neilconway opened a new pull request, #24151: URL: https://github.com/apache/datafusion/pull/24151
## Which issue does this PR close? - Related to #23982 ## Rationale for this change The `bounded_window` benchmark only covered aggregate window expressions (`count` and `sum`). Window functions such as `row_number`, `rank`, `lag`, `lead`, and `nth_value` go through a different code path, for which we had no benchmark coverage. Add four new benchmark cases: - `linear dense row_number 10000 partitions`: per-visit fixed costs on partitions that receive rows in every batch. - `linear sparse row_number 32768 partitions`: per-batch work on quiet partitions dominates. - `linear sparse lead 32768 partitions`: a non-causal function, whose result for the last buffered row of a partition stays pending until that partition receives another row. - `linear dense rank 10000 partitions`: an evaluator that compares ORDER BY values row by row. Also rename the existing case names to be consistent. ## What changes are included in this PR? * Add 4 new benchmark cases covering standard (non-agg) window functions * Refactor window benchmark code to enable this * Rename benchmark case names to be mutually consistent ## Are these changes tested? Yes. ## Are there any user-facing changes? No. -- 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]
