SciML / SciML/RecursiveArrayTools.jl

Multi-threading `VectorOfArray` of `SVector`s is broken on RecursiveArrayTools.jl version 4

Open
#570 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

bug
Dominant language
Julia
Stars
233
Forks
76
Avg merge
2h 17m
Merged PRs (30d)
9

Description

Describe the bug 🐞

julia> import Pkg; Pkg.activate(temp = true); Pkg.add(["FastBroadcast", "Polyester", "RecursiveArrayTools", "StaticArrays"])
  Activating new project
   Resolving package versions...
    Updating
  [7034ab61] + FastBroadcast v1.3.1
  [f517fe37] + Polyester v0.7.19
  [731186ca] + RecursiveArrayTools v4.0.1
  [90137ffa] + StaticArrays v1.9.18
    Updating
  [7d9f7c33] + Accessors v0.1.44
  [79e6a3ab] + Adapt v4.5.0
  [4fba245c] + ArrayInterface v7.23.0
  [62783981] + BitTwiddlingConvenienceFunctions v0.1.6
  [2a0fbf3d] + CPUSummary v0.2.7
  [fb6a15b2] + CloseOpenIntervals v0.1.13
  [f70d9fcc] + CommonWorldInvalidations v1.0.0
  [34da2185] + Compat v4.18.1
  [a33af91c] + CompositionsBase v0.1.2
  [187b0558] + ConstructionBase v1.6.0
  [adafc99b] + CpuId v0.3.1
  [ffbed154] + DocStringExtensions v0.9.5
  [e2ba6199] + ExprTools v0.1.10
  [7034ab61] + FastBroadcast v1.3.1
  [46192b85] + GPUArraysCore v0.2.0
  [615f187c] + IfElse v0.1.1
  [3587e190] + InverseFunctions v0.1.17
  [10f19ff3] + LayoutPointers v0.1.17
  [1914dd2f] + MacroTools v0.5.16
  [d125e4d3] + ManualMemory v0.1.8
  [f517fe37] + Polyester v0.7.19
  [1d0040c9] + PolyesterWeave v0.2.2
⌅ [aea7be01] + PrecompileTools v1.2.1
  [21216c6a] + Preferences v1.5.2
  [3cdcf5f2] + RecipesBase v1.3.4
  [731186ca] + RecursiveArrayTools v4.0.1
  [ae029012] + Requires v1.3.1
  [7e49a35a] + RuntimeGeneratedFunctions v0.5.17
  [94e857df] + SIMDTypes v0.1.0
  [431bcebd] + SciMLPublic v1.0.1
  [aedffcd0] + Static v1.3.1
  [0d7ed370] + StaticArrayInterface v1.9.0
  [90137ffa] + StaticArrays v1.9.18
  [1e83bf80] + StaticArraysCore v1.4.4
  [7792a7ef] + StrideArraysCore v0.5.8
  [2efcf032] + SymbolicIndexingInterface v0.3.46
  [8290d209] + ThreadingUtilities v0.5.5
  [56f22d72] + Artifacts
  [2a0f44e3] + Base64
  [ade2ca70] + Dates
  [8f399da3] + Libdl
  [37e2e46d] + LinearAlgebra
  [d6f4376e] + Markdown
  [de0858da] + Printf
  [9a3f8284] + Random
  [ea8e919c] + SHA v0.7.0
  [9e88b42a] + Serialization
  [fa267f1f] + TOML v1.0.3
  [cf7118a7] + UUIDs
  [4ec0a83e] + Unicode
  [e66e0078] + CompilerSupportLibraries_jll v1.1.1+0
  [4536629a] + OpenBLAS_jll v0.3.23+5
  [8e850b90] + libblastrampoline_jll v5.11.0+0
        Info Packages marked with ⌅ have new versions available but compatibility constraints restrict them from upgrading. To see why use `status --outdated -m`

julia> using FastBroadcast, Polyester, RecursiveArrayTools, StaticArrays

julia> u = VectorOfArray(fill(SVector(1.0, 1.0), 2, 2))
VectorOfArray{Float64,2}:
2×2 Matrix{SVector{2, Float64}}:
 [1.0, 1.0]  [1.0, 1.0]
 [1.0, 1.0]  [1.0, 1.0]

