders - ^[0-9]+$ native server_default:min_parallel_table_scan_size = '8MB',one_process:max_parallel_maintenance_workers = 0,all_asked:max_parallel_maintenance_workers = 4;min_parallel_table_scan_size = 0 index_mb,ms faster33 no - col_participants_point a parallel `CREATE INDEX` of a `sift_col` index launches a worker for each further 150,000 rows of the table orders default Custom Scan \(SiftAgg native parts:search_path = c_col_participants_point__parts ms,minstr faster33 no - col_participants_scan a parallel `CREATE INDEX` of a `sift_col` index launches a worker for each further 150,000 rows of the table orders default Custom Scan \(SiftAgg native parts:search_path = c_col_participants_scan__parts ms,minstr faster33 no - ordered_key_scan the ordered index-only scan of a `sift_col` index (the first column of the index read alone, in order) is offered events - Index Only Scan native btree:search_path = c_ordered_key_scan__btree ms faster33 no - ordered_window_rank the ordered index-only scan of a `sift_col` index (the first column of the index read alone, in order) is offered events - WindowAgg native btree:search_path = c_ordered_window_rank__btree ms faster33 no - ordered_join_order the ordered index-only scan of a `sift_col` index (the first column of the index read alone, in order) is offered events - Merge Join native btree:search_path = c_ordered_join_order__btree ms faster33 no - ordered_group_key `ORDER BY g, k` (both ascending or both descending) and `WHERE g = x ORDER BY k` over a `sift_col` index on `(g, k)` events - Index Only Scan native btree:search_path = c_ordered_group_key__btree ms faster33 no - grouping_sets_subselect_keys keys that are expressions of a sub-select orders default Pushed Aggregates native - ms faster33 no - lateral_probe_direct the keys of a correlated `LIMIT 1` child are probed in its btree a batch at a time events - child direct probes native executor:sift.agg_probe_direct = off ms faster33 no - lateral_probe_index_only the keys of a correlated `LIMIT 1` child that is an Index Only Scan are probed in its btree a batch at a time events - child direct probes native executor:sift.agg_probe_direct = off ms faster33 no - plan_cost_keys_1 - events - parents native - ms,minstr faster33 no - plan_cost_keys_20 - events - parents native - ms,minstr faster33 no - plan_cost_keys_200 - events - parents native warm_backend:sift.enable_agg = on ms,minstr faster33 no - plan_cost_lateral_pull_up a statement whose lateral sub-selects can be pulled up is planned once events - Limit native - ms,minstr faster33 no - key_cut_overlap_few_keys a `key_cut` index whose runs are as large as partitions is rewritten by key range orders default Custom Scan $SiftAgg$ native cut:search_path = c_key_cut_overlap_few_keys__cut ms,minstr faster33 no - group_sort_ranks a `Sort` of many rows above the groups of a scan orders default Custom Scan $SiftSort$ native - ms faster33 no - group_join_limit_startup the normal plan that gives its first group at once events - Index Only Scan native - ms faster33 no - group_union_limit_correlated reads through a condition on the table events - Custom Scan $SiftAgg$[\s\S]*Custom Scan $SiftAgg$ native - ms faster33 no - lookup_in_keyed the keys of the scan's IN list that a joined btree table is keyed by are read by one scan of its btree events - Lookup: .* index scans native whole:sift.agg_lookup_pushdown = off ms faster33 no - group_order_many_keys takes less work for each group orders default $[1-9][0-9]* order keys$ native - ms,minstr faster33 no - group_join_dim_filter_keys looked up in the filtered table before the groups are made events - Index (Only )?Scan (using|on) gk_fact_grp native - ms faster33 no - lookup_index_only the rows of a table joined by the scan's IN list are taken from the entries of a btree that holds their columns events - Lookup: .