kazuk opened a new issue, #1660:
URL: https://github.com/apache/arrow-rs/issues/1660
**Which part is this question about**
I benchmarked the CSV to Parquet conversion workload.
We detected more than 4% performance degradation between versions 11.0 and
12.0 of arrow-rs.
**Describe your question**
Do you know anything about the cause of this performance degradation?
I would like to see data that tracks the performance of this project release.
Is it published somewhere?
**Additional context**
Upgrade parquet/arrow 11.0 to 11.1: No change in performance
Upgrage parquet/arrow 11.1 to 12.0: 4.2% performance degrade.
```
kazuk:~/c2p$ cargo bench --bench convert_perf -- --save-baseline base
Compiling parquet v11.0.0
Compiling arrow v11.0.0
Compiling c2p v1.2.3 (/home/kazuk/c2p)
Finished bench [optimized] target(s) in 1m 33s
Running unittests (target/release/deps/convert_perf-37297eafe0dbd682)
WARNING: HTML report generation will become a non-default optional feature
in Criterion.rs 0.4.0.
This feature is being moved to cargo-criterion
(https://github.com/bheisler/cargo-criterion) and will be optional in a future
version of Criterion.rs. To silence this warning, either switch to
cargo-criterion or enable the 'html_reports' feature in your Cargo.toml.
Gnuplot not found, using plotters backend
Benchmarking csv_to_parquet: Warming up for 3.0000 s
Warning: Unable to complete 100 samples in 5.0s. You may wish to increase
target time to 45.0s, or reduce sample count to 10.
csv_to_parquet time: [430.04 ms 434.42 ms 439.34 ms]
Found 4 outliers among 100 measurements (4.00%)
2 (2.00%) high mild
2 (2.00%) high severe
kazuk:~/c2p$ cargo upgrade [email protected]
Updating 'https://github.com/rust-lang/crates.io-index' index
c2p:
Upgrading arrow v11 -> v11.1.0
kazuk:~/c2p$ cargo upgrade [email protected]
Updating 'https://github.com/rust-lang/crates.io-index' index
c2p:
Upgrading parquet v11 -> v11.1.0
kazuk:~/c2p$ cargo bench --bench convert_perf -- --baseline base
Finished bench [optimized] target(s) in 0.18s
Running unittests (target/release/deps/convert_perf-7067da5dce3ef542)
WARNING: HTML report generation will become a non-default optional feature
in Criterion.rs 0.4.0.
This feature is being moved to cargo-criterion
(https://github.com/bheisler/cargo-criterion) and will be optional in a future
version of Criterion.rs. To silence this warning, either switch to
cargo-criterion or enable the 'html_reports' feature in your Cargo.toml.
Gnuplot not found, using plotters backend
Benchmarking csv_to_parquet: Warming up for 3.0000 s
Warning: Unable to complete 100 samples in 5.0s. You may wish to increase
target time to 45.8s, or reduce sample count to 10.
csv_to_parquet time: [436.29 ms 436.61 ms 436.97 ms]
change: [-0.6269% +0.5052% +1.5525%] (p = 0.37 >
0.05)
No change in performance detected.
Found 7 outliers among 100 measurements (7.00%)
5 (5.00%) high mild
2 (2.00%) high severe
kazuk:~/c2p$ cargo upgrade [email protected]
Updating 'https://github.com/rust-lang/crates.io-index' index
c2p:
Upgrading arrow v11.1.0 -> v12.0.0
kazuk:~/c2p$ cargo upgrade [email protected]
Updating 'https://github.com/rust-lang/crates.io-index' index
c2p:
Upgrading parquet v11.1.0 -> v12.0.0
kazuk:~/c2p$ cargo bench --bench convert_perf -- --baseline base
Finished bench [optimized] target(s) in 0.09s
Running unittests (target/release/deps/convert_perf-505089516b94adaa)
WARNING: HTML report generation will become a non-default optional feature
in Criterion.rs 0.4.0.
This feature is being moved to cargo-criterion
(https://github.com/bheisler/cargo-criterion) and will be optional in a future
version of Criterion.rs. To silence this warning, either switch to
cargo-criterion or enable the 'html_reports' feature in your Cargo.toml.
Gnuplot not found, using plotters backend
Benchmarking csv_to_parquet: Warming up for 3.0000 s
Warning: Unable to complete 100 samples in 5.0s. You may wish to increase
target time to 45.2s, or reduce sample count to 10.
csv_to_parquet time: [448.57 ms 452.73 ms 457.75 ms]
change: [+2.6943% +4.2162% +5.8502%] (p = 0.00 <
0.05)
Performance has regressed.
Found 7 outliers among 100 measurements (7.00%)
2 (2.00%) high mild
5 (5.00%) high severe
```
--
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]
