1 zstd(1) -- zstd, zstdmt, unzstd, zstdcat - Compress or decompress .zst files
2 ============================================================================
7 `zstd` [*OPTIONS*] [-|_INPUT-FILE_] [-o _OUTPUT-FILE_]
9 `zstdmt` is equivalent to `zstd -T0`
11 `unzstd` is equivalent to `zstd -d`
13 `zstdcat` is equivalent to `zstd -dcf`
18 `zstd` is a fast lossless compression algorithm and data compression tool,
19 with command line syntax similar to `gzip (1)` and `xz (1)`.
20 It is based on the **LZ77** family, with further FSE & huff0 entropy stages.
21 `zstd` offers highly configurable compression speed,
22 with fast modes at > 200 MB/s per core,
23 and strong modes nearing lzma compression ratios.
24 It also features a very fast decoder, with speeds > 500 MB/s per core.
26 `zstd` command line syntax is generally similar to gzip,
27 but features the following differences :
29 - Source files are preserved by default.
30 It's possible to remove them automatically by using the `--rm` command.
31 - When compressing a single file, `zstd` displays progress notifications
32 and result summary by default.
33 Use `-q` to turn them off.
34 - `zstd` does not accept input from console,
35 but it properly accepts `stdin` when it's not the console.
36 - `zstd` displays a short help page when command line is an error.
37 Use `-q` to turn it off.
39 `zstd` compresses or decompresses each _file_ according to the selected
41 If no _files_ are given or _file_ is `-`, `zstd` reads from standard input
42 and writes the processed data to standard output.
43 `zstd` will refuse to write compressed data to standard output
44 if it is a terminal : it will display an error message and skip the _file_.
45 Similarly, `zstd` will refuse to read compressed data from standard input
48 Unless `--stdout` or `-o` is specified, _files_ are written to a new file
49 whose name is derived from the source _file_ name:
51 * When compressing, the suffix `.zst` is appended to the source filename to
52 get the target filename.
53 * When decompressing, the `.zst` suffix is removed from the source filename to
54 get the target filename
56 ### Concatenation with .zst files
57 It is possible to concatenate `.zst` files as is.
58 `zstd` will decompress such files as if they were a single `.zst` file.
63 ### Integer suffixes and special values
64 In most places where an integer argument is expected,
65 an optional suffix is supported to easily indicate large integers.
66 There must be no space between the integer and the suffix.
69 Multiply the integer by 1,024 (2\^10).
70 `Ki`, `K`, and `KB` are accepted as synonyms for `KiB`.
72 Multiply the integer by 1,048,576 (2\^20).
73 `Mi`, `M`, and `MB` are accepted as synonyms for `MiB`.
76 If multiple operation mode options are given,
77 the last one takes effect.
81 This is the default operation mode when no operation mode option is specified
82 and no other operation mode is implied from the command name
83 (for example, `unzstd` implies `--decompress`).
84 * `-d`, `--decompress`, `--uncompress`:
87 Test the integrity of compressed _files_.
88 This option is equivalent to `--decompress --stdout` except that the
89 decompressed data is discarded instead of being written to standard output.
90 No files are created or removed.
92 Benchmark file(s) using compression level #
94 Use FILEs as a training set to create a dictionary.
95 The training set should contain a lot of small files (> 100).
97 Display information related to a zstd compressed file, such as size, ratio, and checksum.
98 Some of these fields may not be available.
99 This command can be augmented with the `-v` modifier.
101 ### Operation modifiers
104 `#` compression level \[1-19] (default: 3)
106 switch to ultra-fast compression levels.
107 If `=#` is not present, it defaults to `1`.
108 The higher the value, the faster the compression speed,
109 at the cost of some compression ratio.
110 This setting overwrites compression level if one was set previously.
111 Similarly, if a compression level is set after `--fast`, it overrides it.
113 unlocks high compression levels 20+ (maximum 22), using a lot more memory.
114 Note that decompression will also require more memory when using these levels.
116 enables long distance matching with `#` `windowLog`, if not `#` is not
117 present it defaults to `27`.
118 This increases the window size (`windowLog`) and memory usage for both the
119 compressor and decompressor.
120 This setting is designed to improve the compression ratio for files with
121 long matches at a large distance.
123 Note: If `windowLog` is set to larger than 27, `--long=windowLog` or
124 `--memory=windowSize` needs to be passed to the decompressor.
125 * `-T#`, `--threads=#`:
126 Compress using `#` working threads (default: 1).
127 If `#` is 0, attempt to detect and use the number of physical CPU cores.
128 In all cases, the nb of threads is capped to ZSTDMT_NBTHREADS_MAX==200.
129 This modifier does nothing if `zstd` is compiled without multithread support.
131 Does not spawn a thread for compression, use a single thread for both I/O and compression.
132 In this mode, compression is serialized with I/O, which is slightly slower.
133 (This is different from `-T1`, which spawns 1 compression thread in parallel of I/O).
