Fighting Cancer: The Unexpected Benefit Of Open Sourcing Our Code
Dr. Igor Kozin from The Institute of Cancer Research in London contacted Vlad Krasnov regarding the optimal way to compile CloudFlare's open source fork of zlib. Zlib is widely used for compressing DNA sequencing files, and it turns out that CloudFlare's zlib fork is the best open source solution for this file format. The benefits of open-source software extend far beyond one's imagination, making sophisticated algorithms and software accessible to individuals and organizations without the resources or financial means to develop them independently. CloudFlare's zlib fork includes several improvements over the default implementation, such as using uint64_t as the standard type, an improved hash function, searching for matches of at least 4 bytes instead of the suggested 3 bytes, utilizing SIMD instructions for window rolling, and optimizing the longest-match function. Independent benchmarks show that CloudFlare's implementation outperforms competitors in most settings, with the experimental branch offering even greater performance improvements on certain levels.
Company
Cloudflare
Date published
July 8, 2015
Author(s)
Vlad Krasnov
Word count
1058
Language
English
Hacker News points
154