Skip to content


Subversion checkout URL

You can clone with
Download ZIP
A compressed alternative to the Java BitSet class


Build Status docs-badge

(c) 2009-2015 Daniel Lemire (, Cliff Moon, David McIntosh (, Robert Becho (, Colby Ranger (, Veronika Zenz (, Owen Kaser (, Gregory Ssi-Yan-Kai (, and Rory Graves (

This code is licensed under Apache License, Version 2.0 (ASL2.0). (GPL 2.0 derivatives are allowed.)

This is a word-aligned compressed variant of the Java Bitset class. We provide both a 64-bit and a 32-bit RLE-like compression scheme. It can be used to implement bitmap indexes.

The goal of word-aligned compression is not to achieve the best compression, but rather to improve query processing time. Hence, we try to save CPU cycles, maybe at the expense of storage. However, the EWAH scheme we implemented is always more efficient storage-wise than an uncompressed bitmap (as implemented in the java BitSet class by Sun).

JavaEWAH offers competitive speed. In an exhaustive comparison, Guzun et al. (ICDE 2014) found that "EWAH offers the best query time for all distributions."

JavaEWAH also supports memory-mapped files: we can serialize the bitmaps to disk and then map them to memory using the java.nio classes. This may avoid wasteful serialization/deserialization routines.

The library also provides a drop-in replacement for the standard BitSet class. Like the other bitmap classes in JavaEWAH, this uncompressed BitSet class supports memory-mapped files as well as many other conviences.

For better performance, use a 64-bit JVM over 64-bit CPUs when using the 64-bit scheme (javaewah.EWAHCompressedBitmap). The 32-bit version (javaewah32.EWAHCompressedBitmap32) should compress better but be comparatively slower. It is recommended however that you run your own benchmark.

Java 6 or better is required. We found the very latest OpenJDK release offered the best performance.

Real-world usage

JavaEWAH is part of Apache Hive and its derivatives (e.g., Apache Spark) and Eclipse JGit. It has been used in production systems for many years. It is part of major Linux distributions.

EWAH is used to accelerate the distributed version control system Git ( You can find the C port of EWAH written by the Git team at

When should you use a bitmap?

Sets are a fundamental abstraction in software. They can be implemented in various ways, as hash sets, as trees, and so forth. In databases and search engines, sets are often an integral part of indexes. For example, we may need to maintain a set of all documents or rows (represented by numerical identifier) that satisfy some property. Besides adding or removing elements from the set, we need fast functions to compute the intersection, the union, the difference between sets, and so on.

To implement a set of integers, a particularly appealing strategy is the bitmap (also called bitset or bit vector). Using n bits, we can represent any set made of the integers from the range [0,n): it suffices to set the ith bit is set to one if integer i is present in the set. Commodity processors use words of W=32 or W=64 bits. By combining many such words, we can support large values of n. Intersections, unions and differences can then be implemented as bitwise AND, OR and ANDNOT operations. More complicated set functions can also be implemented as bitwise operations.

When the bitset approach is applicable, it can be orders of magnitude faster than other possible implementation of a set (e.g., as a hash set) while using several times less memory.

When should you use compressed bitmaps?

An uncompress BitSet can use a lot of memory. For example, if you take a BitSet and set the bit at position 1,000,000 to true and you have just over 100kB. That's over 100kB to store the position of one bit. This is wasteful even if you do not care about memory: suppose that you need to compute the intersection between this BitSet and another one that has a bit at position 1,000,001 to true, then you need to go through all these zeroes, whether you like it or not. That can become very wasteful.

This being said, there are definitively cases where attempting to use compressed bitmaps is wasteful. For example, if you have a small universe size. E.g., your bitmaps represent sets of integers from [0,n) where n is small (e.g., n=64 or n=128). If you are able to uncompressed BitSet and it does not blow up your memory usage, then compressed bitmaps are probably not useful to you. In fact, if you do not need compression, then a BitSet offers remarkable speed. One of the downsides of a compressed bitmap like those provided by JavaEWAH is slower random access: checking whether a bit is set to true in a compressed bitmap takes longer.

How does EWAH compares with the alternatives?

EWAH is part of a larger family of compressed bitmaps that are run-length-encoded bitmaps. They identify long runs of 1s or 0s and they represent them with a marker word. If you have a local mix of 1s and 0, you use an uncompressed word.

There are many formats in this family beside EWAH:

  • Oracle's BBC is an obsolete format at this point: though it may provide good compression, it is likely much slower than more recent alternatives due to excessive branching.
  • WAH is a patented variation on BBC that provides better performance.
  • Concise is a variation on the patented WAH. It some specific instances, it can compress much better than WAH (up to 2x better), but it is generally slower.
  • EWAH is both free of patent, and it is faster than all the above. On the downside, it does not compress quite as well. It is faster because it allows some form of "skipping" over uncompressed words. So though none of these formats are great at random access, EWAH is better than the alternatives.

There are other alternatives however. For example, the Roaring format ( is not a run-length-encoded hybrid. It provides faster random access than even EWAH.

Data format

For more details regarding the compression format, please see Section 3 of the following paper:

Daniel Lemire, Owen Kaser, Kamel Aouiche, Sorting improves word-aligned bitmap indexes. Data & Knowledge Engineering 69 (1), pages 3-28, 2010.

(The PDF file is freely available on the arXiv site.)


For a simple comparison between this library and other libraries such as WAH, ConciseSet, BitSet and other options, please see

However, this is very naive. It is recommended that you run your own benchmarks.

Unit testing

As of October 2011, this packages relies on Maven. To test it:

mvn test

See for details.


