Github user saketj commented on the issue:

    https://github.com/apache/incubator-quickstep/pull/114
  
    @pateljm Sure Jignesh, I have rebased this PR now with the master.
    
    Thanks @hbdeshmukh for helping in re-running the performance numbers for 
this PR over the updated master after recent commits. Here are the updated 
numbers for the entire TPC-H suite of queries now:
    
    |              | Execution time w/o PR (in ms) | Execution time w/ PR (in 
ms) |
    
|--------------|-------------------------------|------------------------------|
    | Query 01.sql | 16,405                        | 14,901                     
  |
    | Query 02.sql | 5,603                         | 5,582                      
  |
    | Query 03.sql | 8,267                         | 6,018                      
  |
    | Query 04.sql | 4,823                         | 2,741                      
  |
    | Query 05.sql | 5,210                         | 4,592                      
  |
    | Query 06.sql | 396                           | 402                        
  |
    | Query 07.sql | 23,907                        | 23,154                     
  |
    | Query 08.sql | 3,324                         | 3,268                      
  |
    | Query 09.sql | 9,641                         | 15,749                     
  |
    | Query 10.sql | 15,215                        | 14,823                     
  |
    | Query 11.sql | 2,141                         | 2,160                      
  |
    | Query 12.sql | 2,479                         | 2,121                      
  |
    | Query 13.sql | 34,369                        | 34,159                     
  |
    | Query 14.sql | 795                           | 822                        
  |
    | Query 15.sql | 4,696                         | 5,109                      
  |
    | Query 16.sql | 9,442                         | 9,040                      
  |
    | Query 17.sql | 149,132                       | 142,230                    
  |
    | Query 18.sql | 81,042                        | 76,434                     
  |
    | Query 19.sql | 1,453                         | 1,564                      
  |
    | Query 20.sql | 53,268                        | 59,553                     
  |
    | Query 21.sql | 119,326                       | 124,905                    
  |
    | Query 22.sql | 6,812                         | 7,187                      
  |
    | Total        | 557745                        | 556515                     
  |
    
    Overall, there is certainly some benefit to be gained by introducing a 
branchless codepath. Most queries seem to improve in general. However, the 
results for Query 09, 20, and 21 are definitely surprising. I will spend some 
time understanding why this is the case.


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---

Reply via email to