Run CodSpeed benchmarks in parallel shards (#4597)

Author: patrick91

.github/workflows/test.yml

@@ -124,11 +124,16 @@ jobs:
           path: coverage_html_report
 
   benchmarks:
-    name: 📈 Benchmarks
+    name: 📈 Benchmarks (${{ matrix.shard }}/4)
 
     # Using this version because CodSpeed doesn't support Ubuntu 24.04 LTS yet
     runs-on: ubuntu-22.04
 
+    strategy:
+      fail-fast: false
+      matrix:
+        shard: [1, 2, 3, 4]
+
     permissions:
       contents: read
       id-token: write
@@ -143,13 +148,17 @@ jobs:
           enable-cache: true
           python-version: "3.14"
 
-      - run: uv sync
+      - run: uv sync --locked
 
       - name: Run benchmarks
         uses: CodSpeedHQ/action@4296e51e7041e24dadb86d1d6e8b9320d223dbe8 # v5.0.3
         with:
           mode: simulation
-          run: uv run pytest tests/benchmarks --codspeed -p no:codeflash-benchmark
+          run: >-
+            uv run pytest tests/benchmarks --codspeed
+            -p no:codeflash-benchmark
+            --test-group=${{ matrix.shard }}
+            --test-group-count=4
 
   lint:
     name: ✨ Lint

pyproject.toml

@@ -83,6 +83,7 @@ dev = [
   "pytest-emoji>=0.2.0,<0.3.0",
   "pytest-mock>=3.10,<4.0",
   "pytest-snapshot>=0.9.0,<0.10.0",
+  "pytest-test-groups>=1.2.1,<2.0.0",
   "pytest-xdist[psutil]>=3.1.0,<4.0.0",
   "python-multipart>=0.0.7",
   "rich>=12.5.1",