damccorm opened a new pull request, #41126:
URL: https://github.com/apache/beam/pull/41126
### Description
Optimizes `_IntervalWindowBase.__init__` in
`sdks/python/apache_beam/utils/windowed_value.py` and `windowed_value.pxd`:
1. **Fast-path `int` and `Timestamp` constructor arguments:**
- Previously, `_IntervalWindowBase.__init__` unconditionally called
`Timestamp.of(start).micros` and `Timestamp.of(end).micros` when `start` or
`end` were provided.
- Checking `type(start) is Timestamp` and `isinstance(start, int)`
directly populates `_start_micros` and `_end_micros` without function call and
`Timestamp` wrapper allocation overhead.
### Benchmarks & Performance Impact
- **Microbenchmark:** [`bench_03_interval_window_base_init`
(`c2feb961eef`)](https://github.com/damccorm/beam/blob/c2feb961eef2e7d2410825aab9537c62a95e6790/sdks/python/apache_beam/testing/benchmarks/python_sdk_perf/microbenchmarks.py#L96-L104)
(on branch
[`perf/py-benchmarks-suite`](https://github.com/damccorm/beam/tree/perf/py-benchmarks-suite))
— 500,000 `IntervalWindow(start, end)` instantiations with `int` and
`Timestamp` bounds improved by **6.23x (`-83.9%`)**.
- **On-Worker Dataflow Subsystem Benchmark
(`01_sliding_windows_assign_ms`):** In [`WorkerSubsystemBenchDoFn`
(`c2feb961eef`)](https://github.com/damccorm/beam/blob/c2feb961eef2e7d2410825aab9537c62a95e6790/sdks/python/apache_beam/testing/benchmarks/python_sdk_perf/dataflow_e2e_benchmarks.py#L48-L56),
200,000 `SlidingWindows(60, 10).assign` calls (1.2M `IntervalWindow`s) on a
Cloud Dataflow `n2-standard-4` worker improved by **8.82x (`-88.7%`, `6,095 ms`
-> `691 ms`)**.
- **End-to-End Cloud Dataflow Benchmark (`core_window_row_combine`):**
Included in [`run_core_window_row_combine_pipeline`
(`c2feb961eef`)](https://github.com/damccorm/beam/blob/c2feb961eef2e7d2410825aab9537c62a95e6790/sdks/python/apache_beam/testing/benchmarks/python_sdk_perf/dataflow_e2e_benchmarks.py#L274-L302)
(3,000,000 generated events -> 9,000,000 sliding-windowed elements on a single
`n2-standard-4` worker), which improved end-to-end worker stage execution time
by **1.80x (`-44.5%`, `599.6s` -> `332.9s`)** and reduced total billable
Dataflow vCPU/Memory/PD usage by **`-26.1%` (`1,134` -> `838` vCPU-s)**.
### Testing
- Verified with `pytest sdks/python/apache_beam/transforms/window_test.py
sdks/python/apache_beam/utils/windowed_value_test.py`.
--
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]
