Skip to content
Merged
Show file tree
Hide file tree
Changes from 24 commits
Commits
Show all changes
28 commits
Select commit Hold shift + click to select a range
06645e1
Update dependencies and migrate Trixi and CUDA APIs
huiyuxie Oct 15, 2026
06ddcfc
Use descriptive dimension names in compatibility tests
huiyuxie Oct 15, 2026
cac47bf
Apply JuliaFormatter 2 formatting
huiyuxie Oct 15, 2026
8789a2a
Fix GPU array dispatch and CUDA 6 device attributes
huiyuxie Oct 16, 2026
51e5dc9
Support multidimensional GPU states without dispatch ambiguity
huiyuxie Oct 16, 2026
d9d8258
Resolve the local package in documentation CI
huiyuxie Oct 16, 2026
3b2965f
Use the current finite volume flux field in shock capturing
huiyuxie Oct 16, 2026
35e34c9
Restore Aqua quality tests and update dependency compatibility
huiyuxie Oct 20, 2026
6cfce50
Resolve GPU callback ambiguities and run Aqua in dependency CI
huiyuxie Oct 20, 2026
3f6ad8e
Consolidate CI tests and documentation checks in Buildkite
huiyuxie Oct 20, 2026
48a4be3
Test TrixiCUDA.jl on Julia 1.10 through 1.13
huiyuxie Oct 20, 2026
128fd50
Preserve resolved dependencies in persistent-task checks
huiyuxie Oct 21, 2026
5f613e1
Defer Aqua integration to a separate pull request
huiyuxie Oct 21, 2026
c5be89d
Remove dependency compatibility check and standardize TreeMesh format…
huiyuxie Oct 21, 2026
758b17c
Restore comma separators in TreeMesh constructors
huiyuxie Oct 21, 2026
f9a6b91
Apply batched suggestions from code review
huiyuxie Oct 23, 2026
2482f7f
Apply Julia formatting
huiyuxie Oct 23, 2026
37b369d
Remove unused imports and group module dependencies
huiyuxie Oct 23, 2026
c3d0764
Consolidate imports within each category
huiyuxie Oct 23, 2026
cabf2bd
Fix GPU array dispatch ambiguity
huiyuxie Oct 23, 2026
30ab723
Align GPU DGSEM method signatures with upstream
huiyuxie Oct 23, 2026
f956b89
Restrict GPU solver methods to DGSEM
huiyuxie Oct 23, 2026
c255899
Use explicit DGSEM type construction
huiyuxie Oct 23, 2026
d192787
Apply Julia formatting
huiyuxie Oct 23, 2026
760553c
Apply Julia formatting
huiyuxie Oct 23, 2026
5a44e6d
Use inverse quadrature weights directly in GPU surface integrals
huiyuxie Oct 23, 2026
d1a4376
Remove extra surface integral test
huiyuxie Oct 23, 2026
eee3606
Remove unused AbstractGPUVector import
huiyuxie Oct 23, 2026
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
26 changes: 14 additions & 12 deletions .buildkite/pipeline.yml
Original file line number Diff line number Diff line change
@@ -1,16 +1,20 @@
steps:
- label: "CUDA Test - Julia 1.10"
- label: "Test TrixiCUDA.jl - Julia {{matrix.julia}}"
matrix:
setup:
julia: ["1.10", "1.11", "1.12", "1.13"]
plugins:
- JuliaCI/julia#v1:
version: "1.10"
version: "{{matrix.julia}}"
command: |
julia -e 'using Pkg
julia --project=. -e 'using Pkg

println("--- :julia: Instantiating environment")
Pkg.develop(path=".")
Pkg.update()
Pkg.status()

println("+++ :julia: Running tests")
Pkg.test("TrixiCUDA")'
Pkg.test()'
agents:
queue: "cuda"
if: build.message !~ /\[skip tests\]/
Expand All @@ -21,18 +25,16 @@ steps:
- JuliaCI/julia#v1:
version: "1.10"
command: |
julia --project -e '
julia --project=docs -e '
println("--- :julia: Instantiating project")
using Pkg
Pkg.develop(path=pwd())
Pkg.instantiate()
Pkg.build()
Pkg.activate("docs")
Pkg.instantiate()
push!(LOAD_PATH, @__DIR__)

