Github user ehnalis commented on the pull request:
https://github.com/apache/spark/pull/6082#issuecomment-102139174
I would not accept you evaluation, since as you see you experience a too
high deviation. To converge to 4.8 seconds you would have needed to run it much
more times. Too many other factors can effect run times, as you see, back-off
with 4.8 seconds of advantage actually completed later.
I would refer to this paper to emphasize how many ways things can go wrong:
Characterizing and Subsetting Big Data Workloads -
http://arxiv.org/pdf/1409.0792.pdf
Performance Benefits of DataMPI: A Case Study with BigDataBench -
http://arxiv.org/ftp/arxiv/papers/1403/1403.3480.pdf
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