julia> v = copy(u)
VectorOfArray{Float64,2}:
2×2 Matrix{SVector{2, Float64}}:
 [1.0, 1.0]  [1.0, 1.0]
 [1.0, 1.0]  [1.0, 1.0]

julia> @.. thread=true v = v + u
ERROR: setindex!(::SVector{2, Float64}, value, ::Int) is not defined.
 Hint: Use `MArray` or `SizedArray` to create a mutable static array
Stacktrace:
  [1] error(s::String)
    @ Base ./error.jl:35
  [2] setindex!(a::SVector{2, Float64}, value::Float64, i::Int64)
    @ StaticArrays ~/.julia/packages/StaticArrays/0cEwi/src/indexing.jl:3
  [3] setindex!
    @ ~/.julia/packages/RecursiveArrayTools/0peF1/src/vector_of_array.jl:904 [inlined]
  [4] setindex!
    @ ./subarray.jl:331 [inlined]
  [5] _setindex!
    @ ./abstractarray.jl:1425 [inlined]
  [6] setindex!
    @ ./abstractarray.jl:1395 [inlined]
  [7] __fast_materialize!
    @ ~/.julia/packages/FastBroadcast/WRPAc/src/FastBroadcast.jl:152 [inlined]
  [8] _fast_materialize!
    @ ~/.julia/packages/FastBroadcast/WRPAc/src/FastBroadcast.jl:164 [inlined]
  [9] fast_materialize!
    @ ~/.julia/packages/FastBroadcast/WRPAc/src/FastBroadcast.jl:191 [inlined]
 [10] _batch_broadcast_fn
    @ ~/.julia/packages/FastBroadcast/WRPAc/ext/FastBroadcastPolyesterExt.jl:10 [inlined]
 [11] macro expansion
    @ ~/.julia/packages/Polyester/GQ6c9/src/batch.jl:246 [inlined]
 [12] _batch_no_reserve
    @ ~/.julia/packages/Polyester/GQ6c9/src/batch.jl:168 [inlined]
 [13] batch
    @ ~/.julia/packages/Polyester/GQ6c9/src/batch.jl:335 [inlined]
 [14] batch
    @ ~/.julia/packages/Polyester/GQ6c9/src/batch.jl:286 [inlined]
 [15] fast_materialize_threaded!(dst::VectorOfArray{…}, bc::Base.Broadcast.Broadcasted{…})
    @ FastBroadcastPolyesterExt ~/.julia/packages/FastBroadcast/WRPAc/ext/FastBroadcastPolyesterExt.jl:32
 [16] fast_materialize!(::Threaded, dst::VectorOfArray{…}, bc::Base.Broadcast.Broadcasted{…})
    @ FastBroadcast ~/.julia/packages/FastBroadcast/WRPAc/src/FastBroadcast.jl:216
 [17] top-level scope
    @ REPL[5]:1
Some type information was truncated. Use `show(err)` to see complete types.

julia> Threads.nthreads()
2

Expected behavior

As in version v3.54.0 of RecursiveArrayTools.jl, the output is

julia> @.. thread=true v = v + u
VectorOfArray{Float64,2}:
2×2 Matrix{SVector{2, Float64}}:
 [2.0, 2.0]  [2.0, 2.0]
 [2.0, 2.0]  [2.0, 2.0]

Minimal Reproducible Example 👇

See above

Error & Stacktrace ⚠️

See above

Environment (please complete the following information):

See above

Additional context

Successor of #564. Does RecursiveArrayTools.jl use multiple threads in CI to test the new extensions properly?

Contributor guide

Open the contributing guide

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

Reproduce the example with RecursiveArrayTools v4.0.1 and inspect vector_of_array.jl around line 904, where the stack trace reaches setindex!. Compare the behavior with RecursiveArrayTools v3.54.0, and review the FastBroadcastPolyesterExt and Polyester batch frames. Done means the threaded broadcast over VectorOfArray of SVector values produces the expected doubled values without an error.

Written by the indexing model from the issue text.

Assessment

Tech stack
julia
Domain
data
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Quiet
Clarity
Clearly specified
Newbie friendliness
45/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.