* index scans native whole:sift.agg_lookup_pushdown = off ms faster33 no - lookup_dense_classes is kept as classes of its keys events - Lookup: .* rows classed native rows:sift.agg_lookup_columns = off ms faster33 no - group_lookup_scalar_key a row of the table looked up by its unique key events - Custom Scan $SiftAgg$ native - ms faster33 no - group_filter_exists_key its `FILTER` of a grouped query events - Custom Scan $SiftAgg$ native - ms faster33 no - group_two_keys_hash probes the table with the comparison unrolled orders default Custom Scan $SiftAgg$ native - ms,minstr faster33 no - filter_scalar_subselect_bound a scalar sub-select that gives one row events - Custom Scan $SiftAgg$[\s\S]*InitPlan native - ms faster33 no - float_step_buckets a float column grouped by the buckets of its values is evaluated for each bucket events - Group Keys: value_a: $floor native executor:sift.agg_float_step = off ms faster33 no - group_spill_two_keys costs less for each group it writes to the partitions and merges back orders default Custom Scan \(SiftAgg$ native - ms,minstr faster33 no - group_direct_arrays_passes whose direct arrays (the capacities of the keys multiply) are more than hash memory puts its groups in the hash table at once orders default Custom Scan $SiftAgg$ native - ms,minstr faster33 no - lookup_probe_join a join to a large btree table on a range of the scan's keys is a probe of the btree for each key the scan meets events - Lookup Joins: sift_probe native whole:sift.agg_probe_rewrite = off ms faster33 no - float_sum_ungrouped the sum of a float column that is not grouped events - Custom Scan $SiftAgg$ native - ms faster33 no - group_right_join_leaf_agg and no longer builds an id for each distinct value events - Custom Scan $SiftAgg$ native - ms faster33 no - lookup_probe_not_in a NOT IN over a large btree table (neither side NULL) is a probe of the btree for each key the scan meets events - Lookup Joins: .*sift_probe native whole:sift.agg_probe_rewrite = off ms faster33 no - agg_sum_scalar_factor a sum of a numeric column times a value asked once events - Custom Scan $SiftAgg$[\s\S]*InitPlan native - ms faster33 no - group_passes_upfront chooses the number of passes from the planner's estimate of the groups and reads the table once for each pass orders default Custom Scan $SiftAgg$ native - ms,minstr faster33 no - agg_sum_lookup_product a sum of a numeric column times a value of a table joined by its unique key events - Custom Scan $SiftAgg$ native - ms faster33 no - fresh_request_lookup - events - child direct probes native warm_backend:sift.enable_agg = on ms,minstr faster33 no - entity_list_btree_plan the heap fetch of a top-N is priced at a page of the b-tree plan with a LIMIT, so a list of rows of one entity that a b-tree on the entity and the key gives in order keeps that plan events - fallback:Index Scan using el_bt|Incremental Sort native warm_forced:sift.agg_force = on ms faster33 no - agg_case_filter `sum`, `avg`, `min`, `max` and `count` over `CASE WHEN p THEN x` events - Pushed Aggregates: count$\*$, sum$amount$ native whole:sift.agg_case_filter = off ms,minstr faster33 no - col_units_per_key the merge of the small overlay segments also serves an index that is not cut into partitions orders default Custom Scan \(SiftAgg native layered:search_path = c_col_units_per_key__layered ms,minstr faster33 no - agg_pred_normalize a condition that wraps a column in `coalesce(col, c)` events - Filters: amount > native - ms,minstr faster33 no - col_overlay_wide_units the merge of the small overlay segments counts the rows of the units a request scans orders default Custom Scan $SiftAgg native layered:search_path = c_col_overlay_wide_units__layered ms,minstr faster33 no - col_wbuf_read_priced backends that read it have paid for its promotion orders default Custom Scan \(SiftAgg native buffered:search_path = c_col_wbuf_read_priced__buffered ms,minstr