ND, feature transform, tests, performance improvements

.github/workflows/bench_nd_profile.yml

@@ -0,0 +1,108 @@
+name: ND Benchmark Profile
+
+on:
+  workflow_dispatch:
+
+jobs:
+  bench:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Set up Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.11"
+
+      - name: Install dependencies
+        run: |
+          python -m pip install --upgrade pip
+          python -m pip install numpy
+
+      - name: Install package (editable)
+        run: python -m pip install -e .
+
+      - name: Run legacy vs ND benchmark
+        run: |
+          set -euo pipefail
+          # Dense (structured) templates
+          python scripts/bench_explicit_threads.py \
+            --parallels 1,2,4,8 --reps 20 --dtype uint16 --profile \
+            --shapes 128x128,512x512,1024x1024,2048x2048 --template structured --grid-base 8x8 \
+            --output benchmarks/legacy_vs_nd_explicit_ci.template_structured_2d.csv
+          python scripts/bench_explicit_threads.py \
+            --parallels 1,2,4,8 --reps 20 --dtype uint16 --profile \
+            --shapes 96x96x96,192x192x192 --template structured --grid-base 8x8x8 \
+            --output benchmarks/legacy_vs_nd_explicit_ci.template_structured_3d.csv
+
+          # Sparse (mod8) templates
+          python scripts/bench_explicit_threads.py \
+            --parallels 1,2,4,8 --reps 20 --dtype uint16 --profile \
+            --shapes 128x128,512x512,1024x1024,2048x2048 --template structured_mod8 --grid-base 8x8 \
+            --output benchmarks/legacy_vs_nd_explicit_ci.template_structured_mod8_2d.csv
+          python scripts/bench_explicit_threads.py \
+            --parallels 1,2,4,8 --reps 20 --dtype uint16 --profile \
+            --shapes 96x96x96,192x192x192 --template structured_mod8 --grid-base 8x8x8 \
+            --output benchmarks/legacy_vs_nd_explicit_ci.template_structured_mod8_3d.csv
+
+          # Circle templates (2D only)
+          python scripts/bench_explicit_threads.py \
+            --parallels 1,2,4,8 --reps 20 --dtype uint16 --profile \
+            --shapes 128x128,512x512,1024x1024,2048x2048 --template circles_small \
+            --output benchmarks/legacy_vs_nd_explicit_ci.template_circles_small_2d.csv
+          python scripts/bench_explicit_threads.py \
+            --parallels 1,2,4,8 --reps 20 --dtype uint16 --profile \
+            --shapes 128x128,512x512,1024x1024,2048x2048 --template circles_large \
+            --output benchmarks/legacy_vs_nd_explicit_ci.template_circles_large_2d.csv
+
+      - name: Summarize results
+        run: |
+          python - <<'PY'
+          import csv
+          import os
+          from pathlib import Path
+
+          header = [
+              "shape",
+              "dims",
+              "parallel",
+              "legacy_ms",
+              "nd_ms",
+              "ratio",
+              "nd_parallel_used",
+              "max_abs_diff",
+          ]
+
+          def render_table(title, csv_path):
+              title = title.replace("\n", " ").strip()
+              if not csv_path.exists():
+                  return f"### {title}\n\nMissing CSV: `{csv_path}`\n"
+              with csv_path.open() as fp:
+                  reader = csv.DictReader(fp)
+                  rows = list(reader)
+              if not rows:
+                  return f"### {title}\n\nNo benchmark rows captured.\n"
+              table = [f"### {title}\n", "| " + " | ".join(header) + " |"]
+              table.append("| " + " | ".join(["---"] * len(header)) + " |")
+              for row in rows:
+                  table.append("| " + " | ".join(row[h] for h in header) + " |")
+              return "\n".join(table) + "\n"
+
+          sections = []
+          sections.append(render_table("Structured (dense) 2D", Path("benchmarks/legacy_vs_nd_explicit_ci.template_structured_2d.csv")))
+          sections.append(render_table("Structured (dense) 3D", Path("benchmarks/legacy_vs_nd_explicit_ci.template_structured_3d.csv")))
+          sections.append(render_table("Structured mod8 (sparse) 2D", Path("benchmarks/legacy_vs_nd_explicit_ci.template_structured_mod8_2d.csv")))
+          sections.append(render_table("Structured mod8 (sparse) 3D", Path("benchmarks/legacy_vs_nd_explicit_ci.template_structured_mod8_3d.csv")))
+          sections.append(render_table("Circles small 2D", Path("benchmarks/legacy_vs_nd_explicit_ci.template_circles_small_2d.csv")))
+          sections.append(render_table("Circles large 2D", Path("benchmarks/legacy_vs_nd_explicit_ci.template_circles_large_2d.csv")))
+
+          summary_path = Path(os.environ["GITHUB_STEP_SUMMARY"])
+          summary_path.write_text("## ND Benchmark Profile\n\n" + "\n".join(sections))
+          PY
+
+      - name: Upload benchmark CSV
+        uses: actions/upload-artifact@v4
+        with:
+          name: legacy-vs-nd-explicit-results
+          path: benchmarks/legacy_vs_nd_explicit_ci.template_*.csv
+          if-no-files-found: error

