[ 
https://issues.apache.org/jira/browse/CASSANDRA-8714?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14346508#comment-14346508
 ] 

Robert Stupp commented on CASSANDRA-8714:
-----------------------------------------

Hm - the micro bench itself tells that neither approach has any benefit.
The microbench uses direct and heap BBs. _branch_ means the patch and _trunk_ - 
yea trunk.
{noformat:title=64kB}
     [java] Benchmark                                                 Mode  
Samples  Score   Error   Units
     [java] o.a.c.t.m.SetBytesAllocationBench.branch                 thrpt      
 10  0,957 ± 0,036  ops/us
     [java] o.a.c.t.m.SetBytesAllocationBench.branch:branchDirect    thrpt      
 10  0,483 ± 0,013  ops/us
     [java] o.a.c.t.m.SetBytesAllocationBench.branch:branchHeap      thrpt      
 10  0,474 ± 0,025  ops/us
     [java] o.a.c.t.m.SetBytesAllocationBench.trunk                  thrpt      
 10  0,938 ± 0,053  ops/us
     [java] o.a.c.t.m.SetBytesAllocationBench.trunk:trunkDirect      thrpt      
 10  0,473 ± 0,024  ops/us
     [java] o.a.c.t.m.SetBytesAllocationBench.trunk:trunkHeap        thrpt      
 10  0,466 ± 0,030  ops/us
{noformat}
{noformat:title=1kB}
     [java] Benchmark                                                 Mode  
Samples   Score   Error   Units
     [java] o.a.c.t.m.SetBytesAllocationBench.branch                 thrpt      
 10  58,268 ± 2,936  ops/us
     [java] o.a.c.t.m.SetBytesAllocationBench.branch:branchDirect    thrpt      
 10  30,058 ± 1,649  ops/us
     [java] o.a.c.t.m.SetBytesAllocationBench.branch:branchHeap      thrpt      
 10  28,209 ± 1,525  ops/us
     [java] o.a.c.t.m.SetBytesAllocationBench.trunk                  thrpt      
 10  55,066 ± 2,410  ops/us
     [java] o.a.c.t.m.SetBytesAllocationBench.trunk:trunkDirect      thrpt      
 10  28,169 ± 1,449  ops/us
     [java] o.a.c.t.m.SetBytesAllocationBench.trunk:trunkHeap        thrpt      
 10  26,897 ± 1,206  ops/us
{noformat}


> row-cache: use preloaded jemalloc w/ Unsafe
> -------------------------------------------
>
>                 Key: CASSANDRA-8714
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-8714
>             Project: Cassandra
>          Issue Type: Sub-task
>            Reporter: Robert Stupp
>            Assignee: Robert Stupp
>             Fix For: 3.0
>
>         Attachments: 8714-2.txt, 8714-3.txt, 8714-4.txt, 8714.txt
>
>
> Using jemalloc via Java's {{Unsafe}} is a better alternative on Linux than 
> using jemalloc via JNA.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to