Google's new Brotli compression algorithm beats Zopfli by up to 26%
- 23 September, 2015 13:59
Google's Compression Team has unveiled a new algorithm that compresses and decompresses data better than its Zopfli.
The Brotli algorithm, meaning 'small bread' in Swiss German, which produces smaller compressed files than other current algorithms, beats its predecessor Zopfli with a 20-26 per cent higher compression ratio.
“We hope that this format will be supported by major browsers in the near future, as the smaller compressed size would give additional benefits to mobile users, such as lower data transfer fees and reduced battery use,” wrote Zoltan Szabadka, software engineer at Google, in a blog post.
It produces smaller file sizes than Gzip, which is the most common compression algorithm for compressing Web content sent to a browser.
It's decompression is much faster than current compression algorithms, which means compressed content is decompressed and rendered more quickly in the browser.
The algorithm uses a different data format to the Zopfli deflate data format. Szabadka describes this data format as “2nd order context modeling, re-use of entropy codes, larger memory window of past data and joint distribution codes”, resulting in higher data density.
What also makes it different to other algorithms is it includes a 122,784-byte sized static dictionary of 13,504 words or syllables in English, Spanish, Chinese, Hindi, Russian and Arabic. Some 121 transforms (takes into account variations to a base word in the static dictionary) are used to construct 1,633,984 sequences of words.
Szabadka tested his algorithm against Zopfli and other compression algorithms on the Canterbury corpus, an ad hoc crawled Web content corpus of 1,285 files totalling 70,611,753 bytes in size, and the enwik8 corpus used in the Hutter prize.
“While Zopfli is now well accepted in the field, there were opinions expressed that we should move on from the deflate file format to a modern solution. Brotli is our attempt at building a compression format and an example implementation of this format that is fundamentally more efficient than deflate [Zopfli],” he wrote in a paper on the Brotli algorithm.
The algorithm has been released as open source on GitHub.