https://github.com/murrellgroup/bitpacking.jl

Fast, device-agnostic bitpacking in Julia
https://github.com/murrellgroup/bitpacking.jl

array bitpacking

Last synced: 3 months ago
JSON representation

Fast, device-agnostic bitpacking in Julia

• Host: GitHub
• URL: https://github.com/murrellgroup/bitpacking.jl
• Owner: MurrellGroup
• License: mit
• Created: 2025-09-08T13:32:23.000Z (9 months ago)
• Default Branch: main
• Last Pushed: 2025-09-17T09:29:19.000Z (9 months ago)
• Last Synced: 2025-10-21T11:58:36.940Z (7 months ago)
• Topics: array, bitpacking
• Language: Julia
• Homepage: https://murrellgroup.github.io/BitPacking.jl/
• Size: 243 KB
• Stars: 1
• Watchers: 1
• Forks: 0
• Open Issues: 4
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# BitPacking.jl

[![Stable](https://img.shields.io/badge/docs-stable-blue.svg)](https://MurrellGroup.github.io/BitPacking.jl/stable/)

[![Dev](https://img.shields.io/badge/docs-dev-blue.svg)](https://MurrellGroup.github.io/BitPacking.jl/dev/)

[![Build Status](https://github.com/MurrellGroup/BitPacking.jl/actions/workflows/CI.yml/badge.svg?branch=main)](https://github.com/MurrellGroup/BitPacking.jl/actions/workflows/CI.yml?query=branch%3Amain)

[![Coverage](https://codecov.io/gh/MurrellGroup/BitPacking.jl/branch/main/graph/badge.svg)](https://codecov.io/gh/MurrellGroup/BitPacking.jl)

BitPacking.jl is a Julia package that implements bitpacking and unpacking arrays with element bit-widths from 1 to 8, imposing mild size limitations on the first dimension. Bitpacking methods are designed to be device-agnostic, meaning they run fast on GPUs.

## Usage

```julia

using BitPacking

x = rand(UInt8, 32) .& 0b1111 # 4-bit unpacked values

y = bitpacked(x, 4)           # half the memory size

x == y == bitunpacked(y)      # true

# broadcasting assignment works

y .= rand(UInt8, 32) .& 0b1111 # assign new values

```

## Limitations

- The first dimension of the input array must currently be divisible by the least common multiple of 8 and the bitwidth.

- Broadcasting assignment first materializes the unpacked result.

## Installation

```julia

using Pkg

Pkg.add("BitPacking")

```

## Acknowledgements

BitPacking.jl is partially inspired by [IntArrays.jl](https://github.com/bicycle1885/IntArrays.jl)