Use KernelAbstractions.jl by kylebeggs · Pull Request #41 · JuliaMeshless/RadialBasisFunctions.jl

kylebeggs · 2025-07-04T12:59:18Z

This PR simply changes the construction of the operator to use KernelAbstractions.jl rather than plain threads, so in the future when we add support for GPU-based neighbor lists we can build the operator on the GPU.

Copilot

Pull Request Overview

This PR replaces the manual threading approach in _build_weights with a batched GPU/CPU kernel via KernelAbstractions.jl, adds a helper for vector-dimension detection, updates autoselect_k to accept any AbstractVector, and brings in KernelAbstractions at the module level.

Swap out Threads.@threads stencil construction for a @kernel-based batch processor.
Introduce _get_vector_dim to compute point dimensionality generically and update autoselect_k.
Add using KernelAbstractions in the main module and clean up old tests.

Reviewed Changes

Copilot reviewed 5 out of 6 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
test/solve.jl	Removed legacy `_calculate_matrix_entry!` tests after pruning the old threaded implementation.
src/utils.jl	Added `_get_vector_dim` methods and widened `autoselect_k` to `AbstractVector`.
src/solve.jl	Replaced threaded loops with a batched `build_stencils_kernel`; defaults to CPU but supports GPU.
src/RadialBasisFunctions.jl	Added `using KernelAbstractions` to enable kernel dispatch.
CLAUDE.md	New guidance file for automated code assistants.

Comments suppressed due to low confidence (4)

test/solve.jl:46

The legacy test set for _calculate_matrix_entry! was removed, leaving no direct coverage for individual matrix-entry logic. Consider adding new tests that validate the outputs of the build_stencils_kernel or the resulting sparse matrix entries to maintain coverage.

end

src/solve.jl:50

The variable nmon is not defined or passed into the kernel; this will error. You should pass nmon as an argument to build_stencils_kernel or declare it as a compile-time constant.

        n = k + nmon

src/solve.jl:51

The type parameter TD is referenced inside the kernel but never passed in or defined within its scope. You need to supply TD (e.g., element type) as an argument or use a constant within the kernel.

        A = Symmetric(zeros(TD, n, n), :U)

src/solve.jl:39

The J array (column indices) is passed into the kernel but never assigned or updated, so it remains at its initial state. Add logic inside the kernel to populate J[(i-1)*k + idx] alongside I and V.

    @kernel function build_stencils_kernel(

src/utils.jl

codecov · 2025-07-04T13:00:54Z

Codecov Report

❌ Patch coverage is 96.55172% with 1 line in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
src/solve.jl	95.65%	1 Missing ⚠️

Files with missing lines	Coverage Δ
src/RadialBasisFunctions.jl	`100.00% <ø> (ø)`
src/utils.jl	`78.57% <100.00%> (+3.57%)`	⬆️
src/solve.jl	`98.87% <95.65%> (+0.11%)`	⬆️

