IntArrays

IntArrays.jl is a package for packed integer arrays. An array type, IntArray, is exported from this package and some methods in Base are extended for it.

The IntArray type is defined as follows:

type IntArray{w,T<:Unsigned,n} <: AbstractArray{T,n}

where

• w: the bit width of integers (i.e. the number of bits to encode an integer)
• T: the type of integers
• n: the number of dimensions in the array.

This works like normal arrays, but each element is packed in a buffer as compact as possible. That means the total memory footprint can be reduced if you specify small w value: the total size is about w * length(int_array) bits. You can think of it as a generalization of BitArray defined in the standard library: IntArray can store any (unsigned) integers, whereas BitArray is restricted to Bool. It is your responsibility to keep values between 0 and 2^w-1; otherwise values will be truncated to w bits with no warning.

Like Vector{T} and Matrix{T} in Base, IntVector{w,T} and IntMatrix{w,T} are also defined as a type alias of IntArray{w,T,n}.

Example

julia> using IntArrays

julia> ivec = IntVector{2}([0x00, 0x01, 0x03, 0x02])
4-element IntArrays.IntArray{2,UInt8,1}:
 0x00
 0x01
 0x03
 0x02

julia> ivec[2]
0x01

julia> ivec[2] = 0x03
0x03

julia> ivec[2]
0x03

See tutorial.ipynb for more details.

Benchmark

Micro benchmarks can be found in the benchmark directory.

The getindex and setindex! methods of IntArray are 2-7 times slower than those of raw arrays due to the heavy bit operations. Any pull request that fills the gap is welcome!

UInt8

UInt64

julia> versioninfo()
Julia Version 0.4.0-dev+6866
Commit e0b4510* (2015-08-20 22:06 UTC)
DEBUG build
Platform Info:
  System: Darwin (x86_64-apple-darwin14.4.0)
  CPU: Intel(R) Core(TM) i5-4288U CPU @ 2.60GHz
  WORD_SIZE: 64
  BLAS: libopenblas (USE64BITINT DYNAMIC_ARCH NO_AFFINITY Haswell)
  LAPACK: libopenblas
  LIBM: libopenlibm
  LLVM: libLLVM-3.3

Plan

• Behave more like normal arrays
• Store signed integers