View source for Deflate

{{Short description|Data compression algorithm}}
{{About|the data compression algorithm||Deflation (disambiguation)}}
{{Original research|date=May 2025}}
{{Infobox file format
| name = Deflate
| icon = 
| iconcaption = 
| icon_size = 
| screenshot = 
| screenshot_size = 
| caption = 
|_noextcode = 
| extension = <!-- or: | extensions = -->
|_nomimecode = 
| mime = 
| type_code = 
| uniform_type = 
| conforms_to = 
| magic = 
| max_size = 
| developer = [[Phil Katz]], [[PKWare]]
| released = {{Start date and age|1990|08|21|df=yes}}
| latest_release_version = 
| latest_release_date = <!-- {{Start date and age|202y|mm|dd|df=yes}} -->
| type = 
| compression = [[LZ77 and LZ78|LZ77]], [[Huffman coding]]
| container_for = 
| contained_by = 
| extended_from = 
| extended_to = 
| standard = <!-- or: | standards = -->
| free = Yes
| open = Yes
| url = {{URL|www.pkware.com}}
}}

In [[computing]], '''Deflate''' (stylized as '''DEFLATE''', and also called '''Flate'''<ref>{{cite web |author1=The Go Authors |title=flate package - compress/flate - Go Packages |url=https://pkg.go.dev/compress/flate |website=The Go Programming Language |publisher=Google |access-date=5 September 2023 |quote=Package flate implements the Deflate compressed data format, described in RFC issue 1951.}}</ref><ref>{{cite web |author1=<!-- Human Name --> |author1-link= |title=PDF 32000-1:2008: Document management — Portable document format — Part 1: PDF 1.7 |url=https://opensource.adobe.com/dc-acrobat-sdk-docs/pdfstandards/PDF32000_2008.pdf |website=Adobe Open Source |publisher=[[Adobe Inc.]] |access-date=5 September 2023 |page=23 |quote=FlateDecode [...] Decompresses data encoded using the zlib/deflate compression method}}</ref>) is a [[Lossless compression|lossless]] [[data compression]] [[file format]] that uses a combination of [[LZ77 and LZ78|LZ77]] and [[Huffman coding]]. It was designed by [[Phil Katz]], for version 2 of his [[PKZIP]] archiving tool. Deflate was later specified in [[Request for Comments]] (RFC) 1951 (1996).<ref name="IETF">{{cite IETF |last1=Deutsch |first1=L. Peter |author1-link=L. Peter Deutsch |date=May 1996 |title=Deflate Compressed Data Format Specification version 1.3 |rfc=1951 |section=Abstract |page=1 |publisher=[[Internet Engineering Task Force]] (IETF) |access-date=2014-04-23}}</ref>

Katz also designed the original [[algorithm]] used to construct Deflate streams. This algorithm received [[software patent]] {{US patent|5051745}}, assigned to [[PKWare]], Inc.<ref name="patent">{{cite patent |country=US |number=5051745 |inventor1-last=Katz |inventor1-first=Phillip W. |inventorlink1=Phil Katz |status=patent |publication-date=1991-09-24 |issue-date=1991-09-24 |title=String Searcher, and Compressor Using Same |assign1=PKWare Inc.}}</ref><ref>{{cite book |last1=Salomon |first1=David |year=2007 |title=Data Compression: The Complete Reference |edition=4 |page=241 |publisher=Springer |isbn=978-1-84628-602-5 |url=https://books.google.com/books?id=ujnQogzx_2EC&pg=PA241}}</ref> As stated in the RFC document, an algorithm producing Deflate files was widely thought to be implementable in a manner not covered by patents.<ref name="IETF"/> This led to its widespread use. For example, in [[gzip]] compressed files and Portable Network Graphics ([[PNG]]) image files, in addition to the [[ZIP (file format)|ZIP]] file format for which Katz originally designed it. The patent has since expired.

== Stream format ==
A Deflate stream consists of a series of blocks. Each block is preceded by a 3-[[bit]] header:
* First bit: Last-block-in-stream marker:
** <code>1</code>: This is the last block in the stream.
** <code>0</code>: There are more blocks to process after this one.
* Second and third bits: Encoding method used for this block type:
** <code>00</code>: A stored (a.k.a. raw or literal) section, between 0 and 65,535 bytes in length
** <code>01</code>: A ''static Huffman'' compressed block, using a pre-agreed [[Huffman coding|Huffman tree]] defined in the RFC
** <code>10</code>: A ''dynamic Huffman'' compressed block, complete with the Huffman table supplied
** <code>11</code>: Reserved: don't use

The ''stored'' block option adds minimal overhead and is used for data that is incompressible.