println("+++ :julia: Building documentation")
include("docs/make.jl")'
artifact_paths: "docs/build/**/*"
agents:
queue: "cuda"
if: build.message !~ /\[skip docs\]/
timeout_in_minutes: 10
if: build.pull_request.id != null && build.message !~ /\[skip docs\]/
timeout_in_minutes: 15
8 changes: 6 additions & 2 deletions .github/workflows/Documenter.yml
Original file line number Diff line number Diff line change
Expand Up @@ -5,7 +5,11 @@ on:
branches:
- main
tags: "*"
pull_request:
workflow_dispatch:

concurrency:
group: ${{ github.workflow }}-${{ github.ref }}
cancel-in-progress: true

jobs:
build:
Expand Down Expand Up @@ -35,4 +39,4 @@ jobs:
env:
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }} # for Documenter via GitHub Actions
DOCUMENTER_KEY: ${{ secrets.DOCUMENTER_KEY }} # optional: SSH deploy key
run: julia --project=docs/ --color=yes docs/make.jl
run: julia --project=docs/ --color=yes docs/make.jl
21 changes: 0 additions & 21 deletions .github/workflows/FormatCheck.yml

This file was deleted.

16 changes: 0 additions & 16 deletions .github/workflows/SpellCheck.yml

This file was deleted.

26 changes: 26 additions & 0 deletions .github/workflows/Style.yml
Original file line number Diff line number Diff line change
@@ -0,0 +1,26 @@
name: Style

on:
pull_request:
push:
branches: [main]
workflow_dispatch:

concurrency:
group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }}
cancel-in-progress: true

jobs:
style:
name: Formatting and spelling
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v5
- name: Check spelling
uses: crate-ci/typos@v1.39.0
with:
config: .typos.toml
- name: Check formatting
uses: julia-actions/julia-format@v4
with:
version: '2.14.0'
15 changes: 8 additions & 7 deletions Project.toml
Original file line number Diff line number Diff line change
Expand Up @@ -5,23 +5,24 @@ version = "1.0.0-DEV"

[deps]
CUDA = "052768ef-5323-5732-b1bb-66c8b64840ba"
GPUArraysCore = "46192b85-c4d5-4398-a991-12ede77f4527"
Printf = "de0858da-6303-5e67-8744-51eddeeeb8d7"
SciMLBase = "0bca4576-84f4-4d90-8ffe-ffa030f20462"
StaticArrays = "90137ffa-7385-5640-81b9-e52037218182"
Trixi = "a7f1ee26-1774-49b1-8366-f1abc58fbfcb"

[compat]
Aqua = "0.8"
CUDA = "5"
SciMLBase = "2"
StaticArrays = "1"
CUDA = "6.4.2"
GPUArraysCore = "0.2.1"
Printf = "1.10"
SciMLBase = "3.58"
StaticArrays = "1.9.23"
Test = "1.10"
Trixi = "0.11 - 0.13.5"
Trixi = "0.17.16"
julia = "1.10"

[extras]
Aqua = "4c88cf16-eb10-579e-8560-4a9242c79595"
Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"

[targets]
test = ["Aqua", "Test"]
test = ["Test"]
2 changes: 1 addition & 1 deletion README.md
Original file line number Diff line number Diff line change
Expand Up @@ -92,7 +92,7 @@ coordinates_max = 1.0

mesh = TreeMesh(coordinates_min, coordinates_max,
initial_refinement_level = 4,
n_cells_max = 30_000)
periodicity = true)

semi = SemidiscretizationHyperbolicGPU(mesh, equations, initial_condition_convergence_test,
solver)
Expand Down
12 changes: 7 additions & 5 deletions benchmark/advection_basic_1d.jl
Original file line number Diff line number Diff line change
Expand Up @@ -17,11 +17,13 @@ coordinates_max = 1.0f0

mesh = TreeMesh(coordinates_min, coordinates_max,
initial_refinement_level = 3,
n_cells_max = 30_000, RealT = RealT)
periodicity = true,
RealT = RealT)

# Cache initialization
semi = SemidiscretizationHyperbolic(mesh, equations,
initial_condition_convergence_test, solver)
initial_condition_convergence_test, solver,
boundary_conditions = boundary_condition_periodic)
semi_gpu = SemidiscretizationHyperbolicGPU(mesh, equations,
initial_condition_convergence_test, solver_gpu)

Expand Down Expand Up @@ -51,7 +53,7 @@ u_gpu = TrixiCUDA.wrap_array(u_gpu_, mesh_gpu, equations_gpu, solver_gpu, cache_
du_gpu = TrixiCUDA.wrap_array(du_gpu_, mesh_gpu, equations_gpu, solver_gpu, cache_gpu)

