Skip to content
Long Range Zip
C C++ Perl Shell Assembly
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.


lrzip README

Long Range ZIP or Lzma RZIP

This is a compression program optimised for large files. The larger the file
and the more memory you have, the better the compression advantage this will
provide, especially once the files are larger than 100MB. The advantage can
be chosen to be either size (much smaller than bzip2) or speed (much faster
than bzip2).

Quick lowdown of the most used options:

 lrztar directory
This will produce an archive directory.tar.lrz compressed with lzma

 lrzuntar directory.tar.lrz
This will completely extract an archived directory

 lrzip filename
This will produce an archive filename.lrz compressed with lzma (best all
round) giving slow compression and fast decompression

 lrzip -z filename
This will produce an archive filename.lrz compressed with ZPAQ giving extreme
compression but which takes ages to compress and decompress

 lrzip -l filename
This will produce an archive filename.lrz compressed with LZO giving very
fast compression and fast decompression

 lrunzip filename.lrz
This will decompress filename.lrz into filename

Lrzip uses an extended version of rzip which does a first pass long distance
redundancy reduction. The lrzip modifications make it scale according to
memory size.
The data is then either:
1. Compressed by lzma (default) which gives excellent compression
at approximately twice the speed of bzip2 compression
2. Compressed by a number of other compressors chosen for different reasons,
in order of likelihood of usefulness:
2a. ZPAQ: Extreme compression up to 20% smaller than lzma but ultra slow
at compression AND decompression.
2b. LZO: Extremely fast compression and decompression which on most machines
compresses faster than disk writing making it as fast (or even faster) than
simply copying a large file
2c. GZIP: Almost as fast as LZO but with better compression.
2d. BZIP2: A defacto linux standard of sorts but is the middle ground between
lzma and gzip and neither here nor there.
3. Leaving it uncompressed and rzip prepared. This form improves substantially
any compression performed on the resulting file in both size and speed (due to
the nature of rzip preparation merging similar compressible blocks of data and
creating a smaller file). By "improving" I mean it will either speed up the
very slow compressors with minor detriment to compression, or greatly increase
the compression of simple compression algorithms.

The major disadvantages are:
1. The main lrzip application only works on single files so it requires the
lrztar wrapper to fake a complete archiver.
2. It requires a lot of memory to get the best performance out of, and is not
really usable (for compression) with less than 256MB. Decompression requires
less ram and works on smaller ram machines.
3. Only stdin in compression works well. The other combinations of
stdin/stdout work but in a very inefficient manner generating temporary files
on disk so this method of using lrzip is not recommended.

The unique feature of lrzip is that it tries to make the most of the available
ram in your system at all times for maximum benefit. It does this by default,
choosing the largest sized window possible without running out of memory. It
also has a unique "sliding mmap" feature which makes it possible to even use
a compression window larger than your ramsize, if the file is that large. It
does this (with the -U option) by implementing one large mmap buffer as per
normal, and a smaller moving buffer to track which part of the file is
currently being examined, emulating a much larger single mmapped buffer.
Unfortunately this mode is many times slower.

See the file README.benchmarks in doc/ for performance examples and what kind
of data lrzip is very good with.

(nasm on 32bit x86)

To build/install:
make install


Q. How do I make a static build?
A. make static

Q. I want the absolute maximum compression I can possibly get, what do I do?
A. Try the command line options -MUz. This will use all available ram and ZPAQ
compression, and even use a compression window larger than you have ram.
Expect serious swapping to occur if your file is larger than your ram and for
it to take many times longer.

Q. I want the absolute fastest decent compression I can possibly get.
A. Try the command line options -Ml. This will use the maximum possible
memory, lzo backend compression, and level 7 compression (1 isn't much faster).

Q. How much slower is the unlimited mode?
A. It depends on 2 things. First, just how much larger than your ram the file
is, as the bigger the difference, the slower it will be. The second is how much
redundant data there is. The more there is, the slower, but ultimately the
better the compression. Using the example of a 10GB virtual image on a machine
with 8GB ram, it would allocate about 5.5GB by default, yet is capable of
allocating all the ram for the 10GB file in -M mode.

Options		Size		Compress	Decompress
-l		1793312108	05m13s		3m12s
-lM		1413268368	04m18s		2m54s
-lU		1413268368	06m05s		2m54s

As you can see, the -U option gives the same compression in this case as the
-M option, and for about 50% more time. The advantage to using -U is that it
will work even when the size can't be encompassed by -M, but progressively
slower. Why isn't it on by default? If the compression window is a LOT larger
than ram, with a lot of redundant information it can be drastically slower. I
may revisit this possibility in the future if I can make it any faster.

Q. Can I use your tool for even more compression than lzma offers?
A. Yes, the rzip preparation of files makes them more compressible by every
other compression technique I have tried. Using the -n option will generate
a .lrz file smaller than the original which should be more compressible, and
since it is smaller it will compress faster than it otherwise would have.