Most compressible data will end up being encoded using method <code>10</code>, the ''dynamic Huffman'' encoding, which produces an optimized Huffman tree customized for each block of data individually. Instructions to generate the necessary Huffman tree immediately follow the block header. The static Huffman option is used for short messages, where the fixed saving gained by omitting the tree outweighs the percentage compression loss due to using a non-optimal (thus, not technically Huffman) code.

Compression is achieved through two steps:
* Matching and replacing duplicate strings with pointers
* Replacing symbols with new, weighted symbols based on use frequency

=== Duplicate string elimination ===
{{Further|LZ77 and LZ78|LZSS}}
Within compressed blocks, if a duplicate series of bytes is spotted (a repeated string), then a back-[[Reference (computer science)|reference]] is inserted, linking to the prior location of that identical string instead. An encoded match to an earlier string consists of an [[8-bit computing|8-bit]] length (3–258 bytes) and a 15-bit distance (1–32,768 bytes) to the start of the duplicate. Relative back-references can be made across any number of blocks, as long as the distance appears within the last 32&nbsp;[[Kibibyte|KiB]] of uncompressed data decoded (termed the ''sliding window'').

If the distance is less than the length, the duplicate overlaps itself, indicating repetition. For example, a run of 10 identical bytes can be encoded as one byte, followed by a duplicate of length 9, starting with the prior byte.

Searching the preceding text for duplicate substrings is the most computationally expensive part of the Deflate algorithm, and the operation which compression level settings affect.

=== Bit reduction ===
{{Further|Huffman coding}}
The second compression stage consists of replacing commonly used symbols with shorter representations and less commonly used symbols with longer representations. The method used is [[Huffman coding]] which creates an unprefixed tree of non-overlapping intervals, where the length of each sequence is inversely proportional to the logarithm of the probability of that symbol needing to be encoded. The more likely it is that a symbol has to be encoded, the shorter its bit-sequence will be.

A tree is created, containing space for 288 symbols:
* 0–255: represent the literal bytes/symbols 0–255.
* 256: end of block – stop processing if last block, otherwise start processing next block.
* 257–285: combined with extra-bits, a match length of 3–258 bytes.
* 286, 287: not used, reserved and illegal but still part of the tree.

A match length code will always be followed by a distance code. Based on the distance code read, further "extra" bits may be read in order to produce the final distance. The distance tree contains space for 32 symbols:
* 0–3: distances 1–4
* 4–5: distances 5–8, 1 extra bit
* 6–7: distances 9–16, 2 extra bits
* 8–9: distances 17–32, 3 extra bits
* ...
* 26–27: distances 8,193–16,384, 12 extra bits
* 28–29: distances 16,385–32,768, 13 extra bits
* 30–31: not used, reserved and illegal but still part of the tree

For the match distance symbols 2–29, the number of extra bits can be calculated as <math>\left\lfloor\frac{n}{2}\right\rfloor-1</math>.

The two codes (the 288-symbol length/literal tree and the 32-symbol distance tree) are themselves encoded as [[canonical Huffman code]]s by giving the bit length of the code for each symbol. The bit lengths are themselves [[Run-length encoding|run-length encoded]] to produce as compact a representation as possible. As an alternative to including the tree representation, the "static tree" option provides standard fixed Huffman trees. The compressed size using the static trees can be computed using the same statistics (the number of times each symbol appears) as are used to generate the dynamic trees, so it is easy for a compressor to choose whichever is smaller.

== Encoder–compressor ==
During the compression stage, it is the ''encoder'' that chooses the amount of time spent looking for matching strings. The zlib/gzip [[reference implementation]] allows the user to select from a sliding scale of likely resulting compression-level vs. speed of encoding. Options range from <code>0</code> (do not attempt compression, just store uncompressed) to <code>9</code> representing the maximum capability of the reference implementation in zlib/gzip.

Other Deflate encoders have been produced, all of which will also produce a compatible [[bitstream]] capable of being decompressed by any existing Deflate decoder. Differing implementations will likely produce variations on the final encoded bit-stream produced. The focus with non-zlib versions of an encoder has normally been to produce a more efficiently compressed and smaller encoded stream.