faster33 no - col_wbuf_read_priced_set backends that read it have paid for its promotion orders default Custom Scan \(SiftAgg native buffered:search_path = c_col_wbuf_read_priced_set__buffered ms,minstr faster33 no - col_wbuf_read_priced_all backends that read it have paid for its promotion orders default Custom Scan \(SiftAgg native buffered:search_path = c_col_wbuf_read_priced_all__buffered ms,minstr faster33 no - col_log_block_cache the tombstone log is asked once for a heap block orders default Custom Scan \(SiftAgg native - ms,minstr faster33 no - agg_jsonb_contain_object a containment `doc @> '{...}'` with a boolean or JSON null member on an object document events - Custom Scan \(SiftAgg$ native - ms,minstr faster33 no - set_key_skew_dimension multiplied by the skew of the column events - Custom Scan $SiftAgg native - ms faster33 no - group_int_expr_halves `GROUP BY id / 2` events - Custom Scan \(SiftAgg$ native - ms faster33 no - group_bigint_key_dense an `int`, `bigint`, `date`, `timestamp` or `timestamptz` column events - Custom Scan $SiftAgg$ native - ms faster33 no - agg_window_topn `row_number() OVER (PARTITION BY key ORDER BY ..) <= N` keeps only the first N rows of each group in the scan events - array_agg$wn_fact\.id ORDER BY.*\[1:3\] native - ms,minstr faster33 no - point_pk a lookup of one row by its primary key events - Index Scan using pk_fact_pkey native - ms,minstr faster33 no - agg_stat_variance_groups `stddev`, `variance` events - Custom Scan \(SiftAgg$ native - ms faster33 no - agg_stat_corr_groups `corr`, `regr_slope` events - Custom Scan $SiftAgg$ native - ms faster33 no - agg_mode_text_groups `mode() WITHIN GROUP` events - Custom Scan $SiftAgg$ native - ms faster33 no - percentile_distinct_values `percentile_disc` events - Custom Scan $SiftAgg$ native - ms faster33 no - agg_window_rank keeps in the scan the rows of rank <= N of each group events - array_agg$rk_fact\.id ORDER BY.*\[rank <= 3\] native - ms,minstr faster33 no - agg_probe_anti the child's rows are loaded once instead of a probe of its table for each outer row events - Lookup Joins: pa_items \(anti$ native - ms,minstr faster33 no - set_list_unit_span `IN` list of 75 keys events - Custom Scan $SiftAgg$ native - ms faster33 no - agg_left_partial the groups of the table by the keys of the join are made in the scan and joined to the pairs events - Subquery Scan on e native - ms,minstr faster33 no - topn_with_ties `FETCH FIRST n ROWS WITH TIES` by an integer column events - Top-N: dim_a_id DESC native - ms faster33 no - agg_percentile_array the percentiles of an array of fractions are picked from the one set of values the scan keeps events - Pushed Aggregates: percentile_cont$0.25$ native - ms,minstr faster33 no - agg_topn_chunk a tuple of a topn that comes after the last kept one of its group is dropped on the first sort key alone events - array_agg$tk_events\.id ORDER BY.*\[1:5\] native - ms,minstr faster33 no - agg_distinct_ordered the distinct values of a group ordered by the argument itself are a set that is sorted once events - array_agg\(DISTINCT do_events\.kind_id ORDER BY native - ms,minstr faster33 no - jsonb_key_exists `doc ? 'k'` of a document column that declares the path k events - Pushed Aggregates: count\(\*$ native - ms faster33 no - agg_having_where a HAVING condition with a sub-select that reads no aggregate is a condition of the rows events - Custom Scan $SiftAgg$ native - ms,minstr faster33 no - search_bitmap_recheck checks the match again only on the rows of a lossy page docs - Recheck Cond: .