134 This mode is the only one available when multithread support is disabled.
135 Single-thread mode features lower memory usage.
136 Final compressed result is slightly different from `-T1`.
137 * `--adapt[=min=#,max=#]` :
138 `zstd` will dynamically adapt compression level to perceived I/O conditions.
139 Compression level adaptation can be observed live by using command `-v`.
140 Adaptation can be constrained between supplied `min` and `max` levels.
141 The feature works when combined with multi-threading and `--long` mode.
142 It does not work with `--single-thread`.
143 It sets window size to 8 MB by default (can be changed manually, see `wlog`).
144 Due to the chaotic nature of dynamic adaptation, compressed result is not reproducible.
145 _note_ : at the time of this writing, `--adapt` can remain stuck at low speed
146 when combined with multiple worker threads (>=2).
148 use `file` as Dictionary to compress or decompress FILE(s)
150 do not store dictionary ID within frame header (dictionary compression).
151 The decoder will have to rely on implicit knowledge about which dictionary to use,
152 it won't be able to check if it's correct.
154 save result into `file` (only possible with a single _INPUT-FILE_)
156 overwrite output without prompting, and (de)compress symbolic links
158 force write to standard output, even if it is the console
160 enable / disable sparse FS support,
161 to make files with many zeroes smaller on disk.
162 Creating sparse files may save disk space and speed up decompression by
163 reducing the amount of disk I/O.
164 default: enabled when output is into a file,
165 and disabled when output is stdout.
166 This setting overrides default and can force sparse mode over stdout.
168 remove source file(s) after successful compression or decompression
170 keep source file(s) after successful compression or decompression.
171 This is the default behavior.
173 operate recursively on dictionaries
175 compress and decompress in other formats. If compiled with
176 support, zstd can compress to or decompress from other compression algorithm
177 formats. Possibly available options are `zstd`, `gzip`, `xz`, `lzma`, and `lz4`.
178 If no such format is provided, `zstd` is the default.
179 * `-h`/`-H`, `--help`:
180 display help/long help and exit
182 display version number and exit.
183 Advanced : `-vV` also displays supported formats.
184 `-vvV` also displays POSIX support.
188 suppress warnings, interactivity, and notifications.
189 specify twice to suppress errors too.
190 * `-C`, `--[no-]check`:
191 add integrity check computed from uncompressed data (default: enabled)
193 All arguments after `--` are treated as files
198 `zstd` offers _dictionary_ compression,
199 which greatly improves efficiency on small files and messages.
200 It's possible to train `zstd` with a set of samples,
201 the result of which is saved into a file called a `dictionary`.
202 Then during compression and decompression, reference the same dictionary,
203 using command `-D dictionaryFileName`.
204 Compression of small files similar to the sample set will be greatly improved.
207 Use FILEs as training set to create a dictionary.
208 The training set should contain a lot of small files (> 100),
209 and weight typically 100x the target dictionary size
210 (for example, 10 MB for a 100 KB dictionary).
212 Supports multithreading if `zstd` is compiled with threading support.
213 Additional parameters can be specified with `--train-fastcover`.
214 The legacy dictionary builder can be accessed with `--train-legacy`.
215 The cover dictionary builder can be accessed with `--train-cover`.
216 Equivalent to `--train-fastcover=d=8,steps=4`.
218 Dictionary saved into `file` (default name: dictionary).
220 Limit dictionary to specified size (default: 112640).
222 Use `#` compression level during training (optional).
223 Will generate statistics more tuned for selected compression level,
224 resulting in a _small_ compression ratio improvement for this level.
226 Split input files in blocks of size # (default: no split)
228 A dictionary ID is a locally unique ID that a decoder can use to verify it is
229 using the right dictionary.
230 By default, zstd will create a 4-bytes random number ID.
231 It's possible to give a precise number instead.
232 Short numbers have an advantage : an ID < 256 will only need 1 byte in the
233 compressed frame header, and an ID < 65536 will only need 2 bytes.
234 This compares favorably to 4 bytes default.
235 However, it's up to the dictionary manager to not assign twice the same ID to
236 2 different dictionaries.
237 * `--train-cover[=k#,d=#,steps=#,split=#]`:
238 Select parameters for the default dictionary builder algorithm named cover.
239 If _d_ is not specified, then it tries _d_ = 6 and _d_ = 8.
240 If _k_ is not specified, then it tries _steps_ values in the range [50, 2000].
241 If _steps_ is not specified, then the default value of 40 is used.
242 If _split_ is not specified or split <= 0, then the default value of 100 is used.
243 Requires that _d_ <= _k_.
245 Selects segments of size _k_ with highest score to put in the dictionary.
246 The score of a segment is computed by the sum of the frequencies of all the
247 subsegments of size _d_.
