cloud-fan commented on PR #59027:
URL: https://github.com/apache/spark/pull/59027#issuecomment-6554108959
Thanks for running these on a cluster. Two notes. The config doc at the
commit I reviewed (53307903c4e) still cites the fixed ratios (0.8X-1.2X,
0.7X-0.8X, 0.5X-0.6X), so the rewording you describe isn't in that commit;
could you push it so it can be checked against the results files? Also, your
runs still show the boxed-key gap: with double keys the range join takes about
17-19% longer than the nested loop join (2.7 against 2.3 minutes at 5% overlap,
31 against 26 at 90%), while long keys are at parity or faster. That is much
smaller than in the single-JVM benchmark, but it points the same way in both
double-key cases, so it doesn't look like noise.
<!-- SPARK_DEV_REVIEW_REPLY
{"feedback_id":"conversation:6544954116","thread_id":"conversation:6544954116","verdict_sha256":"d02bae59a6e33bf74b1c2bfa9b6b1ffa9ae19b97a003ba17aef99205783a7b82"}
-->
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