Q. 32bit?
A. 32bit machines have a limit of 2GB sized compression windows due to
userspace limitations on mmap and malloc, so even if you have much more ram
you will not be able to use compression windows larger than 2GB. Also you
may be unable to decompress files compressed on 64bit machines which have
used windows larger than 2GB.

Q. How about 64bit?
A. 64bit machines with their ability to address massive amounts of ram will
excel with lrzip due to being able to use compression windows limited only in
size by the amount of physical ram.

Q. Other operating systems?
A. The code is POSIXy with GNU extensions. Patches are welcome. Version 0.43+
should build on MacOSX 10.5+

Q. Does it work on stdin/stdout?
A. Yes it does. Compression from stdin works nicely.. However the other
combinations of stdin and stdout use temporary files on disk because of
seeking requirements so the performance of these mode is low. Not recommended!

Q. I have another compression format that is even better than zpaq, can you
use that?
A. You can use it yourself on rzip prepared files (see above). Alternatively
if the source code is compatible with the GPL license it can be added to the
lrzip source code. Libraries with functions similar to compress() and
decompress() functions of zlib would make the process most painless. Please
tell me if you have such a library so I can include it :)

Q. What's this "Starting lzma back end compression thread..." message?
A. While I'm a big fan of progress percentage being visible, unfortunately
lzma compression can't currently be tracked when handing over 100+MB chunks
over to the lzma library. Therefore you'll see progress percentage until
each chunk is handed over to the lzma library.

Q. What's this "lzo testing for incompressible data" message?
A. Other compression is much slower, and lzo is the fastest. To help speed up
the process, lzo compression is performed on the data first to test that the
data is at all compressible. If a small block of data is not compressible, it
tests progressively larger blocks until it has tested all the data (if it fails
to compress at all). If no compressible data is found, then the subsequent
compression is not even attempted. This can save a lot of time during the
compression phase when there is incompressible data. Theoretically it may be
possible that data is compressible by the other backend (zpaq, lzma etc) and not
at all by lzo, but in practice such data achieves only minuscule amounts of
compression which are not worth pursuing. Most of the time it is clear one way
or the other that data is compressible or not. If you wish to disable this
test and force it to try compressing it anyway, use -T 0.

Q. I Have truckloads of ram so I can compress files much better, but can my
generated file be decompressed on machines with less ram?
A. Yes. Ram requirements for decompression go up only by the -L compression
option with lzma and are never anywhere near as large as the compression
requirements. However if you're on 64bit and you use a compression window
greater than 2GB, it might not be possible to decompress it on 32bit machines.

Q. Why are you including bzip2 compression?
A. To maintain a similar compression format to the original rzip (although the
other modes are more useful).

Q. What about multimedia?
A. Most multimedia is already in a heavily compressed "lossy" format which by
its very nature has very little redundancy. This means that there is not
much that can actually be compressed. If your video/audio/picture is in a
high bitrate, there will be more redundancy than a low bitrate one making it
more suitable to compression. None of the compression techniques in lrzip are
optimised for this sort of data. However, the nature of rzip preparation
means that you'll still get better compression than most normal compression
algorithms give you if you have very large files. ISO images of dvds for
example are best compressed directly instead of individual .VOB files. ZPAQ is
the only compression format that can do any significant compression of

Q. Is this multithreaded?
A. As of version 0.530, it is HEAVILY multithreaded with the back end
compression phase, and will continue to process the rzip pre-processing phase
so when using one of the more CPU intensive backend compressions like lzma or
zpaq, SMP machines will show massive speed improvements. Lrzip will detect the
number of CPUs to use, but it can be overridden with the -p option if the
slightly better compression is desired more than speed. Decompression at the
moment is not multithreaded, but is already much faster than the compression
phase (except for zpaq), but enhancements for decompression are planned.

Q. This uses heaps of memory, can I make it use less?
A. Well you can by setting -w to the lowest value (1) but the huge use of
memory is what makes the compression better than ordinary compression
programs so it defeats the point. You'll still derive benefit with -w 1 but
not as much.

Q. What CFLAGS should I use?
A. With a recent enough compiler (gcc>4) setting both CFLAGS and CXXFLAGS to
	-O3 -march=native -fomit-frame-pointer

Q. What compiler does this work with?
A. It was been tested on gcc, ekopath and the intel compiler successfully.
Whether the commercial compilers help or not, I could not tell you.

Q. What codebase are you basing this on?
A. rzip v2.1 and lzma sdk907, but it should be possible to stay in sync with
each of these in the future.

Q. Do we really need yet another compression format?
A. It's not really a new one at all; simply a reimplementation of a few very
good performing ones that will scale with memory and file size.

Q. How do you use lrzip yourself?
A. Two basic uses. I compress large files currently on my drive with the
-l option since it is so quick to get a space saving, and when archiving
data for permanent storage I compress it with the default options.