248 Generally _d_ should be in the range [6, 8], occasionally up to 16, but the
249 algorithm will run faster with d <= _8_.
250 Good values for _k_ vary widely based on the input data, but a safe range is
252 If _split_ is 100, all input samples are used for both training and testing
253 to find optimal _d_ and _k_ to build dictionary.
254 Supports multithreading if `zstd` is compiled with threading support.
258 `zstd --train-cover FILEs`
260 `zstd --train-cover=k=50,d=8 FILEs`
262 `zstd --train-cover=d=8,steps=500 FILEs`
264 `zstd --train-cover=k=50 FILEs`
266 `zstd --train-cover=k=50,split=60 FILEs`
268 * `--train-fastcover[=k#,d=#,f=#,steps=#,split=#,accel=#]`:
269 Same as cover but with extra parameters _f_ and _accel_ and different default value of split
270 If _split_ is not specified, then it tries _split_ = 75.
271 If _f_ is not specified, then it tries _f_ = 20.
272 Requires that 0 < _f_ < 32.
273 If _accel_ is not specified, then it tries _accel_ = 1.
274 Requires that 0 < _accel_ <= 10.
275 Requires that _d_ = 6 or _d_ = 8.
277 _f_ is log of size of array that keeps track of frequency of subsegments of size _d_.
278 The subsegment is hashed to an index in the range [0,2^_f_ - 1].
279 It is possible that 2 different subsegments are hashed to the same index, and they are considered as the same subsegment when computing frequency.
280 Using a higher _f_ reduces collision but takes longer.
284 `zstd --train-fastcover FILEs`
286 `zstd --train-fastcover=d=8,f=15,accel=2 FILEs`
288 * `--train-legacy[=selectivity=#]`:
289 Use legacy dictionary builder algorithm with the given dictionary
290 _selectivity_ (default: 9).
291 The smaller the _selectivity_ value, the denser the dictionary,
292 improving its efficiency but reducing its possible maximum size.
293 `--train-legacy=s=#` is also accepted.
297 `zstd --train-legacy FILEs`
299 `zstd --train-legacy=selectivity=8 FILEs`
306 benchmark file(s) using compression level #
308 benchmark file(s) using multiple compression levels, from `-b#` to `-e#` (inclusive)
310 minimum evaluation time, in seconds (default: 3s), benchmark mode only
311 * `-B#`, `--block-size=#`:
312 cut file(s) into independent blocks of size # (default: no block)
314 set process priority to real-time
316 **Output Format:** CompressionLevel#Filename : IntputSize -> OutputSize (CompressionRatio), CompressionSpeed, DecompressionSpeed
318 **Methodology:** For both compression and decompression speed, the entire input is compressed/decompressed in-memory to measure speed. A run lasts at least 1 sec, so when files are small, they are compressed/decompressed several times per run, in order to improve measurement accuracy.
320 ADVANCED COMPRESSION OPTIONS
321 ----------------------------
322 ### --zstd[=options]:
323 `zstd` provides 22 predefined compression levels.
324 The selected or default predefined compression level can be changed with
325 advanced compression options.
326 The _options_ are provided as a comma-separated list.
327 You may specify only the options you want to change and the rest will be
328 taken from the selected or default compression level.
329 The list of available _options_:
331 - `strategy`=_strat_, `strat`=_strat_:
332 Specify a strategy used by a match finder.
334 There are 8 strategies numbered from 1 to 8, from faster to stronger:
335 1=ZSTD\_fast, 2=ZSTD\_dfast, 3=ZSTD\_greedy, 4=ZSTD\_lazy,
336 5=ZSTD\_lazy2, 6=ZSTD\_btlazy2, 7=ZSTD\_btopt, 8=ZSTD\_btultra.
338 - `windowLog`=_wlog_, `wlog`=_wlog_:
339 Specify the maximum number of bits for a match distance.
341 The higher number of increases the chance to find a match which usually
342 improves compression ratio.
343 It also increases memory requirements for the compressor and decompressor.
344 The minimum _wlog_ is 10 (1 KiB) and the maximum is 30 (1 GiB) on 32-bit
345 platforms and 31 (2 GiB) on 64-bit platforms.
347 Note: If `windowLog` is set to larger than 27, `--long=windowLog` or
348 `--memory=windowSize` needs to be passed to the decompressor.
350 - `hashLog`=_hlog_, `hlog`=_hlog_:
351 Specify the maximum number of bits for a hash table.
353 Bigger hash tables cause less collisions which usually makes compression
354 faster, but requires more memory during compression.