=== Deflate64 ===
Deflate64, specified by PKWARE, is a proprietary variant of Deflate. It's fundamentally the same algorithm. What has changed is the increase in dictionary size from 32&nbsp;KB to 64&nbsp;KB, an extension of the distance codes to 16&nbsp;bits so that they may address a range of 64&nbsp;KB, and the length code, which is extended to [[16-bit computing|16-bit]], so that it may define lengths of three to 65,538 bytes.<ref>{{Cite web |url=http://www.binaryessence.com/dct/imp/en000225.htm |title=Binary Essence – Deflate64 |access-date=22 May 2011 |archive-url=https://web.archive.org/web/20170621195505/http://www.binaryessence.com/dct/imp/en000225.htm |archive-date=21 June 2017 |url-status=bot: unknown}}</ref> This leads to Deflate64 having a longer compression time, and potentially a slightly higher compression ratio, than Deflate.<ref>{{Cite web |url=http://www.binaryessence.com/dct/apc/en000263.htm |title=Binary Essence – "Calgary Corpus" compression comparisons |access-date=22 May 2011 |archive-url=https://web.archive.org/web/20171227131819/http://www.binaryessence.com/dct/apc/en000263.htm |archive-date=27 December 2017 |url-status=bot: unknown}}</ref> Several free and/or open source projects support Deflate64, such as [[7-Zip]],<ref>{{Cite web|url=https://sevenzip.osdn.jp/chm/cmdline/switches/method.htm|title=-m (Set compression Method) switch|website=sevenzip.osdn.jp|access-date=2023-01-21|archive-date=2022-04-09|archive-url=https://web.archive.org/web/20220409225619/https://sevenzip.osdn.jp/chm/cmdline/switches/method.htm|url-status=dead}}</ref> while others, such as [[zlib]], do not, because the procedure is proprietary,<ref>History of Lossless Data Compression Algorithms – [https://ieeeghn.org/wiki/index.php/History_of_Lossless_Data_Compression_Algorithms#DEFLATE64 Deflate64]</ref> and the performance increase over Deflate is small.<ref>zlib FAQ – [https://www.zlib.net/zlib_faq.html#faq40 Does zlib support the new "Deflate64" format introduced by PKWare?]</ref>

== Using Deflate in new software ==
Implementations of Deflate are freely available in many languages. Apps written in [[C (programming language)|C]] typically use the [[zlib]] [[Library (computing)|library]] (under the permissive [[zlib License]]). Apps in [[Borland Pascal]] (and compatible languages) can use paszlib. Apps in [[C++]] can take advantage of the improved Deflate library in [[7-Zip]]. Both [[Java (software platform)|Java]] and [[.NET]] framework offer out-of-the-box support for Deflate in their libraries (respectively, <code>java.util.zip</code> and [https://docs.microsoft.com/en-us/dotnet/api/system.io.compression.deflatestream System.IO.Compression]). Apps in [[Ada (programming language)|Ada]] can use [https://unzip-ada.sourceforge.net/ Zip-Ada] (pure) or [https://zlib-ada.sourceforge.net/ ZLib-Ada].

