[
https://issues.apache.org/jira/browse/HIVE-12547?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15086089#comment-15086089
]
Sergey Shelukhin commented on HIVE-12547:
-
[~jdere] can you comment about the distributed hashjoin
[
https://issues.apache.org/jira/browse/HIVE-12547?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15053469#comment-15053469
]
Sergey Shelukhin commented on HIVE-12547:
-
[~mmccline] what do you think? I looked at the code, it
[
https://issues.apache.org/jira/browse/HIVE-12547?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15053547#comment-15053547
]
Matt McCline commented on HIVE-12547:
-
That makes sense.
> VectorMapJoinFastBytesHashTable fails
[
https://issues.apache.org/jira/browse/HIVE-12547?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15047684#comment-15047684
]
Sergey Shelukhin commented on HIVE-12547:
-
Actually, I wonder how we get into this situation. It's
[
https://issues.apache.org/jira/browse/HIVE-12547?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15041438#comment-15041438
]
Matt McCline commented on HIVE-12547:
-
At the time I wrote it I was using long to hold the hash code
[
https://issues.apache.org/jira/browse/HIVE-12547?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15041441#comment-15041441
]
Matt McCline commented on HIVE-12547:
-
(HIVE-12492 MapJoin: 4 million unique integers seems to be a
[
https://issues.apache.org/jira/browse/HIVE-12547?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15042073#comment-15042073
]
Sergey Shelukhin commented on HIVE-12547:
-
Why? Otherwise, we get query failures and the only
[
https://issues.apache.org/jira/browse/HIVE-12547?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15042093#comment-15042093
]
Matt McCline commented on HIVE-12547:
-
2 parallel arrays means a double CPU Data Cache miss.
>
[
https://issues.apache.org/jira/browse/HIVE-12547?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15042154#comment-15042154
]
Sergey Shelukhin commented on HIVE-12547:
-
Beats query failures in my book :P Also, these would be
[
https://issues.apache.org/jira/browse/HIVE-12547?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15042165#comment-15042165
]
Matt McCline commented on HIVE-12547:
-
I want to make sure the 99% normal case when we aren't creating
[
https://issues.apache.org/jira/browse/HIVE-12547?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15042026#comment-15042026
]
Sergey Shelukhin commented on HIVE-12547:
-
Well, that's a separate problem. For this JIRA, I think
[
https://issues.apache.org/jira/browse/HIVE-12547?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15042059#comment-15042059
]
Matt McCline commented on HIVE-12547:
-
No I don't think we should do that.
>
[
https://issues.apache.org/jira/browse/HIVE-12547?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15038855#comment-15038855
]
Sergey Shelukhin commented on HIVE-12547:
-
[~mmccline] is there a reason why 3 longs are used per
[
https://issues.apache.org/jira/browse/HIVE-12547?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15034429#comment-15034429
]
Sergey Shelukhin commented on HIVE-12547:
-
It's probably time to use multiple arrays.
>
14 matches
Mail list logo