356 The minimum _hlog_ is 6 (64 B) and the maximum is 26 (128 MiB).
358 - `chainLog`=_clog_, `clog`=_clog_:
359 Specify the maximum number of bits for a hash chain or a binary tree.
361 Higher numbers of bits increases the chance to find a match which usually
362 improves compression ratio.
363 It also slows down compression speed and increases memory requirements for
365 This option is ignored for the ZSTD_fast strategy.
367 The minimum _clog_ is 6 (64 B) and the maximum is 28 (256 MiB).
369 - `searchLog`=_slog_, `slog`=_slog_:
370 Specify the maximum number of searches in a hash chain or a binary tree
371 using logarithmic scale.
373 More searches increases the chance to find a match which usually increases
374 compression ratio but decreases compression speed.
376 The minimum _slog_ is 1 and the maximum is 26.
378 - `searchLength`=_slen_, `slen`=_slen_:
379 Specify the minimum searched length of a match in a hash table.
381 Larger search lengths usually decrease compression ratio but improve
384 The minimum _slen_ is 3 and the maximum is 7.
386 - `targetLen`=_tlen_, `tlen`=_tlen_:
387 The impact of this field vary depending on selected strategy.
389 For ZSTD\_btopt and ZSTD\_btultra, it specifies the minimum match length
390 that causes match finder to stop searching for better matches.
391 A larger `targetLen` usually improves compression ratio
392 but decreases compression speed.
394 For ZSTD\_fast, it triggers ultra-fast mode when > 0.
395 The value represents the amount of data skipped between match sampling.
396 Impact is reversed : a larger `targetLen` increases compression speed
397 but decreases compression ratio.
399 For all other strategies, this field has no impact.
401 The minimum _tlen_ is 0 and the maximum is 999.
403 - `overlapLog`=_ovlog_, `ovlog`=_ovlog_:
404 Determine `overlapSize`, amount of data reloaded from previous job.
405 This parameter is only available when multithreading is enabled.
406 Reloading more data improves compression ratio, but decreases speed.
408 The minimum _ovlog_ is 0, and the maximum is 9.
409 0 means "no overlap", hence completely independent jobs.
410 9 means "full overlap", meaning up to `windowSize` is reloaded from previous job.
411 Reducing _ovlog_ by 1 reduces the amount of reload by a factor 2.
412 Default _ovlog_ is 6, which means "reload `windowSize / 8`".
413 Exception : the maximum compression level (22) has a default _ovlog_ of 9.
415 - `ldmHashLog`=_ldmhlog_, `ldmhlog`=_ldmhlog_:
416 Specify the maximum size for a hash table used for long distance matching.
418 This option is ignored unless long distance matching is enabled.
420 Bigger hash tables usually improve compression ratio at the expense of more
421 memory during compression and a decrease in compression speed.
423 The minimum _ldmhlog_ is 6 and the maximum is 26 (default: 20).
425 - `ldmSearchLength`=_ldmslen_, `ldmslen`=_ldmslen_:
426 Specify the minimum searched length of a match for long distance matching.
428 This option is ignored unless long distance matching is enabled.
430 Larger/very small values usually decrease compression ratio.
432 The minimum _ldmslen_ is 4 and the maximum is 4096 (default: 64).
434 - `ldmBucketSizeLog`=_ldmblog_, `ldmblog`=_ldmblog_:
435 Specify the size of each bucket for the hash table used for long distance
438 This option is ignored unless long distance matching is enabled.
440 Larger bucket sizes improve collision resolution but decrease compression
443 The minimum _ldmblog_ is 0 and the maximum is 8 (default: 3).
445 - `ldmHashEveryLog`=_ldmhevery_, `ldmhevery`=_ldmhevery_:
446 Specify the frequency of inserting entries into the long distance matching
449 This option is ignored unless long distance matching is enabled.
451 Larger values will improve compression speed. Deviating far from the
452 default value will likely result in a decrease in compression ratio.
454 The default value is `wlog - ldmhlog`.
457 The following parameters sets advanced compression options to something
458 similar to predefined level 19 for files bigger than 256 KB:
460 `--zstd`=wlog=23,clog=23,hlog=22,slog=6,slen=3,tlen=48,strat=6
463 Select the size of each compression job.
464 This parameter is available only when multi-threading is enabled.
465 Default value is `4 * windowSize`, which means it varies depending on compression level.
466 `-B#` makes it possible to select a custom value.
467 Note that job size must respect a minimum value which is enforced transparently.
468 This minimum is either 1 MB, or `overlapSize`, whichever is largest.
472 Report bugs at: https://github.com/facebook/zstd/issues