[
https://issues.apache.org/jira/browse/BEAM-9085?focusedWorklogId=400818&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-400818
]
ASF GitHub Bot logged work on BEAM-9085:
----------------------------------------
Author: ASF GitHub Bot
Created on: 10/Mar/20 16:24
Start Date: 10/Mar/20 16:24
Worklog Time Spent: 10m
Work Description: kamilwu commented on issue #11092: [BEAM-9085] Fix
performance regression in SyntheticSource
URL: https://github.com/apache/beam/pull/11092#issuecomment-597180032
R: @tvalentyn
I did many tests and gathered the results in one single document:
https://docs.google.com/document/d/1AegjUCc5w4B90_rvR8WAfIeL65PHUi4oGYkodgPS0o0/edit?usp=sharing.
The previous solution (https://github.com/apache/beam/pull/10885) didn't work
very well on big `element_size`, which was most probably the cause of failures
and slowdowns.
Although this PR is still a bit slower than we have now using numpy 1.16 and
Python 2.7, I'd rather avoid downgrading numpy via supplying an additional
requirements.txt file to a Dataflow job. According to this document
(https://numpy.org/neps/nep-0029-deprecation_policy.html), the support for
numpy 1.16 will be dropped on Jan 13, 2021 - and this looks like a temporary
solution.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to 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: 400818)
Time Spent: 4h (was: 3h 50m)
> Performance regression in np.random.RandomState() skews performance test
> results across Python 2/3 on Dataflow
> --------------------------------------------------------------------------------------------------------------
>
> Key: BEAM-9085
> URL: https://issues.apache.org/jira/browse/BEAM-9085
> Project: Beam
> Issue Type: Bug
> Components: testing
> Reporter: Kamil Wasilewski
> Assignee: Kamil Wasilewski
> Priority: Major
> Time Spent: 4h
> Remaining Estimate: 0h
>
> Tests show that the performance of core Beam operations in Python 3.x on
> Dataflow can be a few time slower than in Python 2.7. We should investigate
> what's the cause of the problem.
> Currently, we have one ParDo test that is run both in Py3 and Py2 [1]. A
> dashboard with runtime results can be found here [2].
> [1] sdks/python/apache_beam/testing/load_tests/pardo_test.py
> [2] https://apache-beam-testing.appspot.com/explore?dashboard=5678187241537536
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