*match_idx native - ms,minstr faster33 no - agg_output_subplan a sub-select in the select list or the HAVING of a grouped query is evaluated by the scan node once for each group events - Custom Scan $SiftAgg$ native - ms,minstr faster33 no - agg_lookup_range_bounds the table of a lookup join that has its own bounds events - Custom Scan $SiftAgg$ native - ms,minstr faster33 no - jsonb_key_forms answered by the index in a FILTER, an OR or a CASE events - Pushed Aggregates: count$\*$, sum native - ms faster33 no - jsonb_key_read the value of a key read beside the test of the key events - Pushed Aggregates: sum$attrs\.a$ native - ms faster33 no - agg_case_else an average, a minimum or a count of a CASE with an ELSE that is a value events - Pushed Aggregates: .*sum$qty$ native - ms faster33 no - subselect_max_per_entity a correlated `min()` or `max()` sub-select of a table is the first row of its entity orders default Global Order: true native btree:search_path = c_subselect_max_per_entity__btree ms,minstr faster33 no - dimmap_shared_classes once a backend has read them from the leaves events - Lookup: .*[1-9][0-9]* maps (built|shared) native nomap:sift.agg_dimmap_kb = 0 ms,minstr faster33 no - count_distinct_big_set gets its pairs sorted once the hash set outgrows the cache events - Pushed Aggregates native - ms faster33 no - lookup_expr_column is read from the leaf of the index that holds it events - Lookup: .* column loads native rows:sift.agg_lookup_columns = off ms faster33 no - semi_lut_bigint by an `int8` key is answered from a byte table events - [1-9][0-9]* semi tables native generic_join:sift.agg_semi_lut = off ms faster33 no - lookup_bool_filters are columns of the lookup table read from its index events - Lookup: .* column loads native rows:sift.agg_lookup_columns = off ms faster33 no - lookup_dense_kept is read from the cache of an earlier statement as classes of its keys events - Lookup: .*cached tables native rows:sift.agg_lookup_columns = off ms faster33 no - lookup_scoped_keys is read for the keys of the list alone events - Lookup: [0-9]+ rows read native rows:sift.agg_lookup_pushdown = off ms faster33 no - lookup_dense_tests are columns of the lookup table computed once for each of its rows events - Lookup: .*(rows classed|cached tables) native rows:sift.agg_lookup_columns = off ms faster33 no - join_count_per_parent `count(x)` of a column of a joined table keeps a flag for each joined tuple events - Custom Scan $SiftAgg$ native - ms faster33 no - distinct_split share one set of ids: the scan groups by the outcome of each filter events - Group Keys: .*lookup_id: $NULLIF native - ms faster33 no - found_path_exists is answered from the flag the segments keep that the parent row was found events - [1-9][0-9]* found proofs native join:sift.agg_connect_found = off ms,minstr faster33 no - found_path_under_parent_updates_1pct is answered from the flag the segments keep that the parent row was found events - [1-9][0-9]* found proofs native join:sift.agg_connect_found = off ms,minstr faster33 no - found_path_under_parent_updates_10pct is answered from the flag the segments keep that the parent row was found events - [1-9][0-9]* found proofs native join:sift.agg_connect_found = off ms,minstr faster33 no - invert_exists_child_filter is planned also as the count of the distinct keys of the child's rows that have a parent events - count\(DISTINCT dim_id$ native join:sift.agg_child_first = off ms,minstr faster33 no - invert_not_exists_child_filter is planned also as the count of the distinct keys of the child's rows that have a parent events - count$DISTINCT dim_id$ native join:sift.agg_child_first = off ms,minstr faster33 no - child_first_left_join_groups made for each key of the child first and added up over the parents events - sum$iv_events\.a1$ native join:sift.agg_child_first = off ms,minstr faster33 no - parent_filter_child_sum - events - Custom Scan $SiftAgg$ native - ms,minstr faster33 no - parent_child_distinct - events - Custom Scan $SiftAgg$ native - ms,minstr faster33 no - parent_child_id_list - events - Custom Scan $SiftAgg$ native - ms,minstr faster33 no - parent_lateral_top1 - events - Custom Scan $SiftAgg$ native - ms,minstr faster33 no -