[
https://issues.apache.org/jira/browse/ARTEMIS-1977?focusedWorklogId=190104&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-190104
]
ASF GitHub Bot logged work on ARTEMIS-1977:
-------------------------------------------
Author: ASF GitHub Bot
Created on: 25/Jan/19 16:57
Start Date: 25/Jan/19 16:57
Worklog Time Spent: 10m
Work Description: franz1981 commented on issue #2523: ARTEMIS-1977
ASYNCIO can reduce sys-calls to retrieve I/O events
URL: https://github.com/apache/activemq-artemis/pull/2523#issuecomment-457639282
@michaelandrepearce I suppose this one could be of some interest to you :)
@orpiske Please mate I need some help to compile this one for 32 bit, I'm
not the good to configure cross-compilation on my box and I haven't an old box
to do it...
If it won't work for 32 bit architecture I can put a flag to just skip the
optimization :+1:
It should reduce (for very fast disks) the amount of context switches while
reading the completed IO events: with Specter/Meltdown increasing the cost of
context switches it should brings some nice improvement/less CPU usage for free.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
Issue Time Tracking
-------------------
Worklog Id: (was: 190104)
Time Spent: 0.5h (was: 20m)
> ASYNCIO can reduce sys-calls to retrieve I/O events
> ---------------------------------------------------
>
> Key: ARTEMIS-1977
> URL: https://issues.apache.org/jira/browse/ARTEMIS-1977
> Project: ActiveMQ Artemis
> Issue Type: Improvement
> Components: Broker
> Affects Versions: 2.6.2
> Reporter: Francesco Nigro
> Assignee: Francesco Nigro
> Priority: Minor
> Fix For: 2.7.0
>
> Time Spent: 0.5h
> Remaining Estimate: 0h
>
> On LibAIO is possible to retrieve the I/O completion events without
> using io_getevents sys-calls by reading the user-space ring buffer used by the
> kernel to store them.
> This is already beneficial for very fast disks and necessary for further
> improvements
> of ASYNCIO Journal to leverage (very) fast low-latency disks by going
> completly
> lock-free.
>
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