        EWAHCompressedBitmap ewahBitmap1 = EWAHCompressedBitmap.bitmapOf(0, 2, 55, 64, 1 << 30);
        EWAHCompressedBitmap ewahBitmap2 = EWAHCompressedBitmap.bitmapOf(1, 3, 64,
                1 << 30);
        System.out.println("bitmap 1: " + ewahBitmap1);
        System.out.println("bitmap 2: " + ewahBitmap2);
        // or
        EWAHCompressedBitmap orbitmap = ewahBitmap1.or(ewahBitmap2);
        System.out.println("bitmap 1 OR bitmap 2: " + orbitmap);
        System.out.println("memory usage: " + orbitmap.sizeInBytes() + " bytes");
        // and
        EWAHCompressedBitmap andbitmap = ewahBitmap1.and(ewahBitmap2);
        System.out.println("bitmap 1 AND bitmap 2: " + andbitmap);
        System.out.println("memory usage: " + andbitmap.sizeInBytes() + " bytes");
        // xor
        EWAHCompressedBitmap xorbitmap = ewahBitmap1.xor(ewahBitmap2);
        System.out.println("bitmap 1 XOR bitmap 2:" + xorbitmap);
        System.out.println("memory usage: " + xorbitmap.sizeInBytes() + " bytes");
        // fast aggregation over many bitmaps
        EWAHCompressedBitmap ewahBitmap3 = EWAHCompressedBitmap.bitmapOf(5, 55,
                1 << 30);
        EWAHCompressedBitmap ewahBitmap4 = EWAHCompressedBitmap.bitmapOf(4, 66,
                1 << 30);
        System.out.println("bitmap 3: " + ewahBitmap3);
        System.out.println("bitmap 4: " + ewahBitmap4);
        andbitmap = EWAHCompressedBitmap.and(ewahBitmap1, ewahBitmap2, ewahBitmap3,
        System.out.println("b1 AND b2 AND b3 AND b4: " + andbitmap);
        // serialization
        ByteArrayOutputStream bos = new ByteArrayOutputStream();
        // Note: you could use a file output steam instead of ByteArrayOutputStream
        ewahBitmap1.serialize(new DataOutputStream(bos));
        EWAHCompressedBitmap ewahBitmap1new = new EWAHCompressedBitmap();
        byte[] bout = bos.toByteArray();
        ewahBitmap1new.deserialize(new DataInputStream(new ByteArrayInputStream(bout)));
        System.out.println("bitmap 1 (recovered) : " + ewahBitmap1new);
        if (!ewahBitmap1.equals(ewahBitmap1new)) throw new RuntimeException("Will not happen");
        // we can use a ByteBuffer as backend for a bitmap
        // which allows memory-mapped bitmaps
        ByteBuffer bb = ByteBuffer.wrap(bout);
        EWAHCompressedBitmap rmap = new EWAHCompressedBitmap(bb);
        System.out.println("bitmap 1 (mapped) : " + rmap);

        if (!rmap.equals(ewahBitmap1)) throw new RuntimeException("Will not happen");
        // support for threshold function (new as of version 0.8.0):
        // mark as true a bit that occurs at least T times in the source
        // bitmaps
        EWAHCompressedBitmap threshold2 = EWAHCompressedBitmap.threshold(2,
                ewahBitmap1, ewahBitmap2, ewahBitmap3, ewahBitmap4);
        System.out.println("threshold 2 : " + threshold2);


You can use our drop-in replacement for the BitSet class in a memory-mapped file context as follows:

        final FileOutputStream fos = new FileOutputStream(tmpfile);
        BitSet Bitmap = BitSet.bitmapOf(0, 2, 55, 64, 512);
        Bitmap.serialize(new DataOutputStream(fos));
        RandomAccessFile memoryMappedFile = new RandomAccessFile(tmpfile, "r");
        ByteBuffer bb = memoryMappedFile.getChannel().map(
                FileChannel.MapMode.READ_ONLY, 0, totalcount);
        ImmutableBitSet mapped = new ImmutableBitSet(bb.asLongBuffer());

There are more examples in the "examples" folder (e.g., for memory-file mapping).

Maven central repository

You can download JavaEWAH from the Maven central repository:

You can also specify the dependency in the Maven "pom.xml" file:


Naturally, you should replace "version" by the version you desire.

Travis (Continuous integration)

You can check whether the latest version builds on your favorite version of Java using Travis:


Joel Boehland wrote Clojure wrappers:

Frequent questions

Question: How do I build javaewah without testing or signing?

    mvn clean install -DskipTests -Dgpg.skip=true

Question: Will JavaEWAH support long values?

Answer: It might, but it does not at the moment.

Question: How do I check the value of a bit?

Answer: If you need to routinely check the value of a given bit quickly, then EWAH might not be the right format. However, if you must do it, you can proceed as follows:

           * Suppose you have the following bitmap:
          EWAHCompressedBitmap b = EWAHCompressedBitmap.bitmapOf(0,2,64,1<<30);
           * We want to know if bit 64 is set:
          boolean is64set = b.get(64);

API Documentation

Further reading

Daniel Lemire, Owen Kaser, Kamel Aouiche, Sorting improves word-aligned bitmap indexes, Data & Knowledge Engineering 69 (1), 2010.

Owen Kaser and Daniel Lemire, Compressed bitmap indexes: beyond unions and intersections, Software: Practice and Experience, 2014.

Help needed

JavaEWAH has been used in production for many years. However, we still need help writing more tests.

Coverage Status


Special thanks to Shen Liang for optimization advice.

This work was supported by NSERC grant number 26143.

Something went wrong with that request. Please try again.