.github/workflows/build_wheels.yml

@@ -1,45 +1,54 @@
 name: Build Wheels
 
 on:
-  workflow_dispatch:
+  workflow_dispatch: {}
   push:
     tags:
       - '*'
 env:
-  CIBW_SKIP: pp38* pp39* pp310* *-musllinux*
+  # Supported stable CPython versions (adjust as versions EOL).
+  CIBW_BUILD: "cp310-* cp311-* cp312-* cp313-* cp314-*"
+  CIBW_SKIP: "*-musllinux*"
 
 jobs:
   build_wheels:
-    name: Build wheels on ${{matrix.arch}} for ${{ matrix.os }}
+    name: Build ${{ matrix.archs_linux }}${{ matrix.archs_windows }}${{ matrix.archs_macos }} on ${{ matrix.os }}
     runs-on: ${{ matrix.os }}
     strategy:
       matrix:
-        os: [ubuntu-latest, windows-2019, macos-latest]
-        arch: [auto]
         include:
           - os: ubuntu-latest
-            arch: aarch64
+            archs_linux: "x86_64"
+          - os: ubuntu-24.04-arm
+            archs_linux: "aarch64"
+          - os: windows-latest
+            archs_windows: "AMD64"
+          - os: windows-11-arm
+            archs_windows: "ARM64"
+            cibw_build: "cp311-* cp312-* cp313-* cp314-*"  # cp310 lacks win_arm64 support
+          - os: macos-15-intel
+            archs_macos: "x86_64"
+          - os: macos-latest
+            archs_macos: "arm64"
 
     steps:
-      - uses: actions/checkout@v2
-
-      - name: Set up QEMU
-        if:  ${{ matrix.arch == 'aarch64' }}
-        uses: docker/setup-qemu-action@v1
+      - uses: actions/checkout@v4
 
       - name: Build wheels
-        uses: pypa/cibuildwheel@v2.22.0
+        uses: pypa/cibuildwheel@v3.3.1
         with:
           output-dir: ./wheelhouse
-        # to supply options, put them in 'env', like:
         env:
-          CIBW_ARCHS_LINUX: ${{matrix.arch}}
-          CIBW_BEFORE_BUILD: pip install numpy setuptools wheel
-          CIBW_ARCHS_MACOS: "x86_64 arm64"
+          CIBW_BUILD: ${{ matrix.cibw_build || env.CIBW_BUILD }}
+          CIBW_ARCHS_LINUX: ${{ matrix.archs_linux }}
+          CIBW_ARCHS_WINDOWS: ${{ matrix.archs_windows }}
+          CIBW_ARCHS_MACOS: ${{ matrix.archs_macos }}
+          # Avoid date-stamped local versions that can cause wheel filename mismatch.
+          CIBW_ENVIRONMENT: "SETUPTOOLS_SCM_LOCAL_SCHEME=no-local-version EDT_MARCH_NATIVE=0"
 
       - name: Upload built wheels
         uses: actions/upload-artifact@v4
         with:
-          name: built-wheels-${{ matrix.os }}-${{ matrix.arch }}
+          name: built-wheels-${{ matrix.os }}-${{ matrix.archs_linux }}${{ matrix.archs_windows }}${{ matrix.archs_macos }}
           path: ./wheelhouse/*.whl
           if-no-files-found: warn

.github/workflows/tests.yml

@@ -1,6 +1,7 @@
 name: Tests
 
 on:
+  workflow_dispatch:
   push:
     branches:
       - master
@@ -14,24 +15,24 @@ jobs:
     runs-on: ${{ matrix.os }}
     strategy:
       matrix:
-        os: [ubuntu-latest, macos-latest, windows-2019]
-        python-version: ["3.8", "3.9", "3.10", "3.11", "3.12"]
+        os: [ubuntu-latest, macos-latest, windows-latest]
+        python-version: ["3.10", "3.11", "3.12", "3.13"]
 
     steps:
+      - uses: actions/checkout@v4
+
       - name: Set up Python ${{ matrix.python-version }}
-        uses: actions/setup-python@v2
+        uses: actions/setup-python@v5
         with:
           python-version: ${{ matrix.python-version }}
 
-      - uses: actions/checkout@v2
-
       - name: Install dependencies
         run: |
           python -m pip install --upgrade pip
-          python -m pip install pytest numpy scipy setuptools wheel
+          python -m pip install cython pytest numpy scipy setuptools wheel
 
-      - name: Compile
-        run: python setup.py develop
+      - name: Install package (editable)
+        run: python -m pip install -e .
 
       - name: Test with pytest
-        run: python -m pytest -v -x automated_test.py
+        run: python -m pytest -v -x tests

.gitignore

@@ -40,9 +40,12 @@
 .pytest_cache
 cpp/test
 
+# Cython-generated C++ (rebuilt on every install)
 src/edt.cpp
+legacy/edt.cpp
 
 test.py
 test2.py
 test3.py
 test4.py
+benchmarks/*.csv

AUTHORS

@@ -1,3 +1,4 @@
+Kevin Cutler <39454982+kevinjohncutler@users.noreply.github.com>
 Pavlo Hilei <45758974+Pavlik1400@users.noreply.github.com>
 William Silversmith <william.silversmith@gmail.com>
 William Silversmith <ws9@princeton.edu>