WillAyd commented on PR #15041:
URL: https://github.com/apache/arrow/pull/15041#issuecomment-1370288509

   Latest commit isn't idiomatic but pushing for what I think are the "correct" 
numbers first. Here's an interesting comparison of the benchmark for the 
different types:
   
   ```sh
   willayd@willayd-Lemur-Pro:~/clones/arrow/cpp/release$ 
./release/arrow-compute-vector-sort-benchmark 
--benchmark_filter="^ArraySortIndices.*/49152/1$"
   2023-01-03T14:24:23-08:00
   Running ./release/arrow-compute-vector-sort-benchmark
   Run on (12 X 4700 MHz CPU s)
   CPU Caches:
     L1 Data 48 KiB (x6)
     L1 Instruction 32 KiB (x6)
     L2 Unified 1280 KiB (x6)
     L3 Unified 12288 KiB (x1)
   Load Average: 0.83, 3.36, 3.21
   ***WARNING*** CPU scaling is enabled, the benchmark real time measurements 
may be noisy and will incur extra overhead.
   
----------------------------------------------------------------------------------------------
   Benchmark                                    Time             CPU   
Iterations UserCounters...
   
----------------------------------------------------------------------------------------------
   ArraySortIndicesInt64Narrow/49152/1       8.53 us         8.53 us        
81671 bytes_per_second=5.36415G/s items_per_second=719.963M/s null_percent=100 
size=49.152k
   ArraySortIndicesInt64Wide/49152/1         8.53 us         8.53 us        
81569 bytes_per_second=5.36582G/s items_per_second=720.188M/s null_percent=100 
size=49.152k
   ArraySortIndicesBool/49152/1               195 us          195 us         
3555 bytes_per_second=240.604M/s items_per_second=2.01833G/s null_percent=100 
size=49.152k
   ArraySortIndicesString/49152/1            5.01 us         5.01 us       
137257 bytes_per_second=9.13159G/s items_per_second=612.811M/s null_percent=100 
size=49.152k
   ```
   
   Wasn't expecting there to be much of a difference across these - I'm 
guessing the other types may be misreporting their benchmark since they don't 
take into account the proper sizing of an all null array?


-- 
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]

Reply via email to