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>

github-actions · 2025-07-04T13:08:38Z

Benchmark Results

	main	`c11a1fe`...	main / `c11a1fe`...
Directional	2.4 ± 0.089 ms	2.43 ± 0.11 ms	0.986 ± 0.057
Directional (per point)	2.33 ± 0.12 ms	2.42 ± 0.084 ms	0.964 ± 0.06
Gradient	7.89 ± 0.32 ms	7.48 ± 0.29 ms	1.05 ± 0.058
MonomialBasis/dim=1/deg=0	0.0475 ± 0.012 μs	0.0473 ± 0.011 μs	1.01 ± 0.34
MonomialBasis/dim=1/deg=1	0.0752 ± 0.013 μs	0.0757 ± 0.014 μs	0.993 ± 0.25
MonomialBasis/dim=1/deg=2	0.0835 ± 0.018 μs	0.0861 ± 0.019 μs	0.97 ± 0.3
MonomialBasis/dim=2/deg=0	0.0364 ± 0.0098 μs	0.0376 ± 0.0091 μs	0.969 ± 0.35
MonomialBasis/dim=2/deg=1	0.0353 ± 0.012 μs	0.0351 ± 0.012 μs	1 ± 0.47
MonomialBasis/dim=2/deg=2	0.0407 ± 0.011 μs	0.0419 ± 0.012 μs	0.972 ± 0.39
MonomialBasis/dim=3/deg=0	0.0344 ± 0.011 μs	0.0367 ± 0.012 μs	0.938 ± 0.43
MonomialBasis/dim=3/deg=1	0.0405 ± 0.012 μs	0.0417 ± 0.012 μs	0.972 ± 0.4
MonomialBasis/dim=3/deg=2	0.0494 ± 0.011 μs	0.0502 ± 0.012 μs	0.984 ± 0.32
Partial	2.35 ± 0.12 ms	2.39 ± 0.11 ms	0.986 ± 0.069
RBF/Gaussian, exp(-(ε*r)²)
├─Shape factor: ε = 1
└─Polynomial augmentation: degree 0/0/∂	10.7 ± 0.1 ns	10.2 ± 0.19 ns	1.05 ± 0.022
RBF/Gaussian, exp(-(ε*r)²)
├─Shape factor: ε = 1
└─Polynomial augmentation: degree 0/0/∂²	10.6 ± 0.1 ns	10.5 ± 0.18 ns	1.01 ± 0.02
RBF/Gaussian, exp(-(ε*r)²)
├─Shape factor: ε = 1
└─Polynomial augmentation: degree 0/0/∇	17.3 ± 0.13 ns	17.3 ± 0.06 ns	1 ± 0.0083
RBF/Gaussian, exp(-(ε*r)²)
├─Shape factor: ε = 1
└─Polynomial augmentation: degree 0/0/∇²	18.6 ± 0.069 ns	18.3 ± 0.099 ns	1.02 ± 0.0067
RBF/Gaussian, exp(-(ε*r)²)
├─Shape factor: ε = 1
└─Polynomial augmentation: degree 1/1/∂	10.7 ± 0.08 ns	10.1 ± 0.09 ns	1.05 ± 0.012
RBF/Gaussian, exp(-(ε*r)²)
├─Shape factor: ε = 1
└─Polynomial augmentation: degree 1/1/∂²	10.6 ± 0.061 ns	10.5 ± 0.18 ns	1.01 ± 0.018
RBF/Gaussian, exp(-(ε*r)²)
├─Shape factor: ε = 1
└─Polynomial augmentation: degree 1/1/∇	17.3 ± 0.14 ns	17.3 ± 0.06 ns	1 ± 0.0088
RBF/Gaussian, exp(-(ε*r)²)
├─Shape factor: ε = 1
└─Polynomial augmentation: degree 1/1/∇²	18.6 ± 0.061 ns	18.3 ± 0.1 ns	1.02 ± 0.0065
RBF/Gaussian, exp(-(ε*r)²)
├─Shape factor: ε = 1
└─Polynomial augmentation: degree 2/2/∂	10.6 ± 0.08 ns	10.1 ± 0.09 ns	1.05 ± 0.012
RBF/Gaussian, exp(-(ε*r)²)
├─Shape factor: ε = 1
└─Polynomial augmentation: degree 2/2/∂²	10.6 ± 0.1 ns	11.1 ± 0.06 ns	0.957 ± 0.01
RBF/Gaussian, exp(-(ε*r)²)
├─Shape factor: ε = 1
└─Polynomial augmentation: degree 2/2/∇	17.3 ± 0.14 ns	17.3 ± 0.06 ns	1 ± 0.0088
