Riza Suminto has posted comments on this change. ( http://gerrit.cloudera.org:8080/15511 )
Change subject: WIP IMPALA-9434: Implement Robin Hood Hash Table. ...................................................................... Patch Set 11: Patch Set 11 add improvement for insertion into Robin-Hood hash table. If an insertion target a bucket that is an empty, do insertion in-place, bypassing temporary bucket and swap routine. This reduce unnecessary number of bucket swaps. Based on Patch Set 11, I have run initial performance measurement against TPC-DS with scale 60 GB over 3 nodes. The baseline is TPC-DS under quadratic probing hash table. Among 91 queries, 34 queries perform better in Robin-Hood than Quadratic, 57 others show degradation. The breakdown are the following: - 1 query improve by 18.69% - 3 queries improve between 11.29% to 14.21% - 8 queries improve between 5.90% to 9.93% - 22 queries improve between 0.02% to 4.14% - 42 queries degrade 0% to -4.86% - 8 queries degrade between -5.51% to -8.50% - 6 queries degrade between -10.49% to -14.89% - 1 queries degrade between -23.50% The improvement and degradations mostly still within tens and hundreds of miliseconds, so I suspect larger test scale will be more representative. Meanwhile, I will study the query profile from my runs and look if we can find some explanation about this result. -- To view, visit http://gerrit.cloudera.org:8080/15511 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: I28eeccd7f9ccae39e31972391f971901bcbfe986 Gerrit-Change-Number: 15511 Gerrit-PatchSet: 11 Gerrit-Owner: Riza Suminto <riza.sumi...@cloudera.com> Gerrit-Reviewer: David Rorke <dro...@cloudera.com> Gerrit-Reviewer: Impala Public Jenkins <impala-public-jenk...@cloudera.com> Gerrit-Reviewer: Riza Suminto <riza.sumi...@cloudera.com> Gerrit-Reviewer: Tim Armstrong <tarmstr...@cloudera.com> Gerrit-Comment-Date: Mon, 06 Apr 2020 17:31:09 +0000 Gerrit-HasComments: No
