x-at-01 commented on issue #401: URL: https://github.com/apache/parquet-format/issues/401#issuecomment-5516875195
Regarding Gorilla vs ALP for floating-point data: In benchmarks comparing both approaches on time-series datasets (such as sensor, price, and coordinate data), ALP provides higher throughput and compression ratios because it maps floats to scaled integers before bitpacking: - Performance: In [fastalp](https://github.com/webc-site/wedb_embed/tree/main/fastalp) (available on [crates.io](https://crates.io/crates/fastalp)), decompression achieves 55 to 77 GB/s using register SIMD arithmetic, and bulk compression reaches 6.5 GB/s on f64 arrays. - Compression ratio: Averages 2.29x across the 31 standard datasets from the ALP paper, with automatic raw fallback to prevent negative compression expansion on high-entropy data. -- 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]
