1<p align="center"><img src="https://raw.githubusercontent.com/facebook/zstd/dev/doc/images/zstd_logo86.png" alt="Zstandard"></p> 2 3__Zstandard__, or `zstd` as short version, is a fast lossless compression algorithm, 4targeting real-time compression scenarios at zlib-level and better compression ratios. 5It's backed by a very fast entropy stage, provided by [Huff0 and FSE library](https://github.com/Cyan4973/FiniteStateEntropy). 6 7The project is provided as an open-source dual [BSD](LICENSE) and [GPLv2](COPYING) licensed **C** library, 8and a command line utility producing and decoding `.zst`, `.gz`, `.xz` and `.lz4` files. 9Should your project require another programming language, 10a list of known ports and bindings is provided on [Zstandard homepage](http://www.zstd.net/#other-languages). 11 12**Development branch status:** 13 14[![Build Status][travisDevBadge]][travisLink] 15[![Build status][AppveyorDevBadge]][AppveyorLink] 16[![Build status][CircleDevBadge]][CircleLink] 17 18[travisDevBadge]: https://travis-ci.org/facebook/zstd.svg?branch=dev "Continuous Integration test suite" 19[travisLink]: https://travis-ci.org/facebook/zstd 20[AppveyorDevBadge]: https://ci.appveyor.com/api/projects/status/xt38wbdxjk5mrbem/branch/dev?svg=true "Windows test suite" 21[AppveyorLink]: https://ci.appveyor.com/project/YannCollet/zstd-p0yf0 22[CircleDevBadge]: https://circleci.com/gh/facebook/zstd/tree/dev.svg?style=shield "Short test suite" 23[CircleLink]: https://circleci.com/gh/facebook/zstd 24 25## Benchmarks 26 27For reference, several fast compression algorithms were tested and compared 28on a server running Linux Debian (`Linux version 4.14.0-3-amd64`), 29with a Core i7-6700K CPU @ 4.0GHz, 30using [lzbench], an open-source in-memory benchmark by @inikep 31compiled with [gcc] 7.3.0, 32on the [Silesia compression corpus]. 33 34[lzbench]: https://github.com/inikep/lzbench 35[Silesia compression corpus]: http://sun.aei.polsl.pl/~sdeor/index.php?page=silesia 36[gcc]: https://gcc.gnu.org/ 37 38| Compressor name | Ratio | Compression| Decompress.| 39| --------------- | ------| -----------| ---------- | 40| **zstd 1.3.4 -1** | 2.877 | 470 MB/s | 1380 MB/s | 41| zlib 1.2.11 -1 | 2.743 | 110 MB/s | 400 MB/s | 42| brotli 1.0.2 -0 | 2.701 | 410 MB/s | 430 MB/s | 43| quicklz 1.5.0 -1 | 2.238 | 550 MB/s | 710 MB/s | 44| lzo1x 2.09 -1 | 2.108 | 650 MB/s | 830 MB/s | 45| lz4 1.8.1 | 2.101 | 750 MB/s | 3700 MB/s | 46| snappy 1.1.4 | 2.091 | 530 MB/s | 1800 MB/s | 47| lzf 3.6 -1 | 2.077 | 400 MB/s | 860 MB/s | 48 49[zlib]: http://www.zlib.net/ 50[LZ4]: http://www.lz4.org/ 51 52Zstd can also offer stronger compression ratios at the cost of compression speed. 53Speed vs Compression trade-off is configurable by small increments. 54Decompression speed is preserved and remains roughly the same at all settings, 55a property shared by most LZ compression algorithms, such as [zlib] or lzma. 56 57The following tests were run 58on a server running Linux Debian (`Linux version 4.14.0-3-amd64`) 59with a Core i7-6700K CPU @ 4.0GHz, 60using [lzbench], an open-source in-memory benchmark by @inikep 61compiled with [gcc] 7.3.0, 62on the [Silesia compression corpus]. 63 64Compression Speed vs Ratio | Decompression Speed 65---------------------------|-------------------- 66 |  67 68A few other algorithms can produce higher compression ratios at slower speeds, falling outside of the graph. 69For a larger picture including slow modes, [click on this link](doc/images/DCspeed5.png). 70 71 72## The case for Small Data compression 73 74Previous charts provide results applicable to typical file and stream scenarios (several MB). Small data comes with different perspectives. 