# Warm up
Trixi.rhs!(du, u, t, mesh, equations, boundary_conditions, source_terms, solver, cache)
Trixi.rhs_hyperbolic!(nothing, du, u, t, mesh, equations, boundary_conditions, source_terms, solver, cache)
TrixiCUDA.rhs_gpu!(du_gpu, u_gpu, t_gpu, mesh_gpu, equations_gpu, boundary_conditions_gpu,
source_terms_gpu, solver_gpu, cache_gpu, cache_cpu)

Expand All @@ -60,8 +62,8 @@ dofs = Trixi.ndofsglobal(semi)

# Benchmark on CPU and GPU
@info "Benchmarking rhs! on CPU"
cpu_trial = @benchmark Trixi.rhs!(du, u, t, mesh, equations, boundary_conditions, source_terms,
solver, cache)
cpu_trial = @benchmark Trixi.rhs_hyperbolic!(nothing, du, u, t, mesh, equations, boundary_conditions, source_terms,
solver, cache)

@info "Benchmarking rhs! on GPU"
gpu_trial = @benchmark CUDA.@sync TrixiCUDA.rhs_gpu!(du_gpu, u_gpu, t_gpu, mesh_gpu, equations_gpu,
Expand Down
12 changes: 7 additions & 5 deletions benchmark/advection_basic_2d.jl
Original file line number Diff line number Diff line change
Expand Up @@ -17,11 +17,13 @@ coordinates_max = (1.0f0, 1.0f0)

mesh = TreeMesh(coordinates_min, coordinates_max,
initial_refinement_level = 3,
n_cells_max = 30_000, RealT = RealT)
periodicity = true,
RealT = RealT)

# Cache initialization
semi = SemidiscretizationHyperbolic(mesh, equations,
initial_condition_convergence_test, solver)
initial_condition_convergence_test, solver,
boundary_conditions = boundary_condition_periodic)
semi_gpu = SemidiscretizationHyperbolicGPU(mesh, equations,
initial_condition_convergence_test, solver_gpu)

Expand Down Expand Up @@ -51,7 +53,7 @@ u_gpu = TrixiCUDA.wrap_array(u_gpu_, mesh_gpu, equations_gpu, solver_gpu, cache_
du_gpu = TrixiCUDA.wrap_array(du_gpu_, mesh_gpu, equations_gpu, solver_gpu, cache_gpu)

# Warm up
Trixi.rhs!(du, u, t, mesh, equations, boundary_conditions, source_terms, solver, cache)
Trixi.rhs_hyperbolic!(nothing, du, u, t, mesh, equations, boundary_conditions, source_terms, solver, cache)
TrixiCUDA.rhs_gpu!(du_gpu, u_gpu, t_gpu, mesh_gpu, equations_gpu, boundary_conditions_gpu,
source_terms_gpu, solver_gpu, cache_gpu, cache_cpu)

Expand All @@ -60,8 +62,8 @@ dofs = Trixi.ndofsglobal(semi)

# Benchmark on CPU and GPU
@info "Benchmarking rhs! on CPU"
cpu_trial = @benchmark Trixi.rhs!(du, u, t, mesh, equations, boundary_conditions, source_terms,
solver, cache)
cpu_trial = @benchmark Trixi.rhs_hyperbolic!(nothing, du, u, t, mesh, equations, boundary_conditions, source_terms,
solver, cache)

@info "Benchmarking rhs! on GPU"
gpu_trial = @benchmark CUDA.@sync TrixiCUDA.rhs_gpu!(du_gpu, u_gpu, t_gpu, mesh_gpu, equations_gpu,
Expand Down
12 changes: 7 additions & 5 deletions benchmark/advection_basic_3d.jl
Original file line number Diff line number Diff line change
Expand Up @@ -17,11 +17,13 @@ coordinates_max = (1.0f0, 1.0f0, 1.0f0)

mesh = TreeMesh(coordinates_min, coordinates_max,
initial_refinement_level = 3,
n_cells_max = 30_000, RealT = RealT)
periodicity = true,
RealT = RealT)

# Cache initialization
semi = SemidiscretizationHyperbolic(mesh, equations,
initial_condition_convergence_test, solver)
initial_condition_convergence_test, solver,
boundary_conditions = boundary_condition_periodic)
semi_gpu = SemidiscretizationHyperbolicGPU(mesh, equations,
initial_condition_convergence_test, solver_gpu)

