# \[ANN\] DispatchedTuples.jl: A Dict/NamedTuple-like gpu-friendly data structure

**URL:** https://discourse.julialang.org/t/ann-dispatchedtuples-jl-a-dict-namedtuple-like-gpu-friendly-data-structure/57415
**Category:** Package Announcements
**Tags:** package, announcement, gpu, tuple, namedtuple
**Created:** [March 17, 2021, 9:11pm UTC](https://discourse.julialang.org/t/ann-dispatchedtuples-jl-a-dict-namedtuple-like-gpu-friendly-data-structure/57415 "2021-03-17T21:11:20Z")
**Posts on this page:** 1
**Page:** 1

<div class="post-metadata">

### Author: ![charleskawczynski](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/charleskawczynski/32/13783_2.png) [@charleskawczynski](https://discourse.julialang.org/u/charleskawczynski)
#### Post date: [March 17, 2021, 9:11pm UTC](https://discourse.julialang.org/t/ann-dispatchedtuples-jl-a-dict-namedtuple-like-gpu-friendly-data-structure/57415/1 "2021-03-17T21:11:20Z")

</div>

[DispatchedTuples.jl](https://github.com/charleskawczynski/DispatchedTuples.jl) is a _very_ lightweight package that defines an efficient Dict/NamedTuple-like data structure that is gpu-friendly (as they are tuple-backed).

The main difference between a NamedTuple and a DispatchedTuple is that the keys are **instances of types**. The reason for this choice was mostly because Julia’s type system and dispatch are radically awesome.

This package is in _very_ early development, and I would appreciate feedback on the interface and how extending it might be useful for the overall Julia ecosystem.