[
https://issues.apache.org/jira/browse/CASSANDRA-9989?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16596997#comment-16596997
]
Jay Zhuang commented on CASSANDRA-9989:
---------------------------------------
There's a little bit improvement when we pre-compute the child size and split
the left values to the last 2 nodes:
[{{9989-2}}|https://github.com/cooldoger/cassandra/tree/9989-2]. Here is my
benchmark test results (it should only impact the large bTree build):
{noformat}
== calculate child size every round:
[java] Benchmark (dataSize) Mode Cnt Score
Error Units
[java] BTreeBuildBench.buildTreeTest 1 thrpt 16 124595.864 ?
10133.336 ops/ms
[java] BTreeBuildBench.buildTreeTest 2 thrpt 16 120228.601 ?
12859.617 ops/ms
[java] BTreeBuildBench.buildTreeTest 5 thrpt 16 103881.001 ?
8136.400 ops/ms
[java] BTreeBuildBench.buildTreeTest 10 thrpt 16 89141.480 ?
7716.011 ops/ms
[java] BTreeBuildBench.buildTreeTest 20 thrpt 16 67390.602 ?
8057.348 ops/ms
[java] BTreeBuildBench.buildTreeTest 40 thrpt 16 19633.234 ?
1545.773 ops/ms
[java] BTreeBuildBench.buildTreeTest 100 thrpt 16 10334.557 ?
1027.898 ops/ms
[java] BTreeBuildBench.buildTreeTest 1000 thrpt 16 1239.163 ?
173.303 ops/ms
[java] BTreeBuildBench.buildTreeTest 10000 thrpt 16 104.024 ?
12.069 ops/ms
[java] BTreeBuildBench.buildTreeTest 100000 thrpt 16 10.259 ?
1.088 ops/ms
== pre-calculate child size and split the left values to the last 2 nodes:
[java] Benchmark (dataSize) Mode Cnt Score
Error Units
[java] BTreeBuildBench.buildTreeTest 1 thrpt 16 122030.330 ?
10528.782 ops/ms
[java] BTreeBuildBench.buildTreeTest 2 thrpt 16 121939.935 ?
12627.014 ops/ms
[java] BTreeBuildBench.buildTreeTest 5 thrpt 16 104694.942 ?
9031.935 ops/ms
[java] BTreeBuildBench.buildTreeTest 10 thrpt 16 87687.949 ?
9029.432 ops/ms
[java] BTreeBuildBench.buildTreeTest 20 thrpt 16 67941.722 ?
7099.874 ops/ms
[java] BTreeBuildBench.buildTreeTest 40 thrpt 16 19468.380 ?
1640.993 ops/ms
[java] BTreeBuildBench.buildTreeTest 100 thrpt 16 10503.954 ?
980.228 ops/ms
[java] BTreeBuildBench.buildTreeTest 1000 thrpt 16 1374.558 ?
167.329 ops/ms
[java] BTreeBuildBench.buildTreeTest 10000 thrpt 16 111.364 ?
8.896 ops/ms
[java] BTreeBuildBench.buildTreeTest 100000 thrpt 16 10.728 ?
1.107 ops/ms
{noformat}
I would prefer the clearer code.
> Optimise BTree.Buider
> ---------------------
>
> Key: CASSANDRA-9989
> URL: https://issues.apache.org/jira/browse/CASSANDRA-9989
> Project: Cassandra
> Issue Type: Sub-task
> Reporter: Benedict
> Assignee: Jay Zhuang
> Priority: Minor
> Fix For: 4.x
>
> Attachments: 9989-trunk.txt
>
>
> BTree.Builder could reduce its copying, and exploit toArray more efficiently,
> with some work. It's not very important right now because we don't make as
> much use of its bulk-add methods as we otherwise might, however over time
> this work will become more useful.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org
For additional commands, e-mail: commits-h...@cassandra.apache.org
