[MAHOUT-1824] Optimize FlinkOpAtA to use upper triangular matrices

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Implemented
• Affects Version/s: 0.11.2
• Fix Version/s: 0.12.0
• Component/s: classic
• Labels: performance
• Epic Link: Samsara-framework
• Assignee: Andrew Palumbo
• Reporter: Suneel Marthi
• Created: 06/Apr/16 07:52
• Updated: 31/Jan/24 22:18
• Resolved: 08/Apr/16 08:08

Description

Optimize FlinkOpAtA to use upper triangular matrices (similar to what's being done in Spark backend).

Presently dals fails on FlinkOpAtA computation with an OOM

57766 [flink-akka.actor.default-dispatcher-5] ERROR akka.actor.ActorSystemImpl  - exception on LARS’ timer thread
java.lang.OutOfMemoryError: GC overhead limit exceeded
57770 [flink-akka.actor.default-dispatcher-5] ERROR akka.actor.ActorSystemImpl  - Uncaught fatal error from thread [flink-scheduler-1] shutting down ActorSystem [flink]
java.lang.OutOfMemoryError: GC overhead limit exceeded
- dals *** FAILED ***
  org.apache.flink.runtime.client.JobTimeoutException: Timeout while waiting for JobManager answer. Job time exceeded 21474835 seconds
  at org.apache.flink.runtime.client.JobClient.submitJobAndWait(JobClient.java:136)
  at org.apache.flink.runtime.minicluster.FlinkMiniCluster.submitJobAndWait(FlinkMiniCluster.scala:423)
  at org.apache.flink.runtime.minicluster.FlinkMiniCluster.submitJobAndWait(FlinkMiniCluster.scala:409)
  at org.apache.flink.runtime.minicluster.FlinkMiniCluster.submitJobAndWait(FlinkMiniCluster.scala:401)
  at org.apache.flink.client.LocalExecutor.executePlan(LocalExecutor.java:190)
  at org.apache.flink.api.java.LocalEnvironment.execute(LocalEnvironment.java:90)
  at org.apache.flink.api.java.ExecutionEnvironment.execute(ExecutionEnvironment.java:855)
  at org.apache.flink.api.scala.ExecutionEnvironment.execute(ExecutionEnvironment.scala:638)
  at org.apache.flink.api.scala.DataSet.collect(DataSet.scala:546)
  at org.apache.mahout.flinkbindings.blas.FlinkOpAtA$.slim(FlinkOpAtA.scala:53)
  ...
  Cause: akka.pattern.AskTimeoutException: Ask timed out on [Actor[akka://flink/user/$a#372851579]] after [21474835000 ms]
  at akka.pattern.PromiseActorRef$$anonfun$1.apply$mcV$sp(AskSupport.scala:333)
  at akka.actor.Scheduler$$anon$7.run(Scheduler.scala:117)
  at akka.actor.LightArrayRevolverScheduler$TaskHolder.run(Scheduler.scala:476)
  at akka.actor.LightArrayRevolverScheduler$$anonfun$close$1.apply(Scheduler.scala:282)
  at akka.actor.LightArrayRevolverScheduler$$anonfun$close$1.apply(Scheduler.scala:281)
  at scala.collection.Iterator$class.foreach(Iterator.scala:727)
  at scala.collection.AbstractIterator.foreach(Iterator.scala:1157)
  at scala.collection.IterableLike$class.foreach(IterableLike.scala:72)
  at scala.collection.AbstractIterable.foreach(Iterable.scala:54)
  at akka.actor.LightArrayRevolverScheduler.close(Scheduler.scala:280)