wbo4958 commented on code in PR #45794:
URL: https://github.com/apache/spark/pull/45794#discussion_r1546951004
##########
core/src/test/scala/org/apache/spark/scheduler/CoarseGrainedSchedulerBackendSuite.scala:
##########
@@ -179,6 +179,53 @@ class CoarseGrainedSchedulerBackendSuite extends
SparkFunSuite with LocalSparkCo
}
}
+ // Every item corresponds to (CPU resources per task, GPU resources per task,
+ // and the GPU addresses assigned to all tasks).
+ Seq(
+ (1, 1, Array(Array("0"), Array("1"), Array("2"), Array("3"))),
+ (1, 2, Array(Array("0", "1"), Array("2", "3"))),
+ (1, 4, Array(Array("0", "1", "2", "3"))),
+ (2, 1, Array(Array("0"), Array("1"))),
+ (4, 1, Array(Array("0"))),
+ (4, 2, Array(Array("0", "1"))),
+ (2, 2, Array(Array("0", "1"), Array("2", "3"))),
Review Comment:
Yeah, that's true, it will generate the deterministic result. The internal
implementation will offer the sorted GPU addresses to the tasks (to align with
the old version)
--
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]