Expand Down Expand Up @@ -51,7 +53,7 @@ u_gpu = TrixiCUDA.wrap_array(u_gpu_, mesh_gpu, equations_gpu, solver_gpu, cache_
du_gpu = TrixiCUDA.wrap_array(du_gpu_, mesh_gpu, equations_gpu, solver_gpu, cache_gpu)

# Warm up
Trixi.rhs!(du, u, t, mesh, equations, boundary_conditions, source_terms, solver, cache)
Trixi.rhs_hyperbolic!(nothing, du, u, t, mesh, equations, boundary_conditions, source_terms, solver, cache)
TrixiCUDA.rhs_gpu!(du_gpu, u_gpu, t_gpu, mesh_gpu, equations_gpu, boundary_conditions_gpu,
source_terms_gpu, solver_gpu, cache_gpu, cache_cpu)

Expand All @@ -60,8 +62,8 @@ dofs = Trixi.ndofsglobal(semi)

# Benchmark on CPU and GPU
@info "Benchmarking rhs! on CPU"
cpu_trial = @benchmark Trixi.rhs!(du, u, t, mesh, equations, boundary_conditions, source_terms,
solver, cache)
cpu_trial = @benchmark Trixi.rhs_hyperbolic!(nothing, du, u, t, mesh, equations, boundary_conditions, source_terms,
solver, cache)

@info "Benchmarking rhs! on GPU"
gpu_trial = @benchmark CUDA.@sync TrixiCUDA.rhs_gpu!(du_gpu, u_gpu, t_gpu, mesh_gpu, equations_gpu,
Expand Down
12 changes: 7 additions & 5 deletions benchmark/advection_mortar_2d.jl
Original file line number Diff line number Diff line change
Expand Up @@ -20,10 +20,12 @@ refinement_patches = ((type = "box", coordinates_min = (0.0f0, -1.0f0),
mesh = TreeMesh(coordinates_min, coordinates_max,
initial_refinement_level = 2,
refinement_patches = refinement_patches,
n_cells_max = 10_000, RealT = RealT)
periodicity = true,
RealT = RealT)

# Cache initialization
semi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver)
semi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver,
boundary_conditions = boundary_condition_periodic)
semi_gpu = SemidiscretizationHyperbolicGPU(mesh, equations, initial_condition, solver_gpu)

tspan = tspan_gpu = (0.0f0, 1.0f0)
Expand Down Expand Up @@ -52,7 +54,7 @@ u_gpu = TrixiCUDA.wrap_array(u_gpu_, mesh_gpu, equations_gpu, solver_gpu, cache_
du_gpu = TrixiCUDA.wrap_array(du_gpu_, mesh_gpu, equations_gpu, solver_gpu, cache_gpu)

# Warm up
Trixi.rhs!(du, u, t, mesh, equations, boundary_conditions, source_terms, solver, cache)
Trixi.rhs_hyperbolic!(nothing, du, u, t, mesh, equations, boundary_conditions, source_terms, solver, cache)
TrixiCUDA.rhs_gpu!(du_gpu, u_gpu, t_gpu, mesh_gpu, equations_gpu, boundary_conditions_gpu,
source_terms_gpu, solver_gpu, cache_gpu, cache_cpu)

Expand All @@ -61,8 +63,8 @@ dofs = Trixi.ndofsglobal(semi)

# Benchmark on CPU and GPU
@info "Benchmarking rhs! on CPU"
cpu_trial = @benchmark Trixi.rhs!(du, u, t, mesh, equations, boundary_conditions, source_terms,
solver, cache)
cpu_trial = @benchmark Trixi.rhs_hyperbolic!(nothing, du, u, t, mesh, equations, boundary_conditions, source_terms,
solver, cache)

@info "Benchmarking rhs! on GPU"
gpu_trial = @benchmark CUDA.@sync TrixiCUDA.rhs_gpu!(du_gpu, u_gpu, t_gpu, mesh_gpu, equations_gpu,
Expand Down
12 changes: 7 additions & 5 deletions benchmark/advection_mortar_3d.jl
Original file line number Diff line number Diff line change
Expand Up @@ -22,10 +22,12 @@ refinement_patches = ((type = "box", coordinates_min = (0.0f0, -1.0f0, -1.0f0),
mesh = TreeMesh(coordinates_min, coordinates_max,
initial_refinement_level = 2,
refinement_patches = refinement_patches,
n_cells_max = 10_000, RealT = RealT)
periodicity = true,
RealT = RealT)

