Baunsgaard commented on PR #15791:
URL: https://github.com/apache/iceberg/pull/15791#issuecomment-4162201699

   Since i was asked for a benchmark. Here is the results, and links to related 
PRs in the other subpackages:
   
   Related PRs:
   - Java: https://github.com/apache/iceberg/pull/15791
   - C++: https://github.com/apache/iceberg-cpp/pull/608
   - Rust: https://github.com/apache/iceberg-rust/pull/2292
   
   
   ### Benchmark results
   
   `setRange()` (native bulk range) vs a loop of individual `set()` calls
   (JDK 17, 5 iterations after 3 warmup):
   
   | Scenario | Before (set loop) | After (setRange) | Speedup |
   |---|---|---|---|
   | 100 positions | 37.8 µs | 2.4 µs | **16x** |
   | 10k positions | 618.8 µs | 2.4 µs | **258x** |
   | 200k positions | 2.53 ms | 4.9 µs | **514x** |
   | 1M positions | 7.38 ms | 21.5 µs | **344x** |
   | 10k cross-key boundary | 208.9 µs | 2.0 µs | **104x** |
   
   > The cross-key boundary scenario inserts 10k positions spanning the 32-bit 
key boundary (e.g., `(1LL << 32) - 5000` to `(1LL << 32) + 5000`). This 
exercises the key-splitting logic where the range is divided across two 
internal 32-bit Roaring bitmaps.


-- 
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