75 76The smaller the amount of data to compress, the more difficult it is to compress. This problem is common to all compression algorithms, and reason is, compression algorithms learn from past data how to compress future data. But at the beginning of a new data set, there is no "past" to build upon. 77 78To solve this situation, Zstd offers a __training mode__, which can be used to tune the algorithm for a selected type of data. 79Training Zstandard is achieved by providing it with a few samples (one file per sample). The result of this training is stored in a file called "dictionary", which must be loaded before compression and decompression. 80Using this dictionary, the compression ratio achievable on small data improves dramatically. 81 82The following example uses the `github-users` [sample set](https://github.com/facebook/zstd/releases/tag/v1.1.3), created from [github public API](https://developer.github.com/v3/users/#get-all-users). 83It consists of roughly 10K records weighing about 1KB each. 84 85Compression Ratio | Compression Speed | Decompression Speed 86------------------|-------------------|-------------------- 87 |  |  88 89 90These compression gains are achieved while simultaneously providing _faster_ compression and decompression speeds. 91 92Training works if there is some correlation in a family of small data samples. The more data-specific a dictionary is, the more efficient it is (there is no _universal dictionary_). 93Hence, deploying one dictionary per type of data will provide the greatest benefits. 94Dictionary gains are mostly effective in the first few KB. Then, the compression algorithm will gradually use previously decoded content to better compress the rest of the file. 95 96### Dictionary compression How To: 97 981. Create the dictionary 99 100 `zstd --train FullPathToTrainingSet/* -o dictionaryName` 101 1022. Compress with dictionary 103 104 `zstd -D dictionaryName FILE` 105 1063. Decompress with dictionary 107 108 `zstd -D dictionaryName --decompress FILE.zst` 109 110 111## Build instructions 112 113### Makefile 114 115If your system is compatible with standard `make` (or `gmake`), 116invoking `make` in root directory will generate `zstd` cli in root directory. 117 118Other available options include: 119- `make install` : create and install zstd cli, library and man pages 120- `make check` : create and run `zstd`, tests its behavior on local platform 121 122### cmake 123 124A `cmake` project generator is provided within `build/cmake`. 125It can generate Makefiles or other build scripts 126to create `zstd` binary, and `libzstd` dynamic and static libraries. 127 128By default, `CMAKE_BUILD_TYPE` is set to `Release`. 129 130### Meson 131 132A Meson project is provided within [`build/meson`](build/meson). Follow 133build instructions in that directory. 134 135You can also take a look at [`.travis.yml`](.travis.yml) file for an 136example about how Meson is used to build this project. 137 138Note that default build type is **release**. 139 140### Visual Studio (Windows) 141 142Going into `build` directory, you will find additional possibilities: 143- Projects for Visual Studio 2005, 2008 and 2010. 144 + VS2010 project is compatible with VS2012, VS2013, VS2015 and VS2017. 145- Automated build scripts for Visual compiler by [@KrzysFR](https://github.com/KrzysFR), in `build/VS_scripts`, 146 which will build `zstd` cli and `libzstd` library without any need to open Visual Studio solution. 147 148### Buck 149 150You can build the zstd binary via buck by executing: `buck build programs:zstd` from the root of the repo. 151The output binary will be in `buck-out/gen/programs/`. 152 153## Status 154 155Zstandard is currently deployed within Facebook. It is used continuously to compress large amounts of data in multiple formats and use cases. 156Zstandard is considered safe for production environments. 157 158## License 159 160Zstandard is dual-licensed under [BSD](LICENSE) and [GPLv2](COPYING). 161 162## Contributing 163 164The "dev" branch is the one where all contributions are merged before reaching "master". 165If you plan to propose a patch, please commit into the "dev" branch, or its own feature branch. 166Direct commit to "master" are not permitted. 167For more information, please read [CONTRIBUTING](CONTRIBUTING.md). 168