# Cache initialization
semi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver)
semi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver,
boundary_conditions = boundary_condition_periodic)
semi_gpu = SemidiscretizationHyperbolicGPU(mesh, equations, initial_condition, solver_gpu)

tspan = tspan_gpu = (0.0f0, 5.0f0)
Expand Down Expand Up @@ -54,7 +56,7 @@ u_gpu = TrixiCUDA.wrap_array(u_gpu_, mesh_gpu, equations_gpu, solver_gpu, cache_
du_gpu = TrixiCUDA.wrap_array(du_gpu_, mesh_gpu, equations_gpu, solver_gpu, cache_gpu)

# Warm up
Trixi.rhs!(du, u, t, mesh, equations, boundary_conditions, source_terms, solver, cache)
Trixi.rhs_hyperbolic!(nothing, du, u, t, mesh, equations, boundary_conditions, source_terms, solver, cache)
TrixiCUDA.rhs_gpu!(du_gpu, u_gpu, t_gpu, mesh_gpu, equations_gpu, boundary_conditions_gpu,
source_terms_gpu, solver_gpu, cache_gpu, cache_cpu)

Expand All @@ -63,8 +65,8 @@ dofs = Trixi.ndofsglobal(semi)

# Benchmark on CPU and GPU
@info "Benchmarking rhs! on CPU"
cpu_trial = @benchmark Trixi.rhs!(du, u, t, mesh, equations, boundary_conditions, source_terms,
solver, cache)
cpu_trial = @benchmark Trixi.rhs_hyperbolic!(nothing, du, u, t, mesh, equations, boundary_conditions, source_terms,
solver, cache)

@info "Benchmarking rhs! on GPU"
gpu_trial = @benchmark CUDA.@sync TrixiCUDA.rhs_gpu!(du_gpu, u_gpu, t_gpu, mesh_gpu, equations_gpu,
Expand Down
12 changes: 7 additions & 5 deletions benchmark/euler_ec_1d.jl
Original file line number Diff line number Diff line change
Expand Up @@ -22,10 +22,12 @@ coordinates_min = (-2.0f0,)
coordinates_max = (2.0f0,)
mesh = TreeMesh(coordinates_min, coordinates_max,
initial_refinement_level = 3,
n_cells_max = 10_000, RealT = RealT)
periodicity = true,
RealT = RealT)

# Cache initialization
semi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver)
semi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver,
boundary_conditions = boundary_condition_periodic)
semi_gpu = SemidiscretizationHyperbolicGPU(mesh, equations, initial_condition, solver_gpu)

tspan = tspan_gpu = (0.0f0, 0.4f0)
Expand Down Expand Up @@ -54,7 +56,7 @@ u_gpu = TrixiCUDA.wrap_array(u_gpu_, mesh_gpu, equations_gpu, solver_gpu, cache_
du_gpu = TrixiCUDA.wrap_array(du_gpu_, mesh_gpu, equations_gpu, solver_gpu, cache_gpu)

# Warm up
Trixi.rhs!(du, u, t, mesh, equations, boundary_conditions, source_terms, solver, cache)
Trixi.rhs_hyperbolic!(nothing, du, u, t, mesh, equations, boundary_conditions, source_terms, solver, cache)
TrixiCUDA.rhs_gpu!(du_gpu, u_gpu, t_gpu, mesh_gpu, equations_gpu, boundary_conditions_gpu,
source_terms_gpu, solver_gpu, cache_gpu, cache_cpu)

Expand All @@ -63,8 +65,8 @@ dofs = Trixi.ndofsglobal(semi)

# Benchmark on CPU and GPU
@info "Benchmarking rhs! on CPU"
cpu_trial = @benchmark Trixi.rhs!(du, u, t, mesh, equations, boundary_conditions, source_terms,
solver, cache)
cpu_trial = @benchmark Trixi.rhs_hyperbolic!(nothing, du, u, t, mesh, equations, boundary_conditions, source_terms,
solver, cache)

@info "Benchmarking rhs! on GPU"
gpu_trial = @benchmark CUDA.@sync TrixiCUDA.rhs_gpu!(du_gpu, u_gpu, t_gpu, mesh_gpu, equations_gpu,
Expand Down
Loading
Loading