Add benchmarking suite with trackable performance

.github/workflows/benchmarks.yml

@@ -0,0 +1,60 @@
+name: Performance Regression Tests
+on:
+  push:
+    branches:
+      - main
+
+permissions:
+  contents: write
+  deployments: write
+
+jobs:
+  benchmark:
+    name: Run julia benchmark example
+    runs-on: ${{ matrix.os }}
+    strategy:
+      fail-fast: false
+      matrix:
+        version:
+          - '1.10'
+        os:
+          - ubuntu-latest
+        arch:
+          - x64
+    steps:
+      - uses: actions/checkout@v4
+      - uses: julia-actions/setup-julia@v1
+        with:
+          version: ${{ matrix.version }}
+          arch: ${{ matrix.arch }}
+      - uses: actions/cache@v4
+        env:
+          cache-name: cache-artifacts
+        with:
+          path: ~/.julia/artifacts
+          key: runner.os−test−env.cache−name−{{ hashFiles('**/Project.toml') }}
+          restore-keys: |
+            runner.os−test−
+            ${{ env.cache-name }}-
+            ${{ runner.os }}-test-
+            ${{ runner.os }}-
+      - name: Run benchmark
+        run: |
+          julia --project=GalerkinToolkitExamples -e '
+            using Pkg
+            Pkg.develop(path=".")
+            include("GalerkinToolkitExamples/benchmarks/run_benchmarks.jl")'
+
+      - name: Store benchmark result
+        uses: benchmark-action/github-action-benchmark@v1
+        with:
+          name: Julia benchmark result
+          tool: 'julia'
+          output-file-path: output.json
+          # Use personal access token instead of GITHUB_TOKEN due to https://github.community/t/github-action-not-triggering-gh-pages-upon-push/16096
+          github-token: ${{ secrets.GITHUB_TOKEN }}
+          auto-push: true
+          # Show alert with commit comment on detecting possible performance regression
+          alert-threshold: '200%'
+          comment-on-alert: true
+          fail-on-alert: true

GalerkinToolkitExamples/Project.toml

@@ -20,6 +20,7 @@ SparseArrays = "2f01184e-e22b-5df5-ae63-d93ebab69eaf"
 StaticArrays = "90137ffa-7385-5640-81b9-e52037218182"
 TimerOutputs = "a759f4b9-e2f1-59dc-863e-4aeb61b1ea8f"
 WriteVTK = "64499a7a-5c06-52f2-abe2-ccb03c286192"
+BenchmarkTools = "6e4b80f9-dd63-53aa-95a3-0cdb28fa8baf"
 
 [extras]
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"

GalerkinToolkitExamples/benchmarks/run_benchmarks.jl

@@ -0,0 +1,41 @@
+module GalerkinToolkitBenchmarkTests
+
+using BenchmarkTools
+
+import GalerkinToolkit as GT
+using GalerkinToolkitExamples: Poisson
+
+
+function handwritten_poisson(n)
+	"""
+	Runs the hand-written Poisson example code, for a 3D
+	mesh of dimensions n x n x n.
+	"""
+
+	mesh = GT.cartesian_mesh((0,2,0,2,0,2), (n,n,n))
+
+	params = Dict{Symbol,Any}()
+	params[:implementation] = :hand_written
+	params[:mesh] = mesh
+	params[:dirichlet_tags] = ["1-face-1","1-face-2","1-face-3","1-face-4"]
+	params[:discretization_method] = :continuous_galerkin
+	params[:dirichlet_method] = :strong
+	params[:integration_degree] = 2
+
+	Poisson.main(params)
+end
+
+
+# Build a benchmark suite for the Poisson example
+suite = BenchmarkGroup()
+suite["poisson-hand"] = BenchmarkGroup(["Poisson", "handwritten", "3D"])
+suite["poisson-hand"]["n=10"] = @benchmarkable handwritten_poisson(10)
+
+# Run all benchmarks
+tune!(suite)
+results = run(suite, verbose = true)
+
+# Save benchmark results for tracking and visualization
+BenchmarkTools.save("output.json", median(results))
+
+end # module

docs/src/developers_guide.md

@@ -1,3 +1,10 @@
 # Developers guide
 
+## Performance Benchmarks
+There is a benchmark suite defined in `GalerkinToolkitExamples/benchmarks`. This uses [BenchmarkTools.jl](https://github.com/JuliaCI/BenchmarkTools.jl) to perform the timings
+and [github-action-benchmark](https://github.com/benchmark-action/github-action-benchmark) to collect the results and store them in the `gh-pages` branch. Graphs of performance
+changes over time (per commit hash) can be viewed here: <https://galerkintoolkit.github.io/GalerkinToolkit.jl/dev/bench/>.
 
+The github action can be configured (in `.github/workflows/benchmarks.yml`) to fail if the performance change is beyond a given threshold. Look for the `alert-threshold:` and `fail-on-alert:` keys.
+
+More benchmarks can be added (or existing ones modified) in `GalerkinToolkitExamples/benchmarks/run_benchmarks.jl`.