RamakrishnaChilaka commented on PR #15358:
URL: https://github.com/apache/lucene/pull/15358#issuecomment-3458057687

   I re-ran the micro benchmarks using only the encoding changes (keeping the 
existing `DataOuput` interface unchanged).
   
   ### Benchmarks
   ```
   Run-1
   candidate:
   Benchmark                              (bitsPerValue)   Mode  Cnt   Score   
Error   Units
   ForUtilEncodeBulkIntsBenchmark.encode               2  thrpt    5  14.636 ± 
0.031  ops/us
   ForUtilEncodeBulkIntsBenchmark.encode               4  thrpt    5  11.046 ± 
0.100  ops/us
   ForUtilEncodeBulkIntsBenchmark.encode               8  thrpt    5   7.251 ± 
0.010  ops/us
   ForUtilEncodeBulkIntsBenchmark.encode              12  thrpt    5   4.188 ± 
0.008  ops/us
   ForUtilEncodeBulkIntsBenchmark.encode              16  thrpt    5   3.992 ± 
0.009  ops/us
   ForUtilEncodeBulkIntsBenchmark.encode              20  thrpt    5   2.133 ± 
0.004  ops/us
   ForUtilEncodeBulkIntsBenchmark.encode              24  thrpt    5   2.206 ± 
0.002  ops/us
   ForUtilEncodeBulkIntsBenchmark.encode              28  thrpt    5   1.915 ± 
0.002  ops/us
   ForUtilEncodeBulkIntsBenchmark.encode              32  thrpt    5   2.074 ± 
0.005  ops/us
   
   baseline:
   Benchmark                              (bitsPerValue)   Mode  Cnt   Score   
Error   Units
   ForUtilEncodeBulkIntsBenchmark.encode               2  thrpt    5  15.336 ± 
0.234  ops/us
   ForUtilEncodeBulkIntsBenchmark.encode               4  thrpt    5  12.417 ± 
0.010  ops/us
   ForUtilEncodeBulkIntsBenchmark.encode               8  thrpt    5   8.317 ± 
0.014  ops/us
   ForUtilEncodeBulkIntsBenchmark.encode              12  thrpt    5   4.342 ± 
0.006  ops/us
   ForUtilEncodeBulkIntsBenchmark.encode              16  thrpt    5   4.703 ± 
0.002  ops/us
   ForUtilEncodeBulkIntsBenchmark.encode              20  thrpt    5   2.332 ± 
0.003  ops/us
   ForUtilEncodeBulkIntsBenchmark.encode              24  thrpt    5   2.203 ± 
0.003  ops/us
   ForUtilEncodeBulkIntsBenchmark.encode              28  thrpt    5   1.966 ± 
0.007  ops/us
   ForUtilEncodeBulkIntsBenchmark.encode              32  thrpt    5   2.483 ± 
0.052  ops/us
   ```
   
   ### Summary
   After re-benchmarking without the `writeInts()` path, there's no measurable 
speedup (sometimes minor regression). This confirms that the performance gains 
observed in the PR description primarily came from the bulk write optimisation 
itself.
   
   Given that modifying `DataOutput` to introduce Bulk I/O is a significant API 
change, I will close this PR for now.


-- 
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]


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

Reply via email to