Roc DEFLATE library runs within single digits of C libdeflate
Original title: How close can pure functional programming get to C?
On the Silesia corpus, roc-deflate was a median 8.9% slower on Ryzen and 2.6% slower on an M1.
Richard Feldman reports optimizing roc-deflate, a pure Roc DEFLATE package, against libdeflate in C. Across 144 Silesia-corpus benchmarks it was slower on average, with a median gap of 8.9% on a Ryzen 3900X and 2.6% on a 2021 M1 MacBook Pro. Worst cases were about 18% slower and some inputs more than 18% faster. Roc 0.1.0 is still weeks away, and the post says the result may not generalize.
Unconfirmed: The author says it is unknown whether other Roc packages can match this result.
- lean-zip in Lean 4 beat miniz_oxide but not libdeflate.
- roc-deflate has no streaming API yet and was checked for identical output.
- Roc cannot opt out of memory safety, bounds checks, or reference counting.
- The post mentions a Brotli remote code execution found the day before, without a CVE id.
From the original article
8 min read at rtfeldman.comIn Why Lean is Faster Than Rust , the author shares an impressive result: a DEFLATE compression library written in Lean 4 , a functional programming language and proof assistant, outperforming an equivalent Rust package that crates. io says has been downloaded over a billion times. From the article: lean-zip beats every single compression level miniz_oxide provides, i.
By stmonty · rtfeldman.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.