Aleksandr Efimov has posted comments on this change. ( http://gerrit.cloudera.org:8080/24883 )
Change subject: IMPALA-13052: Estimate reservoir sample memory in aggregations ...................................................................... Patch Set 1: Hi! Measurements behind the commit message. Per group: the aggregator's PeakMemoryUsage minus its PeakReservation, divided by the number of groups, on tpch_parquet.lineitem; measured vs estimated bytes: appx_median(l_orderkey) by l_orderkey, ~4 rows 4642 vs 4648 appx_median(l_quantity) by l_orderkey, ~4 rows 5145 vs 5160 appx_median(l_orderkey) by l_suppkey, ~600 rows 72351 vs 72050 appx_median(l_orderkey) by l_suppkey % 100, 60K rows 1384067 vs 1370294 appx_median(l_quantity) by l_suppkey % 100, 60K rows 2432651 vs 2418358 merge of appx_median(l_shipmode) by l_orderkey 4708-4744 vs 4648 merge of two DECIMAL appx_median(), ~1 row 10300-10307 vs 10320 The first five ran on one instance (NUM_NODES=1), the merges on three. By l_partkey, 30 rows per group on average, it is 6244 vs 5680: the estimate sizes the array for 32 samples, but 32% of those groups have more rows, and their state adds a 64-sample block while the 32-sample one stays in FreePool's free list. Weighted by the actual group sizes, the model gives 6332. -- To view, visit http://gerrit.cloudera.org:8080/24883 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: Id6eec3daf7fea9b027ff0be9c95c6ee5a3cf323d Gerrit-Change-Number: 24883 Gerrit-PatchSet: 1 Gerrit-Owner: Aleksandr Efimov <[email protected]> Gerrit-Reviewer: Aleksandr Efimov <[email protected]> Gerrit-Reviewer: Csaba Ringhofer <[email protected]> Gerrit-Reviewer: Impala Public Jenkins <[email protected]> Gerrit-Reviewer: Riza Suminto <[email protected]> Gerrit-Comment-Date: Fri, 18 Sep 2026 10:32:15 +0000 Gerrit-HasComments: No
