1 <p align="center"><img src="https://raw.githubusercontent.com/facebook/zstd/dev/doc/images/zstd_logo86.png" alt="Zstandard"></p>
3 __Zstandard__, or `zstd` as short version, is a fast lossless compression algorithm,
4 targeting real-time compression scenarios at zlib-level and better compression ratios.
5 It's backed by a very fast entropy stage, provided by [Huff0 and FSE library](https://github.com/Cyan4973/FiniteStateEntropy).
7 Zstandard's format is stable and documented in [RFC8878](https://datatracker.ietf.org/doc/html/rfc8878). Multiple independent implementations are already available.
8 This repository represents the reference implementation, provided as an open-source dual [BSD](LICENSE) and [GPLv2](COPYING) licensed **C** library,
9 and a command line utility producing and decoding `.zst`, `.gz`, `.xz` and `.lz4` files.
10 Should your project require another programming language,
11 a list of known ports and bindings is provided on [Zstandard homepage](https://facebook.github.io/zstd/#other-languages).
13 **Development branch status:**
15 [![Build Status][travisDevBadge]][travisLink]
16 [![Build status][CircleDevBadge]][CircleLink]
17 [![Build status][CirrusDevBadge]][CirrusLink]
18 [![Fuzzing Status][OSSFuzzBadge]][OSSFuzzLink]
20 [travisDevBadge]: https://api.travis-ci.com/facebook/zstd.svg?branch=dev "Continuous Integration test suite"
21 [travisLink]: https://travis-ci.com/facebook/zstd
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 [CirrusDevBadge]: https://api.cirrus-ci.com/github/facebook/zstd.svg?branch=dev
25 [CirrusLink]: https://cirrus-ci.com/github/facebook/zstd
26 [OSSFuzzBadge]: https://oss-fuzz-build-logs.storage.googleapis.com/badges/zstd.svg
27 [OSSFuzzLink]: https://bugs.chromium.org/p/oss-fuzz/issues/list?sort=-opened&can=1&q=proj:zstd
31 For reference, several fast compression algorithms were tested and compared
32 on a desktop running Ubuntu 20.04 (`Linux 5.11.0-41-generic`),
33 with a Core i7-9700K CPU @ 4.9GHz,
34 using [lzbench], an open-source in-memory benchmark by @inikep
35 compiled with [gcc] 9.3.0,
36 on the [Silesia compression corpus].
38 [lzbench]: https://github.com/inikep/lzbench
39 [Silesia compression corpus]: https://sun.aei.polsl.pl//~sdeor/index.php?page=silesia
40 [gcc]: https://gcc.gnu.org/
42 | Compressor name | Ratio | Compression| Decompress.|
43 | --------------- | ------| -----------| ---------- |
44 | **zstd 1.5.1 -1** | 2.887 | 530 MB/s | 1700 MB/s |
45 | [zlib] 1.2.11 -1 | 2.743 | 95 MB/s | 400 MB/s |
46 | brotli 1.0.9 -0 | 2.702 | 395 MB/s | 450 MB/s |
47 | **zstd 1.5.1 --fast=1** | 2.437 | 600 MB/s | 2150 MB/s |
48 | **zstd 1.5.1 --fast=3** | 2.239 | 670 MB/s | 2250 MB/s |
49 | quicklz 1.5.0 -1 | 2.238 | 540 MB/s | 760 MB/s |
50 | **zstd 1.5.1 --fast=4** | 2.148 | 710 MB/s | 2300 MB/s |
51 | lzo1x 2.10 -1 | 2.106 | 660 MB/s | 845 MB/s |
52 | [lz4] 1.9.3 | 2.101 | 740 MB/s | 4500 MB/s |
53 | lzf 3.6 -1 | 2.077 | 410 MB/s | 830 MB/s |
54 | snappy 1.1.9 | 2.073 | 550 MB/s | 1750 MB/s |
56 [zlib]: https://www.zlib.net/
57 [lz4]: https://lz4.github.io/lz4/
59 The negative compression levels, specified with `--fast=#`,
60 offer faster compression and decompression speed
61 at the cost of compression ratio (compared to level 1).
63 Zstd can also offer stronger compression ratios at the cost of compression speed.
64 Speed vs Compression trade-off is configurable by small increments.
65 Decompression speed is preserved and remains roughly the same at all settings,
66 a property shared by most LZ compression algorithms, such as [zlib] or lzma.
68 The following tests were run
69 on a server running Linux Debian (`Linux version 4.14.0-3-amd64`)
70 with a Core i7-6700K CPU @ 4.0GHz,
71 using [lzbench], an open-source in-memory benchmark by @inikep
72 compiled with [gcc] 7.3.0,
73 on the [Silesia compression corpus].
75 Compression Speed vs Ratio | Decompression Speed
76 ---------------------------|--------------------
77 ![Compression Speed vs Ratio](doc/images/CSpeed2.png "Compression Speed vs Ratio") | ![Decompression Speed](doc/images/DSpeed3.png "Decompression Speed")
79 A few other algorithms can produce higher compression ratios at slower speeds, falling outside of the graph.
80 For a larger picture including slow modes, [click on this link](doc/images/DCspeed5.png).
83 ## The case for Small Data compression
85 Previous charts provide results applicable to typical file and stream scenarios (several MB). Small data comes with different perspectives.