=== Encoder implementations ===
* [[PKZIP]]: the first implementation, originally done by [[Phil Katz]] as part of PKZip
* [[zlib]]: standard reference implementation adopted in many apps because of its open-source, permissive license. See {{section link|Zlib#Forks}} for higher-performance forks.
* [[Crypto++]]: contains a public-domain implementation in [[C++]] aimed at reducing potential [[Vulnerability (computing)|security vulnerabilities]]. The author, Wei Dai states "<cite>This code is less clever, but hopefully more understandable and maintainable [than zlib]</cite>".
* [[7-Zip]]: written by Igor Pavlov in [[C++]], this version is freely licensed and achieves higher compression than zlib at the expense of [[central processing unit]] (CPU) use. Has an option to use the Deflate64 storage format.
* [[PuTTY]] 'sshzlib.c': a standalone implementation under the [[MIT License]] by Simon Tatham, it has full decoding capability, but only supports static tree only creation
* libflate:<ref>{{Cite web |url=http://plan9.bell-labs.com/sources/plan9/sys/src/libflate/ |title=Plan 9 from Bell Labs's /n/sources/plan9/sys/src/libflate |website=plan9.bell-labs.com |publisher=Lucent Technologies |archive-date=2006-03-15 |archive-url=https://web.archive.org/web/20060315063934/http://plan9.bell-labs.com/sources/plan9/sys/src/libflate/ |url-status=dead}}</ref> part of [[Plan 9 from Bell Labs]], implements deflate compression
* [[Red Gate Software#HyperBac|Hyperbac]]: uses its own proprietary compression library (in C++ and assembly) with an option to implement the Deflate64 storage format
* [[Zopfli]]: C implementation under the [[Apache License]] by [[Google]]; achieves higher compression at the expense of CPU use. ZopfliPNG is a variant of Zopfli for use with [[Portable Network Graphics|PNG]] files.
* igzip: an encoder written in the [[x86 assembly language]], released by [[Intel]] under the [[MIT License]]. 3x faster than zlib -1. Useful for compressing genomic data.<ref>{{cite web |title=High Performance Deflate Compression with Optimizations for Genomic Data Sets |url=https://software.intel.com/en-us/articles/igzip-a-high-performance-deflate-compressor-with-optimizations-for-genomic-data |website=Intel Software |access-date=18 January 2020 |date=1 October 2019}}</ref>
* libdeflate:<ref>{{cite web |title=libdeflate |url=https://github.com/ebiggers/libdeflate |website=Heavily optimized library for DEFLATE/zlib/gzip compression and decompression}}</ref> a library for fast, whole-buffer Deflate-based compression and decompression. Libdeflate is heavily optimized, especially on x86 processors.

AdvanceCOMP uses the higher compression ratio versions of Deflate in 7-Zip, libdeflate, and Zopfli to enable recompression of [[gzip]], [[Portable Network Graphics|PNG]], [[multiple-image Network Graphics]] (MNG) and [[ZIP (file format)|ZIP]] files with the possibility of smaller file sizes than zlib is able to achieve at maximum settings.<ref>{{cite web |last1=Mazzoleni |first1=Andrea |date=21 February 2023 |title=amadvance/advancecomp |website=[[GitHub]] |url=https://github.com/amadvance/advancecomp/blob/fcf71a89265c78fc26243574dda3a872574a5c02/doc/advzip.txt}}</ref>

