by Daniel Lemire, Leonid Boytsov, Owen Kaser, Maxime Caron, Louis Dionne, Michel Lemay, Erik Kruus, Andrea Bedini, Matthias Petri
A research library with integer compression schemes. It is broadly applicable to the compression of arrays of 32-bit integers where most integers are small. The library seeks to exploit SIMD instructions (SSE) whenever possible.
This library can decode at least 4 billions of compressed integers per second on most desktop or laptop processors. That is, it can decompress data at a rate of 15 GB/s. This is significantly faster than generic codecs like gzip, LZO, Snappy or LZ4.
It is used by the zsearch engine (http://victorparmar.github.com/zsearch/) as well as in GMAP and GSNAP (http://research-pub.gene.com/gmap/). It has been ported to Java (https://github.com/lemire/JavaFastPFOR) and Go (https://github.com/reducedb/encoding). The Java port is used by ClueWeb Tools (https://github.com/lintool/clueweb).
Apache Lucene version 4.6.x uses a compression format derived from our FastPFOR scheme (see http://lucene.apache.org/core/4_6_1/core/org/apache/lucene/util/PForDeltaDocIdSet.html).
Myth: SIMD compression requires very large blocks of integers (1024 or more).
Fact: This is not true. Our fastest scheme (SIMDBinaryPacking) works over blocks of 128 integers.
Myth: SIMD compression means high speed but less compression.
Fact: This is wrong. Some schemes cannot easily be accelerated with SIMD instructions, but many that do compress very well.
If you are working primarily with sorted lists of integers, then you might want to use differential coding. That is you may want to compress the deltas instead of the integers themselves. The current library (fastpfor) is generic and was not optimized for this purpose. However, we have another library designed to compress sorted integer lists:
https://github.com/lemire/SIMDCompressionAndIntersection
This other library (SIMDCompressionAndIntersection) also comes complete with new SIMD-based intersection algorithms.
FastPFOR is a C++ research library. For something simpler, written in C, see:
https://github.com/lemire/simdcomp
For a simple example, please see
example.cpp
in the root directory of this project.
Please see:
- Daniel Lemire and Leonid Boytsov, Decoding billions of integers per second through vectorization, Software Practice & Experience (to appear) http://arxiv.org/abs/1209.2137 http://onlinelibrary.wiley.com/doi/10.1002/spe.2203/abstract
- Daniel Lemire, Leonid Boytsov, Nathan Kurz, SIMD Compression and the Intersection of Sorted Integers, arXiv: 1401.6399, 2014 http://arxiv.org/abs/1401.6399
This library was used by the following papers:
- G. Ottaviano, R. Venturini, Partitioned Elias-Fano Indexes, ACM SIGIR 2014 http://www.di.unipi.it/~ottavian/files/elias_fano_sigir14.pdf
- M. Petri, A. Moffat, J. S. Culpepper, Score-Safe Term Dependency Processing With Hybrid Indexes, ACM SIGIR 2014 http://www.culpepper.io/publications/sp074-petri.pdf
This code is licensed under Apache License, Version 2.0 (ASL2.0).
This code requires a compiler supporting C++11. This was a design decision.
It builds under
- clang++ 3.2 (LLVM 3.2) or better,
- Intel icpc (ICC) 13.0.1 or better,
- MinGW32 (x64-4.8.1-posix-seh-rev5)
- Microsoft VS 2012 or better,
- and GNU GCC 4.7 or better.
The code was tested under Windows, Linux and MacOS.
To fully use the library, your processor should support SSSE3. This includes almost every Intel or AMD processor sold after 2006. (Note: the key schemes require merely SSE2.)
Some specific binaries will only run if your processor supports SSE4.1. They have been purely used for specific tests however.
At the root of the project:
-
Create a directory to contain the out-of-source build.
mkdir build cd build -
Generate the files to actually perform the build. Many build systems are supported by CMake, so you may want to look at the documentation.
cmake .. -
Given you generated Unix Makefiles (the default), you can now build the library.
make
You can specify which C++ compiler you are using with the CXX variable.
e.g., under bash type
export CXX=g++-4.7
make
You should not assume that our objects are thread safe. If you have several threads, each thread should have its own IntegerCODEC objects to ensure that there is no concurrency problems.
