peter-toth commented on issue #23731: [SPARK-26572][SQL] fix aggregate codegen 
result evaluation
URL: https://github.com/apache/spark/pull/23731#issuecomment-460313209
 
 
   Here are my benchmark results of q55. I run 3 times on master and 3 times on 
this PR branch against `scale=5` generated data.
   Master:
   ```
   master:
     Stopped after 5 iterations, 29324 ms
   
   Java HotSpot(TM) 64-Bit Server VM 1.8.0_162-b12 on Mac OS X 10.14.2
   Intel(R) Core(TM) i7-4870HQ CPU @ 2.50GHz
   TPCDS Snappy:                            Best/Avg Time(ms)    Rate(M/s)   
Per Row(ns)   Relative
   
------------------------------------------------------------------------------------------------
   q55                                           5683 / 5865          2.6       
  391.2       1.0X
   
     Stopped after 5 iterations, 28914 ms
   
   Java HotSpot(TM) 64-Bit Server VM 1.8.0_162-b12 on Mac OS X 10.14.2
   Intel(R) Core(TM) i7-4870HQ CPU @ 2.50GHz
   TPCDS Snappy:                            Best/Avg Time(ms)    Rate(M/s)   
Per Row(ns)   Relative
   
------------------------------------------------------------------------------------------------
   q55                                           5584 / 5783          2.6       
  384.3       1.0X
   
     Stopped after 5 iterations, 29905 ms
   
   Java HotSpot(TM) 64-Bit Server VM 1.8.0_162-b12 on Mac OS X 10.14.2
   Intel(R) Core(TM) i7-4870HQ CPU @ 2.50GHz
   TPCDS Snappy:                            Best/Avg Time(ms)    Rate(M/s)   
Per Row(ns)   Relative
   
------------------------------------------------------------------------------------------------
   q55                                           5873 / 5981          2.5       
  404.3       1.0X
   ```
   This PR:
   ```
   this PR:
     Stopped after 5 iterations, 32577 ms
   
   Java HotSpot(TM) 64-Bit Server VM 1.8.0_162-b12 on Mac OS X 10.14.2
   Intel(R) Core(TM) i7-4870HQ CPU @ 2.50GHz
   TPCDS Snappy:                            Best/Avg Time(ms)    Rate(M/s)   
Per Row(ns)   Relative
   
------------------------------------------------------------------------------------------------
   q55                                           6226 / 6515          2.3       
  428.5       1.0X
   
   
     Stopped after 5 iterations, 30612 ms
   
   Java HotSpot(TM) 64-Bit Server VM 1.8.0_162-b12 on Mac OS X 10.14.2
   Intel(R) Core(TM) i7-4870HQ CPU @ 2.50GHz
   TPCDS Snappy:                            Best/Avg Time(ms)    Rate(M/s)   
Per Row(ns)   Relative
   
------------------------------------------------------------------------------------------------
   q55                                           5792 / 6122          2.5       
  398.6       1.0X
   
     Stopped after 5 iterations, 32918 ms
   
   Java HotSpot(TM) 64-Bit Server VM 1.8.0_162-b12 on Mac OS X 10.14.2
   Intel(R) Core(TM) i7-4870HQ CPU @ 2.50GHz
   TPCDS Snappy:                            Best/Avg Time(ms)    Rate(M/s)   
Per Row(ns)   Relative
   
------------------------------------------------------------------------------------------------
   q55                                           6415 / 6584          2.3       
  441.5       1.0X
   ```
   Although the results are a bit varying, it seems this patch would introduce 
some performance degradation.
   I will try to modify the patch to evaluate only `Stateful` expressions as 
@maropu suggested and run the benchmark again.

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
[email protected]


With regards,
Apache Git Services

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to