RBF/Gaussian, exp(-(ε*r)²)
├─Shape factor: ε = 1
└─Polynomial augmentation: degree 2/2/∇²	18.6 ± 0.061 ns	18.3 ± 0.09 ns	1.02 ± 0.0061
RBF/Inverse Multiquadrics, 1/sqrt((r*ε)²+1)
├─Shape factor: ε = 1
└─Polynomial augmentation: degree 0/0/∂	6.8 ± 0.14 ns	6.32 ± 0.011 ns	1.08 ± 0.022
RBF/Inverse Multiquadrics, 1/sqrt((r*ε)²+1)
├─Shape factor: ε = 1
└─Polynomial augmentation: degree 0/0/∂²	14.2 ± 0.069 ns	14 ± 0.04 ns	1.02 ± 0.0057
RBF/Inverse Multiquadrics, 1/sqrt((r*ε)²+1)
├─Shape factor: ε = 1
└─Polynomial augmentation: degree 0/0/∇	8.59 ± 0.14 ns	8.65 ± 0.031 ns	0.993 ± 0.017
RBF/Inverse Multiquadrics, 1/sqrt((r*ε)²+1)
├─Shape factor: ε = 1
└─Polynomial augmentation: degree 0/0/∇²	16.1 ± 0.1 ns	16.3 ± 0.061 ns	0.986 ± 0.0072
RBF/Inverse Multiquadrics, 1/sqrt((r*ε)²+1)
├─Shape factor: ε = 1
└─Polynomial augmentation: degree 1/1/∂	6.74 ± 0.14 ns	6.33 ± 0.01 ns	1.06 ± 0.022
RBF/Inverse Multiquadrics, 1/sqrt((r*ε)²+1)
├─Shape factor: ε = 1
└─Polynomial augmentation: degree 1/1/∂²	14.1 ± 0.081 ns	13.9 ± 0.071 ns	1.01 ± 0.0078
RBF/Inverse Multiquadrics, 1/sqrt((r*ε)²+1)
├─Shape factor: ε = 1
└─Polynomial augmentation: degree 1/1/∇	8.59 ± 0.14 ns	8.66 ± 0.06 ns	0.992 ± 0.018
RBF/Inverse Multiquadrics, 1/sqrt((r*ε)²+1)
├─Shape factor: ε = 1
└─Polynomial augmentation: degree 1/1/∇²	16.1 ± 0.089 ns	16.3 ± 0.061 ns	0.986 ± 0.0066
RBF/Inverse Multiquadrics, 1/sqrt((r*ε)²+1)
├─Shape factor: ε = 1
└─Polynomial augmentation: degree 2/2/∂	6.74 ± 0.15 ns	6.32 ± 0.011 ns	1.07 ± 0.024
RBF/Inverse Multiquadrics, 1/sqrt((r*ε)²+1)
├─Shape factor: ε = 1
└─Polynomial augmentation: degree 2/2/∂²	14.1 ± 0.08 ns	14 ± 0.041 ns	1 ± 0.0064
RBF/Inverse Multiquadrics, 1/sqrt((r*ε)²+1)
├─Shape factor: ε = 1
└─Polynomial augmentation: degree 2/2/∇	8.63 ± 0.09 ns	8.66 ± 0.059 ns	0.997 ± 0.012
RBF/Inverse Multiquadrics, 1/sqrt((r*ε)²+1)
├─Shape factor: ε = 1
└─Polynomial augmentation: degree 2/2/∇²	16.1 ± 0.09 ns	16.3 ± 0.061 ns	0.986 ± 0.0067
RBF/Polyharmonic spline (r³)
└─Polynomial augmentation: degree 0/0/∂	3.41 ± 0.02 ns	3.73 ± 0.01 ns	0.914 ± 0.0059
RBF/Polyharmonic spline (r³)
└─Polynomial augmentation: degree 0/0/∂²	4.7 ± 0.01 ns	4.7 ± 0.01 ns	1 ± 0.003
RBF/Polyharmonic spline (r³)
└─Polynomial augmentation: degree 0/0/∇	5.44 ± 0.03 ns	5.49 ± 0.05 ns	0.991 ± 0.011
RBF/Polyharmonic spline (r³)
└─Polynomial augmentation: degree 0/0/∇²	3.11 ± 0.001 ns	3.42 ± 0.01 ns	0.909 ± 0.0027
RBF/Polyharmonic spline (r³)