We support clang, Visual Studio and the Intel compiler, but a common default is GCC 4.7 or better.
Under a recent version of Ubuntu (12.10), you can install GCC 4.7 by typing:
sudo apt-get install gcc-4.7 g++-4.7
Mac Ports supports GCC 4.7. You can install it by typing:
sudo port install gcc47
With minor changes, all schemes will compile fine under compilers that do not support C++11. And porting the code to C should not be a challenge.
In any case, we already support 3 major C++ compilers so portability is not a major issue.
Many schemes cannot be efficiently ported to Java. However some have been. Please see:
https://github.com/lemire/JavaFastPFOR
If you used CMake to generate the build files, the check target will
run the unit tests. For example , if you generated Unix Makefiles
make check
will do it. If you used raw make, type the following:
make
./unit
Note that we are thorough in the unit tests so it can take several minutes to run through them all.
make codecs
./codecs --clusterdynamic
./codecs --uniformdynamic
Typing "make allallall" will install some testing binaries that depend on Google Snappy. If you want to build these, you need to install Google snappy. You can do so on a recent ubuntu machine as:
sudo apt-get install libsnappy-dev
Typing "make" will generate an "inmemorybenchmark" executable that can process data files.
You can use it to process arrays on (sorted!) integers on disk using the following 32-bit format: 1 unsigned 32-bit integer indicating array length followed by the corresponding number of 32-bit integer. Repeat.
( It is assumed that the integers are sorted.)
Once you have such a binary file somefilename you can process it with our inmemorybenchmark:
./inmemorybenchmark --minlength 10000 somefilename
The "minlength" flag skips short arrays. (Warning: timings over short arrays are unreliable.)
As of April 2014, we recommend getting our archive at
http://lemire.me/data/integercompression2014.html
It is the data was used for the following paper:
Daniel Lemire, Leonid Boytsov, Nathan Kurz, SIMD Compression and the Intersection of Sorted Integers, arXiv: 1401.6399, 2014 http://arxiv.org/abs/1401.6399
(Please consider grabbing our new archive at http://lemire.me/data/integercompression2014.html instead.)
Grab the data set from:
http://boytsov.info/datasets/clueweb09gap/
Using the provided software, uncompress it and run the "toflat" executable to create a "clueweb09.bin" file then run:
./inmemorybenchmark --minlength 10000 clueweb09.bin
Note: processing can take over an hour.
(Please consider grabbing our new archive at http://lemire.me/data/integercompression2014.html instead.)
You can test the library over d-gaps data from the TREC GOV2 data set that was made graciously available by Fabrizio Silvestri, Takeshi Yamamuro and Rossano Venturini.
Go to:
http://integerencoding.isti.cnr.it/?page_id=8
Download both the software and the gov.sort.tar file. You might find useful to grow their stable-0.2.0 version from https://github.com/maropu/integer_encoding_library/releases/tag/stable-0.2.0
Untar the tar file:
tar xvf gov2.sort.tar
You may want to make the newly generated files non-writeable (I'm paranoid):
chmod -w gov2.sort.Delta gov2.sort.Delta.TOC
Build the software (you need the decoders executable) and
You need to run this command
./decoders 3 gov2.sort somefilename
where "3" is for delta.
Then you should be able to test with out inmemorybenchmark:
./inmemorybenchmark --minlength 10000 somefilename.DEC
Note: processing can take over an hour.
Our code is thoroughly tested.
One common issue is that people do not provide large enough buffers. Some schemes can have such small compression rates that the compressed data generated will be much larger than the input data.
Also, make sure that all provided buffers are 16-byte aligned or else, some SSE instructions may fail.
I (D. Lemire) did not patent anything.
However, we implemented varint-G8UI which was patented by its authors. DO NOT use varint-G8UI if you want to avoid patents.
The rest of the library should be patent-free.
This work was supported by NSERC grant number 26143.