=== Hardware encoders ===
* AHA361-PCIX/AHA362-PCIX from [https://www.aha.com/ Comtech AHA] {{Webarchive|url=https://web.archive.org/web/20061208005415/http://www.aha.com/ |date=2006-12-08}}. Comtech produced a [[PCI-X]] card (PCI-ID: <code>193f:0001</code>) able to compress streams using Deflate at a rate of up to 3.0&nbsp;Gbit/s (375&nbsp;MB/s) for incoming uncompressed data. Accompanying the [[Linux kernel]] [[device driver]] for the AHA361-PCIX is an "<code>ahagzip</code>" utility and customized "<code>mod_deflate_aha</code>" able to use the hardware compression from [[Apache HTTP Server|Apache]]. The hardware is based on a [[Xilinx]] [[Virtex (FPGA)|Virtex]] [[field-programmable gate array]] (FPGA) and four custom AHA3601 [[application-specific integrated circuit]]s (ASICs). The AHA361/AHA362 boards are limited to only handling static Huffman blocks and require software to be modified to add support. The cards could not support the full Deflate specification, meaning they could only reliably decode their own output (a stream that did not contain any dynamic Huffman type 2 blocks).
* [https://web.archive.org/web/20080204071759/https://www.indranetworks.com/SC300.html StorCompress 300]/[https://www.indranetworks.com/SCMX3.html MX3] from [https://www.indranetworks.com/ Indra Networks]. This is a range of [[Peripheral Component Interconnect]] (PCI, PCI-ID: <code>17b4:0011</code>) or PCI-X cards featuring between one and six compression engines with claimed processing speeds of up to 3.6&nbsp;Gbit/s (450&nbsp;MB/s). A version of the cards are available with the separate brand ''WebEnhance'' specifically designed for web-serving use rather than [[storage area network]] (SAN) or backup use; a [[PCI Express]] (PCIe) revision, the [http://www.indranetworks.com/SCMX4E.html MX4E] is also produced.
* [https://web.archive.org/web/20080912222617/http://www.aha.com/show_prod.php?id=36 AHA363-PCIe]/[https://web.archive.org/web/20090212202014/http://www.aha.com/show_prod.php?id=37 AHA364-PCIe]/[https://web.archive.org/web/20090820184941/http://www.aha.com/show_prod.php?id=38 AHA367-PCIe]. In 2008, Comtech started producing two PCIe cards (<code>PCI-ID: 193f:0363</code>/<code>193f:0364</code>) with a new hardware AHA3610 encoder chip. The new chip was designed to be capable of a sustained 2.5&nbsp;Gbit/s. Using two of these chips, the AHA363-PCIe board can process Deflate at a rate of up to 5.0&nbsp;Gbit/s (625&nbsp;MB/s) using the two channels (two compression and two decompression). The AHA364-PCIe variant is an encode-only version of the card designed for out-going [[load balancer]]s and instead has multiple register sets to allow 32 independent ''virtual'' compression channels feeding two physical compression engines. Linux, [[Microsoft Windows]], and [[OpenSolaris]] kernel device drivers are available for both of the new cards, along with a modified zlib system library so that dynamically linked applications can automatically use the hardware support without internal modification. The AHA367-PCIe board (<code>PCI-ID: 193f:0367</code>) is similar to the AHA363-PCIe but uses four AHA3610 chips for a sustained compression rate of 10&nbsp;Gbit/s (1250&nbsp;MB/s). Unlike the AHA362-PCIX, the decompression engines on the AHA363-PCIe and AHA367-PCIe boards are fully deflate compliant.
* [https://web.archive.org/web/20101203144755/http://www.cavium.com/processor_security_nitrox-III.html Nitrox] and [https://github.com/zerix/Cavium-SDK-2.0/tree/master/examples/zip Octeon]{{Dead link|date=November 2019 |bot=InternetArchiveBot |fix-attempted=yes}} processors from [https://cavium.com Cavium, Inc.] contain high-speed hardware deflate and inflate engines compatible with both ZLIB and GZIP with some devices able to handle multiple simultaneous data streams.
* [https://github.com/tomtor/HDL-deflate HDL-Deflate] GPL FPGA implementation.
*[https://www.cast-inc.com/compression/gzip-lossless-data-compression/zipaccel-c/ ZipAccel-C] from [https://www.cast-inc.com/ CAST Inc]. This is a Silicon IP core supporting Deflate, [[Zlib]] and [[Gzip]] compression. ZipAccel-C can be implemented in [[Application-specific integrated circuit|ASIC]] or [[field-programmable gate array]] (FPGAs), supports both Dynamic and Static Huffman tables, and can provide throughputs in excess of 100&nbsp;Gbit/s. The company offers compression/decompression accelerator board reference designs for Intel FPGA ([https://www.cast-inc.com/compression/gzip-lossless-data-compression/gzip-rd-int/ ZipAccel-RD-INT]) and Xilinx FPGAs ([https://www.cast-inc.com/compression/gzip-lossless-data-compression/gzip-rd-xil/ ZipAccel-RD-XIL]).
* [[Intel Xeon chipsets|Intel Communications Chipset 89xx Series]] (Cave Creek) for the [[Intel]] [[Xeon]] E5-2600 and E5-2400 Processor Series (Sandy Bridge-EP/EN) supports hardware compression and decompression using QuickAssist Technology. Depending on the chipset, compression and decompression rates of 5&nbsp;Gbit/s, 10&nbsp;Gbit/s, or 20&nbsp;Gbit/s are available.<ref name="quickassist">{{cite web
|url=https://www-ssl.intel.com/content/www/us/en/intelligent-systems/crystal-forest-server/embedded-intel-xeon-e5-2600-and-e5-2400-series-with-intel-communications-chipset-89xx.html
|title=Intel Xeon Processor E5-2600 and E5-2400 Series with Intel Communications Chipset 89xx Series
|access-date=2016-05-18
}}</ref>
* [[IBM z15 (microprocessor)|IBM z15]] CPUs incorporate an improved version of the Nest Accelerator Unit (NXU) hardware acceleration from the zEDC Express [[input/output]] (I/O) expansion cards used in z14 systems for hardware Deflate compression and decompression as specified by RFC1951.<ref name="z15_announce">{{cite web
|url=https://www.ibm.com/common/ssi/cgi-bin/ssialias?subtype=ca&infotype=an&supplier=877&letternum=ENUSZG19-0041
|title=Introducing the IBM z15 - The enterprise platform for mission-critical hybrid multicloud
|website=[[IBM]]
|date=12 September 2019
|access-date=2021-11-01
}}</ref><ref name="z15_techmanual">{{cite book
|url=https://books.google.com/books?id=0vr3DwAAQBAJ&dq=%22Nest+accelerator%22&pg=PA97
|title=IBM z15 (8562) Technical Guide, Page 97
|isbn=9780738458991
|access-date=2021-11-01
|last1=Lascu
|first1=Octavian
|date=28 April 2021
|publisher=IBM Redbooks
}}</ref>
* Starting with the [[POWER9]] architecture, IBM added hardware support for compressing and decompressing Deflate (as specified by RFC 1951) to the formerly crypto-centric Nest accelerator (NX) core introduced with [[POWER7]]+. This support is available to programs running with [[IBM AIX|AIX]] 7.2 Technology Level 4 Expansion Pack or AIX 7.2 Technology Level 5 Service Pack 2 through the zlibNX library.<ref name="zlibnx">{{cite web
|url=https://www.ibm.com/docs/en/aix/7.2?topic=management-data-compression-by-using-zlibnx-library
|title=Data compression by using the zlibNX library - IBM Documentation
|website=[[IBM]]
|access-date=2021-11-01
}}</ref><ref name="power7_accel">{{cite web
|url=https://community.ibm.com/community/user/power/blogs/xinya-wang1/2021/02/18/exploitation-of-nest-accelerators-and-in-core-acce
|title=Exploitation of In-Core Acceleration of POWER Processors for AIX
|access-date=2021-11-01
}}</ref>

