Zoltan Martonka has posted comments on this change. ( http://gerrit.cloudera.org:8080/21127 )
Change subject: [ARM] Concurrent binary tree memory barriers fixed. ...................................................................... Patch Set 6: (1 comment) I realized that, the only proper way to add a 2-way load barrier is with std::atomic_thread_fence(std::memory_order_acquire), which is more strict, than the 1-way std::atomic<T>::load(std::memory_order_acquire); I made some performance test. I used TestScanPerformance from cbtree-test and TestMemRowSetUpdatePerformance from memrowset-test. I also added 2 more performance tests, one using basic arena, one using ThreadSafeMemoryTrackingArena (we only use the 2nd in production). I will add g++ tests too (as soon as it compiles). See the branches sheet for name resolution. https://docs.google.com/spreadsheets/d/1whl8h7S0DcjVYp0jksR87P9v4U8TQH9nS1C9N0D-lUo/edit?usp=sharing http://gerrit.cloudera.org:8080/#/c/21127/6//COMMIT_MSG Commit Message: http://gerrit.cloudera.org:8080/#/c/21127/6//COMMIT_MSG@33 PS6, Line 33: The following 6 barriers need to be more strict: > Did you run the TestCBTree.TestScanPerformance scenario from the cbtree-tes See the google sheet. -- To view, visit http://gerrit.cloudera.org:8080/21127 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: kudu Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: Ie7c02dc3b9444599ef680fa9b29ae7a4d39dd382 Gerrit-Change-Number: 21127 Gerrit-PatchSet: 6 Gerrit-Owner: Zoltan Martonka <[email protected]> Gerrit-Reviewer: Alexey Serbin <[email protected]> Gerrit-Reviewer: Anonymous Coward (763) Gerrit-Reviewer: Ashwani Raina <[email protected]> Gerrit-Reviewer: Kudu Jenkins (120) Gerrit-Reviewer: Tidy Bot (241) Gerrit-Reviewer: Zoltan Chovan <[email protected]> Gerrit-Reviewer: Zoltan Martonka <[email protected]> Gerrit-Comment-Date: Tue, 23 Apr 2024 13:00:06 +0000 Gerrit-HasComments: Yes
