JuliaArrays / JuliaArrays/AxisArrays.jl

Unexpected allocation on views of AxisArray

Open
#191 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Julia
Stars
207
Forks
41
PR merge metrics
No merged PRs in 30d

Description

Consider the following simple example:

const A = randn(100);
aa = AxisArray(A;x=SortedVector([i for i in 1:100]));
function axisarrays_alloc(array,xs)
           @view array[xs];
           nothing;
 end
xs = 2:10;
@benchmark axisarrays_alloc(aa,xs)

results in:

BenchmarkTools.Trial: 
  memory estimate:  352 bytes
  allocs estimate:  7
  --------------
  minimum time:     393.592 ns (0.00% GC)
  median time:      399.507 ns (0.00% GC)
  mean time:        443.488 ns (4.64% GC)
  maximum time:     15.236 μs (96.24% GC)
  --------------
  samples:          10000
  evals/sample:     201

This occurs if x is a SortedVector or plain Vector but not if it is a range. I believe this is related to #147 since the axes are being copied even in a view. I have not yet verified whether or not it is fixed by #148.

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start with the provided axisarrays_alloc example and benchmark the view of an AxisArray using a Vector or SortedVector versus a range. Read issues #147 and #148 to understand the suspected axis-copying behavior. Done means the view avoids the unexpected allocations for Vector and SortedVector axes, confirmed by the benchmark.

Written by the indexing model from the issue text.

Assessment

Tech stack
julia
Domain
performance
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
38/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.