== Decoder, decompressor ==
Inflate is the decoding process that takes a Deflate bitstream for decompression and correctly produces the original full-size data or file.

=== Inflate-only implementations ===
The normal intent with an alternative Inflate implementation is highly optimized decoding speed, or extremely predictable [[random-access memory]] (RAM) use for [[microcontroller]] [[embedded system]]s.

* [[Assembly language|Assembly]]
** [https://github.com/pfusik/zlib6502 6502 inflate], written by Piotr Fusik in [[MOS Technology 6502|6502]] assembly language.
** [https://sourceforge.net/projects/samflate/ SAMflate], written by Andrew Collier in [[Zilog Z80]] assembly language with optional memory paging support for the [[SAM Coupé]], and released under a combination of [[software license]]s: [[Berkeley Software Distribution]] ([[BSD licenses|BSD]]), [[GNU General Public License]] (GPL), [[GNU Lesser General Public License]] (LGPL), [[Debian Free Software Guidelines]] (DFSG).
** [https://web.archive.org/web/20160304053236/https://bitbucket.org/grauw/gunzip gunzip], written by Laurens Holst in [[Zilog Z80|Z80]] assembly language for the [[MSX]], licensed under [[BSD licenses|BSD]].
** [https://github.com/keirf/Amiga-Stuff inflate.asm], a fast and efficient implementation in [[Motorola 68000]] machine language, written by Keir Fraser and released into the [[public domain]].
<!-- ** <code>PCUNZP.ASM</code>, by Michael Mefford. written in [[x86]] [[assembly language]] and published in [[PCMag|PC Magazine]] 1992-03-31. need to check, might not have supported the newer ''Deflate'' method -->

* [[C (programming language)|C]], [[C++]]
** [https://web.archive.org/web/20070927122958/http://www.mikekohn.net/file_formats/kunzip.php kunzip] by Michael Kohn and unrelated to "KZIP". Comes with [[C (programming language)|C]] [[source code]] under the [[GNU Lesser General Public License]] (LGPL). Used in the GNU Image Manipulation Program ([[GIMP]]) installer.
** puff.c ([[zlib]]), a small, unencumbered, single-file reference implementation included in the /contrib/puff directory of the zlib distribution.
** [http://www.ibsensoftware.com/download.html tinf] written by Jørgen Ibsen in [[ANSI C]] and comes with zlib license. Adds about 2k code.
** [http://code.google.com/p/miniz/source/browse/trunk/tinfl.c tinfl.c] ([http://code.google.com/p/miniz/ miniz]), Public domain Inflate implementation contained entirely in a single C function.
* <code>PCDEZIP</code>, Bob Flanders and Michael Holmes, published in [[PCMag|PC Magazine]] 1994-01-11.
* [http://opensource.franz.com/deflate/ inflate.cl] by John Foderaro. Self-standing [[Common Lisp]] decoder distributed with a [[GNU Lesser General Public License]] (LGPL).
* [http://www.paul.sladen.org/projects/pyflate/ pyflate], a pure-[[Python (programming language)|Python]] stand-alone Deflate ([[gzip]]) and [[bzip2]] decoder by Paul Sladen. Written for research/prototyping and released under a combination of [[software license]]s: [[Berkeley Software Distribution]] ([[BSD licenses|BSD]]), [[GNU General Public License]] (GPL), [[GNU Lesser General Public License]] (LGPL), [[Debian Free Software Guidelines]] (DFSG).
* [http://lua-users.org/wiki/ModuleCompressDeflateLua deflatelua], a pure-[[Lua (programming language)|Lua]] implementation of Deflate and [[gzip]]/zlib decompression, by David Manura.
* [https://github.com/chrisdickinson/inflate inflate] a pure-[[JavaScript]] implementation of Inflate by Chris Dickinson
* [https://github.com/nodeca/pako pako]: JavaScript speed-optimized port of zlib. Contains separate build with inflate only.