└─Polynomial augmentation: degree 1/1/∂	3.41 ± 0.021 ns	3.73 ± 0.01 ns	0.914 ± 0.0061
RBF/Polyharmonic spline (r³)
└─Polynomial augmentation: degree 1/1/∂²	4.7 ± 0.01 ns	4.7 ± 0.01 ns	1 ± 0.003
RBF/Polyharmonic spline (r³)
└─Polynomial augmentation: degree 1/1/∇	5.44 ± 0.03 ns	5.5 ± 0.05 ns	0.989 ± 0.011
RBF/Polyharmonic spline (r³)
└─Polynomial augmentation: degree 1/1/∇²	3.11 ± 0.001 ns	3.42 ± 0.01 ns	0.909 ± 0.0027
RBF/Polyharmonic spline (r³)
└─Polynomial augmentation: degree 2/2/∂	3.41 ± 0.02 ns	3.73 ± 0.01 ns	0.914 ± 0.0059
RBF/Polyharmonic spline (r³)
└─Polynomial augmentation: degree 2/2/∂²	4.7 ± 0.01 ns	4.7 ± 0.01 ns	1 ± 0.003
RBF/Polyharmonic spline (r³)
└─Polynomial augmentation: degree 2/2/∇	5.44 ± 0.03 ns	5.49 ± 0.041 ns	0.991 ± 0.0092
RBF/Polyharmonic spline (r³)
└─Polynomial augmentation: degree 2/2/∇²	3.11 ± 0.001 ns	3.42 ± 0.01 ns	0.909 ± 0.0027
RBF/Polyharmonic spline (r¹)
└─Polynomial augmentation: degree 0/0/∂	4.28 ± 0.5 ns	4.27 ± 0.01 ns	1 ± 0.12
RBF/Polyharmonic spline (r¹)
└─Polynomial augmentation: degree 0/0/∂²	5.82 ± 0.01 ns	5.78 ± 0.01 ns	1.01 ± 0.0025
RBF/Polyharmonic spline (r¹)
└─Polynomial augmentation: degree 0/0/∇	6.32 ± 0.049 ns	6.25 ± 0.01 ns	1.01 ± 0.008
RBF/Polyharmonic spline (r¹)
└─Polynomial augmentation: degree 0/0/∇²	4.27 ± 0.01 ns	4.5 ± 0.01 ns	0.949 ± 0.0031
RBF/Polyharmonic spline (r¹)
└─Polynomial augmentation: degree 1/1/∂	4.28 ± 0.47 ns	4.27 ± 0.01 ns	1 ± 0.11
RBF/Polyharmonic spline (r¹)
└─Polynomial augmentation: degree 1/1/∂²	5.82 ± 0.01 ns	6.1 ± 0.021 ns	0.954 ± 0.0037
RBF/Polyharmonic spline (r¹)
└─Polynomial augmentation: degree 1/1/∇	6.3 ± 0.04 ns	6.25 ± 0.019 ns	1.01 ± 0.0071
RBF/Polyharmonic spline (r¹)
└─Polynomial augmentation: degree 1/1/∇²	4.27 ± 0.01 ns	4.5 ± 0.01 ns	0.949 ± 0.0031
RBF/Polyharmonic spline (r¹)
└─Polynomial augmentation: degree 2/2/∂	4.28 ± 1 ns	4.27 ± 0.01 ns	1 ± 0.24
RBF/Polyharmonic spline (r¹)
└─Polynomial augmentation: degree 2/2/∂²	5.82 ± 0.01 ns	5.78 ± 0.02 ns	1.01 ± 0.0039
RBF/Polyharmonic spline (r¹)
└─Polynomial augmentation: degree 2/2/∇	6.3 ± 0.05 ns	6.52 ± 0.03 ns	0.966 ± 0.0089
RBF/Polyharmonic spline (r¹)
└─Polynomial augmentation: degree 2/2/∇²	4.27 ± 0.01 ns	4.5 ± 0.01 ns	0.949 ± 0.0031
RBF/Polyharmonic spline (r⁵)
└─Polynomial augmentation: degree 0/0/∂	4.65 ± 0.001 ns	5.26 ± 0.01 ns	0.884 ± 0.0017
RBF/Polyharmonic spline (r⁵)
└─Polynomial augmentation: degree 0/0/∂²	5.26 ± 0.01 ns	4.96 ± 0.001 ns	1.06 ± 0.002
RBF/Polyharmonic spline (r⁵)
