blob: d8766656f534f4aae27eb5a66bb60fd4adfbe770 [file] [log] [blame] [view]
Yann Collet4b100f42015-10-30 15:49:48 +01001 **Zstd**, short for Zstandard, is a fast lossless compression algorithm, targeting real-time compression scenarios at zlib-level compression ratio.
Yann Collet4856a002015-01-24 01:58:16 +01002
3It is provided as a BSD-license package, hosted on Github.
4
5|Branch |Status |
6|------------|---------|
7|master | [![Build Status](https://travis-ci.org/Cyan4973/zstd.svg?branch=master)](https://travis-ci.org/Cyan4973/zstd) |
8|dev | [![Build Status](https://travis-ci.org/Cyan4973/zstd.svg?branch=dev)](https://travis-ci.org/Cyan4973/zstd) |
9
Yann Collet10ba1832015-10-28 14:05:37 +010010For a taste of its performance, here are a few benchmark numbers from a number of compression codecs suitable for real-time. The test was completed on a Core i7-5600U @ 2.6 GHz, using [fsbench 0.14.3](http://encode.ru/threads/1371-Filesystem-benchmark?p=34029&viewfull=1#post34029), an open-source benchmark program by m^2.
Yann Collet4856a002015-01-24 01:58:16 +010011
Yann Collet4b100f42015-10-30 15:49:48 +010012|Name | Ratio | C.speed | D.speed |
13|-----------------|-------|--------:|--------:|
14| | | MB/s | MB/s |
15| **zstd 0.3** |**2.858**|**280**| **670** |
Yann Collet66d22b82015-11-02 02:36:10 +010016| [zlib] 1.2.8 -1 | 2.730 | 70 | 300 |
Yann Collet4b100f42015-10-30 15:49:48 +010017| QuickLZ 1.5.1b6 | 2.237 | 370 | 415 |
18| LZO 2.06 | 2.106 | 400 | 580 |
19| [LZ4] r131 | 2.101 | 450 | 2100 |
20| Snappy 1.1.0 | 2.091 | 330 | 1100 |
21| LZF 3.6 | 2.077 | 200 | 560 |
Yann Collet56213d82015-08-07 20:15:27 +010022
Yann Collet66d22b82015-11-02 02:36:10 +010023[zlib]:http://www.zlib.net/
Yann Collet56213d82015-08-07 20:15:27 +010024[LZ4]:http://www.lz4.org/
Yann Collet4856a002015-01-24 01:58:16 +010025
Yann Collet66d22b82015-11-02 02:36:10 +010026Zstd can also offer stronger compression ratio at the cost of compression speed. Compression speed is highly configurable, by small increment, to fit different situations. Note however that decompression speed is preserved and remain roughly the same at all settings, a property shared by most LZ compression algorithms, such as [zlib]. The following test is run on a Core i7-3930K CPU @ 4.5GHz, using [lzbench], an open-source in-memory benchmark by inikep.
Yann Collet4856a002015-01-24 01:58:16 +010027
Yann Collet66d22b82015-11-02 02:36:10 +010028[lzbench]:https://github.com/inikep/lzbench
Yann Collet8d8d59e2015-11-02 02:44:43 +010029Compression Ratio vs Speed | Decompression Speed
30---------------------------|--------------------
31![Compression Ratio vs Speed](images/CSpeed.png "Compression Ratio vs Speed") | ![Decompression Speed](images/DSpeed.png "Decompression Speed")
32
Yann Collet4856a002015-01-24 01:58:16 +010033
Yann Colletd5d9bc32015-08-23 23:13:49 +010034Zstd entropy stage is provided by [Huff0 and FSE, from Finite State Entrop library](https://github.com/Cyan4973/FiniteStateEntropy).
Yann Colletf506c8b2015-01-24 10:31:47 +010035
Yann Collet66d22b82015-11-02 02:36:10 +010036Its memory requirement can also be configured to fit into low-memory hardware configurations, or servers handling multiple connections/contexts in parallel.
Yann Colletdb9559f2015-08-24 00:14:40 +010037
Yann Collet66d22b82015-11-02 02:36:10 +010038Zstd has not yet reached "stable format" status. It doesn't guarantee yet that its current compressed format will remain stable and supported in future versions. During this period, it can still change to adapt new optimizations still being investigated. "Stable Format" is projected sometimes early 2016.
39
40That being said, the library is now fairly robust, able to withstand hazards situations, including invalid inputs. The library reliability has been tested using [Fuzz Testing](https://en.wikipedia.org/wiki/Fuzz_testing), with both [internal tools](programs/fuzzer.c) and [external ones](http://lcamtuf.coredump.cx/afl). Therefore, it seems now safe to test Zstandard even within production environments.
Yann Collet4856a002015-01-24 01:58:16 +010041
Yann Collet56213d82015-08-07 20:15:27 +010042### Branch Policy
43The "dev" branch is the one where all contributions will be merged before reaching "master". If you plan to propose a patch, please commit into the "dev" branch or its own feature branch. Direct commit to "master" are not permitted.