[
https://issues.apache.org/jira/browse/BEAM-6276?focusedWorklogId=178315&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-178315
]
ASF GitHub Bot logged work on BEAM-6276:
----------------------------------------
Author: ASF GitHub Bot
Created on: 23/Dec/18 10:09
Start Date: 23/Dec/18 10:09
Worklog Time Spent: 10m
Work Description: reuvenlax commented on issue #7331: [BEAM-6276] Fix
performance regression.
URL: https://github.com/apache/beam/pull/7331#issuecomment-449626415
A few comments inline:
On Sun, Dec 23, 2018 at 12:41 AM Gleb Kanterov <[email protected]>
wrote:
> @reuvenlax <https://github.com/reuvenlax> I did few hacks, and I was able
> to achieve almost similar performance to hand-written code, it 8-10 times
> faster than now. However, it requires a few changes:
>
> - SchemaProvider#schemaFor needs to be deterministic, could be
> achieved by deterministically ordering fields, for instance, by name.
>
> I wouldn't worry about this one as the schemas are generated at
graph-construction time, not on the workers. It affects the DirectRunner
benchmarks as they only run for a few seconds, but shouldn't matter for
real pipelines
>
> - FromRowUsingCreator should be generated by byte-buddy and statically
> rely on row schema
>
> pr/7289 should help here. It also lets users annotate their own
constructors, in which case we simply generate a call to that.
>
> - SetterBasedCreatorFactory should be generated by byte-buddy, and
> implement SerializationFunction[2-N] instead of Object
> create(Object... params) so that FromRowUsingCreator can efficiently
> use it.
>
> Ah, are the generate typecasts slowing things down?
>
> I didn't dig deep, but it seems after that, generated code gets inlined
> into the equivalent to hand-written code constructions.
>
> —
> You are receiving this because you were mentioned.
> Reply to this email directly, view it on GitHub
> <https://github.com/apache/beam/pull/7331#issuecomment-449622403>, or mute
> the thread
>
<https://github.com/notifications/unsubscribe-auth/AUGE1Y0TbROfvsmo5ZKWxuL5M6Xi9dsnks5u70HTgaJpZM4ZbvGg>
> .
>
----------------------------------------------------------------
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: 178315)
Time Spent: 2.5h (was: 2h 20m)
> Performance regression caused by extra calls to TypeDescriptor.getRawType
> -------------------------------------------------------------------------
>
> Key: BEAM-6276
> URL: https://issues.apache.org/jira/browse/BEAM-6276
> Project: Beam
> Issue Type: Bug
> Components: sdk-java-core
> Reporter: Reuven Lax
> Assignee: Reuven Lax
> Priority: Major
> Time Spent: 2.5h
> Remaining Estimate: 0h
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