└─Polynomial augmentation: degree 0/0/∇	6.05 ± 0.03 ns	6.03 ± 0.06 ns	1 ± 0.011
RBF/Polyharmonic spline (r⁵)
└─Polynomial augmentation: degree 0/0/∇²	3.11 ± 0.009 ns	3.73 ± 0.01 ns	0.833 ± 0.0033
RBF/Polyharmonic spline (r⁵)
└─Polynomial augmentation: degree 1/1/∂	4.65 ± 0.001 ns	5.26 ± 0.01 ns	0.884 ± 0.0017
RBF/Polyharmonic spline (r⁵)
└─Polynomial augmentation: degree 1/1/∂²	5.26 ± 0.01 ns	4.96 ± 0.001 ns	1.06 ± 0.002
RBF/Polyharmonic spline (r⁵)
└─Polynomial augmentation: degree 1/1/∇	6.05 ± 0.03 ns	6.02 ± 0.051 ns	1 ± 0.0099
RBF/Polyharmonic spline (r⁵)
└─Polynomial augmentation: degree 1/1/∇²	3.11 ± 0.01 ns	3.73 ± 0.01 ns	0.833 ± 0.0035
RBF/Polyharmonic spline (r⁵)
└─Polynomial augmentation: degree 2/2/∂	4.65 ± 0.001 ns	5.26 ± 0.01 ns	0.884 ± 0.0017
RBF/Polyharmonic spline (r⁵)
└─Polynomial augmentation: degree 2/2/∂²	5.26 ± 0.01 ns	4.96 ± 0.001 ns	1.06 ± 0.002
RBF/Polyharmonic spline (r⁵)
└─Polynomial augmentation: degree 2/2/∇	6.05 ± 0.03 ns	6.02 ± 0.051 ns	1 ± 0.0099
RBF/Polyharmonic spline (r⁵)
└─Polynomial augmentation: degree 2/2/∇²	3.11 ± 0.01 ns	3.73 ± 0.01 ns	0.833 ± 0.0035
RBF/Polyharmonic spline (r⁷)
└─Polynomial augmentation: degree 0/0/∂	10.3 ± 0.04 ns	10.3 ± 0.02 ns	0.995 ± 0.0043
RBF/Polyharmonic spline (r⁷)
└─Polynomial augmentation: degree 0/0/∂²	4.96 ± 0.01 ns	4.96 ± 0.01 ns	1 ± 0.0029
RBF/Polyharmonic spline (r⁷)
└─Polynomial augmentation: degree 0/0/∇	12.6 ± 0.11 ns	12.5 ± 0.14 ns	1.01 ± 0.014
RBF/Polyharmonic spline (r⁷)
└─Polynomial augmentation: degree 0/0/∇²	8.21 ± 0.08 ns	8.19 ± 0.059 ns	1 ± 0.012
RBF/Polyharmonic spline (r⁷)
└─Polynomial augmentation: degree 1/1/∂	10.3 ± 0.031 ns	10.3 ± 0.02 ns	0.995 ± 0.0036
RBF/Polyharmonic spline (r⁷)
└─Polynomial augmentation: degree 1/1/∂²	4.96 ± 0.01 ns	4.96 ± 0.001 ns	1 ± 0.002
RBF/Polyharmonic spline (r⁷)
└─Polynomial augmentation: degree 1/1/∇	12.6 ± 0.11 ns	12.5 ± 0.15 ns	1.01 ± 0.015
RBF/Polyharmonic spline (r⁷)
└─Polynomial augmentation: degree 1/1/∇²	8.21 ± 0.1 ns	8.19 ± 0.06 ns	1 ± 0.014
RBF/Polyharmonic spline (r⁷)
└─Polynomial augmentation: degree 2/2/∂	10.3 ± 0.03 ns	10.3 ± 0.02 ns	0.995 ± 0.0035
RBF/Polyharmonic spline (r⁷)
└─Polynomial augmentation: degree 2/2/∂²	4.96 ± 0.01 ns	4.96 ± 0.01 ns	1 ± 0.0029
RBF/Polyharmonic spline (r⁷)
└─Polynomial augmentation: degree 2/2/∇	12.6 ± 0.1 ns	13.1 ± 0.15 ns	0.963 ± 0.013
RBF/Polyharmonic spline (r⁷)
└─Polynomial augmentation: degree 2/2/∇²	8.23 ± 0.081 ns	8.19 ± 0.059 ns	1 ± 0.012
time_to_load	0.522 ± 0.0083 s	0.583 ± 0.0028 s	0.895 ± 0.015