87 The 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.
89 To solve this situation, Zstd offers a __training mode__, which can be used to tune the algorithm for a selected type of data.
90 Training 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.
91 Using this dictionary, the compression ratio achievable on small data improves dramatically.
93 The 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).
94 It consists of roughly 10K records weighing about 1KB each.
96 Compression Ratio | Compression Speed | Decompression Speed
97 ------------------|-------------------|--------------------
98 ![Compression Ratio](doc/images/dict-cr.png "Compression Ratio") | ![Compression Speed](doc/images/dict-cs.png "Compression Speed") | ![Decompression Speed](doc/images/dict-ds.png "Decompression Speed")
101 These compression gains are achieved while simultaneously providing _faster_ compression and decompression speeds.
103 Training 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_).
104 Hence, deploying one dictionary per type of data will provide the greatest benefits.
105 Dictionary 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.
107 ### Dictionary compression How To:
109 1. Create the dictionary
111 `zstd --train FullPathToTrainingSet/* -o dictionaryName`
113 2. Compress with dictionary
115 `zstd -D dictionaryName FILE`
117 3. Decompress with dictionary
119 `zstd -D dictionaryName --decompress FILE.zst`
122 ## Build instructions
124 `make` is the officially maintained build system of this project.
125 All other build systems are "compatible" and 3rd-party maintained,
126 they may feature small differences in advanced options.
127 When your system allows it, prefer using `make` to build `zstd` and `libzstd`.
131 If your system is compatible with standard `make` (or `gmake`),
132 invoking `make` in root directory will generate `zstd` cli in root directory.
133 It will also create `libzstd` into `lib/`.
135 Other available options include:
136 - `make install` : create and install zstd cli, library and man pages
137 - `make check` : create and run `zstd`, test its behavior on local platform
139 The `Makefile` follows the [GNU Standard Makefile conventions](https://www.gnu.org/prep/standards/html_node/Makefile-Conventions.html),
140 allowing staged install, standard flags, directory variables and command variables.
142 For advanced use cases, specialized compilation flags which control binary generation
143 are documented in [`lib/README.md`](lib/README.md#modular-build) for the `libzstd` library
144 and in [`programs/README.md`](programs/README.md#compilation-variables) for the `zstd` CLI.
148 A `cmake` project generator is provided within `build/cmake`.
149 It can generate Makefiles or other build scripts
150 to create `zstd` binary, and `libzstd` dynamic and static libraries.
152 By default, `CMAKE_BUILD_TYPE` is set to `Release`.
154 #### Support for Fat (Universal2) Output
156 `zstd` can be built and installed with support for both Apple Silicon (M1/M2) as well as Intel by using CMake's Universal2 support.
157 To perform a Fat/Universal2 build and install use the following commands:
160 cmake -B build-cmake-debug -S build/cmake -G Ninja -DCMAKE_OSX_ARCHITECTURES="x86_64;x86_64h;arm64"
168 A Meson project is provided within [`build/meson`](build/meson). Follow
169 build instructions in that directory.
171 You can also take a look at [`.travis.yml`](.travis.yml) file for an
172 example about how Meson is used to build this project.
174 Note that default build type is **release**.
177 You can build and install zstd [vcpkg](https://github.com/Microsoft/vcpkg/) dependency manager:
179 git clone https://github.com/Microsoft/vcpkg.git
182 ./vcpkg integrate install
185 The zstd port in vcpkg is kept up to date by Microsoft team members and community contributors.
186 If the version is out of date, please [create an issue or pull request](https://github.com/Microsoft/vcpkg) on the vcpkg repository.
188 ### Visual Studio (Windows)
190 Going into `build` directory, you will find additional possibilities:
191 - Projects for Visual Studio 2005, 2008 and 2010.
192 + VS2010 project is compatible with VS2012, VS2013, VS2015 and VS2017.
193 - Automated build scripts for Visual compiler by [@KrzysFR](https://github.com/KrzysFR), in `build/VS_scripts`,
194 which will build `zstd` cli and `libzstd` library without any need to open Visual Studio solution.
198 You can build the zstd binary via buck by executing: `buck build programs:zstd` from the root of the repo.
199 The output binary will be in `buck-out/gen/programs/`.
203 You can run quick local smoke tests by running `make check`.
204 If you can't use `make`, execute the `playTest.sh` script from the `src/tests` directory.
205 Two env variables `$ZSTD_BIN` and `$DATAGEN_BIN` are needed for the test script to locate the `zstd` and `datagen` binary.
206 For information on CI testing, please refer to `TESTING.md`.
210 Zstandard is currently deployed within Facebook and many other large cloud infrastructures.
211 It is run continuously to compress large amounts of data in multiple formats and use cases.
212 Zstandard is considered safe for production environments.
216 Zstandard is dual-licensed under [BSD](LICENSE) and [GPLv2](COPYING).
220 The `dev` branch is the one where all contributions are merged before reaching `release`.
221 If you plan to propose a patch, please commit into the `dev` branch, or its own feature branch.
222 Direct commit to `release` are not permitted.
223 For more information, please read [CONTRIBUTING](CONTRIBUTING.md).