vertexclique edited a comment on pull request #8664:
URL: https://github.com/apache/arrow/pull/8664#issuecomment-732497888
@nevi-me @alamb @jorgecarleitao
The good news, I have found a solution for the performance related
considerations. I have experimented on the `sum` and my roofline analysis
brought some good results. Also, criterion benches are here:
Before (current master):
```
sum 2^20 time: [900.13 us 902.01 us 904.02 us]
Found 9 outliers among 100 measurements (9.00%)
2 (2.00%) low mild
3 (3.00%) high mild
4 (4.00%) high severe
sum nulls 2^20 time: [2.5859 ms 2.5909 ms 2.5967 ms]
Found 8 outliers among 100 measurements (8.00%)
2 (2.00%) high mild
6 (6.00%) high severe
```
After:
```
sum 2^20 time: [236.61 us 238.02 us 239.58 us]
change: [-73.888% -73.699% -73.493%] (p = 0.00 <
0.05)
Performance has improved.
Found 11 outliers among 100 measurements (11.00%)
1 (1.00%) low mild
5 (5.00%) high mild
5 (5.00%) high severe
sum nulls 2^20 time: [549.14 us 551.39 us 554.07 us]
change: [-78.784% -78.671% -78.548%] (p = 0.00 <
0.05)
Performance has improved.
Found 12 outliers among 100 measurements (12.00%)
7 (7.00%) high mild
5 (5.00%) high severe
```
Since it is a time-consuming task, I am not going to perform a full rewrite
until we agree that this performance improvement is enough.
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