Benchmark Plots

A plot of the benchmark results have been uploaded as an artifact to the workflow run for this PR.
Go to "Actions"->"Benchmark a pull request"->[the most recent run]->"Artifacts" (at the bottom).

kylebeggs · 2025-07-14T15:16:55Z

@Davide-Miotti could you give this a quick look, it is not too big of a PR

Davide-Miotti

Nice job! Good ideas both the CLAUDE.md and the integration of KernelAbstractions.jl

kylebeggs added 4 commits April 22, 2025 09:22

update to use KA

311a80e

some cleanup

ea4dd10

bump project version

a4e4381

remove unused code and don;t bump version yet

cfc0475

kylebeggs requested review from Davide-Miotti and Copilot July 4, 2025 12:59

Copilot AI reviewed Jul 4, 2025

View reviewed changes

github-actions bot reviewed Jul 4, 2025

View reviewed changes

src/utils.jl Outdated Show resolved Hide resolved

JuliaFormatter suggestion

c11a1fe

Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>

kylebeggs changed the title ~~Kernel abstractions~~ Use KernelAbstractions.jl Jul 4, 2025

Davide-Miotti approved these changes Jul 27, 2025

View reviewed changes

kylebeggs merged commit 5a44e39 into main Jul 28, 2025
22 of 23 checks passed

kylebeggs deleted the kernel-abstractions branch July 28, 2025 13:48

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Use KernelAbstractions.jl#41

Use KernelAbstractions.jl#41
kylebeggs merged 5 commits intomainfrom
kernel-abstractions

kylebeggs commented Jul 4, 2025

Uh oh!

Copilot AI left a comment

Uh oh!

Uh oh!

codecov bot commented Jul 4, 2025 •

edited

Loading

Uh oh!

github-actions bot commented Jul 4, 2025 •

edited

Loading

Uh oh!

kylebeggs commented Jul 14, 2025

Uh oh!

Davide-Miotti left a comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

kylebeggs commented Jul 4, 2025

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull Request Overview

Reviewed Changes

Uh oh!

Uh oh!

codecov bot commented Jul 4, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

github-actions bot commented Jul 4, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Benchmark Results

Benchmark Plots

Uh oh!

kylebeggs commented Jul 14, 2025

Uh oh!

Davide-Miotti left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

codecov bot commented Jul 4, 2025 •

edited

Loading

github-actions bot commented Jul 4, 2025 •

edited

Loading