Q. I found a file that compressed better with plain lzma. How can that be?
A. When the file is more than 5 times the size of the compression window
you have available, the efficiency of rzip preparation drops off as a means
of getting better compression. Eventually when the file is large enough,
plain lzma compression will get better ratios. The lrzip compression will be
a lot faster though. The only way around this is to use as much ram as
possible with the -M option, and going beyond that with the -U option.

Q. Can I use swapspace as ram for lrzip with a massive window?
A. It will indirectly do this with -M mode enabled. If you want the windows
even larger, -U (unlimited) mode will make the compression window as big as
the file itself no matter how big it is, but it will slow down 100 times
during the compression phase once it has reached your full ram.

Q. Why do you nice it to +19 by default? Can I speed up the compression by
changing the nice value?
A. This is a common misconception about what nice values do. They only tell the
cpu process scheduler how to prioritise workloads, and if your application is
the _only_ thing running it will be no faster at nice -20 nor will it be any
slower at +19.

Q. What is the Threshold option, -T ## (1-10)?
A. It is for adjusting the sensitivity of the LZO test that is used when LZMA
compression is selected. When highly random or already-compressed data chunks
are evaluated for LZMA compression, sometimes LZO compression actually will
create a larger chunk than the original.

The Threshold is used to determine a minimum compression amount relative to
the size of the data being evaluated. A value of 1 is the default. This
means that the compression threshold amount is >0% of the size of the
original data. If the threshold is not achieved, the LZMA compression will not
be done and the chunk will not be compressed. Values can be from 0 (bypass the
test) to 10 (maximum compression efficiency expected). The following table can
be used.

For LZO compressor test
T value		Compression %	Compression Ratio
  0		    Ignored
  1		    0-5%	     1.00-1.05	very low compression expected
  2		    5-10%	     1.05-1.10	default value
  3		    10-20%	     1.12-1.25
  4		    20-30%           1.25-1.43
  5		    30-40%	     1.43-1.66
  6		    40-50%           1.66-2.00
  7		    50-60%           2.00-2.50
  8                 60-70%	     2.50-3.33
  9		    70-80%           3.33-5.00
  10                 80+%           5x+

Whenever the data chunk does not compress to the Threshold value, no LZMA
compression will be attempted. For example, if you select -T 5, LZMA
compression will be performed if the projected compression ratio is
less than 1.43. Otherwise, data will be written in rzip format. Setting
a very high T value will result in a lot of uncompressed data in the lrzip
file. However, a lot of time will be saved. For most people you shouldn't ever
need to touch this.

Q. Compression and decompression progress on large archives slows down and
speeds up. There's also a jump in the percentage at the end?
A. Yes, that's the nature of the compression/decompression mechanism. The jump
is because the rzip preparation makes the amount of data much smaller than the
compression backend (lzma) needs to compress.

Q. The percentage counter doesn't always get to 100%.
A. It's quite hard to predict during the rzip phase how long it will take as
lots of redundant data will not count towards the percentage.

Q. Tell me about patented compression algorithms, GPL, lawyers and copyright.
A. No

Q. I receive an error "LZMA ERROR: 2. Try a smaller compression window."
   what does this mean?
A. LZMA requests large amounts of memory. When a higher compression window is
   used, there may not be enough contiguous memory for LZMA. LZMA may request
   up to 25% of TOTAL ram depending on compression level. If contiguous blocks
   of memory are not free, LZMA will return an error. This is not a fatal
   error. However, the current Stream will not be compressed.

Q. Where can I get more information about the internals of LZMA?
A. See and Also, see the file
   ./lzma/C/lzmalib.h which explains the LZMA properties used and the LZMA
   memory requirements and computation.

Due to mmap limitations the maximum size a window can be set to is currently
2GB on 32bit unless the -U option is specified. Files generated on 64 bit
machines with windows >2GB in size might not be decompressible on 32bit

Probably lots. Tell me if you spot any :)


Thanks to Andrew Tridgell for rzip. Thanks to Markus Oberhumer for lzo.
Thanks to Igor Pavlov for lzma. Thanks to Jean-loup Gailly and Mark Adler
for the zlib compression library. Thanks to Christian Leber for lzma
compat layer, Michael J Cohen for Darwin support, Lasse Collin for fix
to LZMALib.cpp and for suggestions, and everyone else who coded
along the way. Huge thanks to Peter Hyman for most of the 0.19-0.24 changes,
and the update to the multithreaded lzma library and all sorts of other
features. Thanks to René Rhéaume for fixing executable stacks and
Ed Avis for various fixes. Thanks to Matt Mahoney for zpaq code. Thanks to
Jukka Laurila for Darwin support. Thanks to George Makrydakis for lrztar.

Con Kolivas <>
Mon, 7 Nov 2010

Also documented by
Peter Hyman <>
Sun, 04 Jan 2009
Something went wrong with that request. Please try again.