=== Hardware decoders ===
* [https://web.archive.org/web/20171010000403/http://www.bitsim.com/en/our-design-model/#Blocks Serial Inflate GPU] from BitSim. Hardware implementation of Inflate. Part of the Bitsim Accelerated Display Graphics Engine (BADGE) controller offering for embedded systems.
* [https://github.com/tomtor/HDL-deflate HDL-Deflate] GPL FPGA implementation.
* [https://www.cast-inc.com/compression/gzip-lossless-data-compression/zipaccel-d/ ZipAccel-D] from [https://www.cast-inc.com/ CAST Inc]. This is a Silicon IP core supporting decompression of Deflate, [[Zlib]] and [[Gzip]] files. The ZipAccel-D IP core that can be implemented in [[Application-specific integrated circuit|ASIC]] or [[Field-programmable gate array|FPGAs]]. The company offers compression/decompression accelerator board reference designs for Intel FPGA ([https://www.cast-inc.com/compression/gzip-lossless-data-compression/gzip-rd-int/ ZipAccel-RD-INT]) and Xilinx FPGAs ([https://www.cast-inc.com/compression/gzip-lossless-data-compression/gzip-rd-xil/ ZipAccel-RD-XIL]).
* [[IBM z15 (microprocessor)|IBM z15]] CPUs incorporate an improved version of the Nest Accelerator Unit (NXU) hardware acceleration from the zEDC Express [[input/output]] (I/O) expansion cards used in z14 systems for hardware Deflate compression and decompression as specified by RFC1951.<ref name="z15_announce"/><ref name="z15_techmanual"/>
* Starting with the [[POWER9]] architecture, IBM added hardware support for compressing and decompressing Deflate (as specified by RFC 1951) to the formerly crypto-centric Nest accelerator (NX) core introduced with [[POWER7]]+. This support is available to programs running with [[IBM AIX|AIX]] 7.2 Technology Level 4 Expansion Pack or AIX 7.2 Technology Level 5 Service Pack 2 through the zlibNX library.<ref name="zlibnx"/><ref name="power7_accel"/>

== See also ==
* [[List of archive formats]]
* [[List of file archivers]]
* [[Comparison of file archivers]]

== References ==
{{Reflist}}

== External links ==
* [[PKWare]], Inc.'s <code>appnote.txt</code>, [http://www.pkware.com/documents/casestudies/APPNOTE.TXT ''.ZIP File Format Specification''] {{Webarchive|url=https://web.archive.org/web/20141205201932/http://www.pkware.com/documents/casestudies/APPNOTE.TXT |date=2014-12-05}}; Section 10, ''X. Deflating – Method 8''.<!-- Really "Section 10?, not "5.5 Deflating - Method 8"? Does this need an archive.org-url?-->
* {{IETF RFC|1951|link=no}} – ''Deflate Compressed Data Format Specification version 1.3''
* [https://www.zlib.net zlib Home Page]
* [https://zlib.net/feldspar.html ''An Explanation of the Deflate Algorithm''] – by Antaeus Feldspar
* [http://www.larsson.dogma.net/dccpaper.pdf ''Extended Application of Suffix Trees to Data Compression''] {{Webarchive|url=https://web.archive.org/web/20160923065920/http://www.larsson.dogma.net/dccpaper.pdf |date=2016-09-23}} – an excellent algorithm to implement Deflate by Jesper Larsson
* [https://www.hanshq.net/zip.html Zip Files: History, Explanation and Implementation] – walk-through of a Deflate implementation

{{Compression methods}}

[[Category:Lossless compression algorithms]]
[[Category:Data compression]]