[jira] [Updated] (SPARK-42613) PythonRunner should set OMP_NUM_THREADS to task cpus instead of executor cores by default

["John Zhuge (Jira)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/SPARK-42613?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

John Zhuge updated SPARK-42613:
-------------------------------
    Description: 
Follow up from [https://github.com/apache/spark/pull/40199#discussion_r1119453996]

 

If OMP_NUM_THREADS is not set explicitly, we should set it to `spark.task.cpus` instead of `spark.executor.cores`. Otherwise, we will still have issues when executer core is set to a very large number but task cpus is 1.

  was:
Coming from [https://github.com/apache/spark/pull/40199#discussion_r1119453996]

 

If OMP_NUM_THREADS is not set explicitly, we should set it to `spark.task.cpus` instead of `spark.executor.cores`. Otherwise, we will still have issues when executer core is set to a very large number but task cpus is 1.


> PythonRunner should set OMP_NUM_THREADS to task cpus instead of executor cores by default
> -----------------------------------------------------------------------------------------
>
>                 Key: SPARK-42613
>                 URL: https://issues.apache.org/jira/browse/SPARK-42613
>             Project: Spark
>          Issue Type: Bug
>          Components: PySpark, YARN
>    Affects Versions: 3.3.0
>            Reporter: John Zhuge
>            Priority: Major
>
> Follow up from [https://github.com/apache/spark/pull/40199#discussion_r1119453996]
>  
> If OMP_NUM_THREADS is not set explicitly, we should set it to `spark.task.cpus` instead of `spark.executor.cores`. Otherwise, we will still have issues when executer core is set to a very large number but task cpus is 1.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org