Adding nj_hint

.github/workflows/C++.yml

@@ -1,11 +1,6 @@
 name: C++ Build
 
 on:
-  push:
-    branches:
-      - master
-    tags:
-      - v*
   pull_request:
     branches:
       - master

.github/workflows/build_cufinufft_wheels.yml

@@ -1,6 +1,22 @@
 name: Build cufinufft Python wheels
 
-on: [push, pull_request]
+on:
+  # Weekly schedule (Sunday 00:00 UTC)
+  schedule:
+    - cron: "0 0 * * 0"
+
+  # On tag push
+  push:
+    tags:
+      - 'v*'
+  # Manual trigger (with an optional deploy toggle)
+  workflow_dispatch:
+    inputs:
+      deploy:
+        description: "Deploy after build?"
+        type: boolean
+        required: false
+        default: false
 
 jobs:
   build_wheels:
@@ -17,9 +33,16 @@ jobs:
       - uses: ilammy/msvc-dev-cmd@v1
       - name: Setup CUDA
         if: ${{ matrix.buildplat[0] == 'windows-2022' }}
-        uses: Jimver/[email protected].21
+        uses: Jimver/[email protected].24
         with:
           cuda: '12.4.0'
+          method: 'network'
+          linux-local-args: '["--toolkit"]'
+          log-file-suffix: '${{matrix.os}}-cuda-install.txt'
+          sub-packages: ${{runner.os == 'Linux' && '["nvcc", "cudart"]' || '["nvcc", "cudart", "cufft", "cufft_dev"]'}}
+          non-cuda-sub-packages: ${{runner.os == 'Linux' && '["libcufft","libcufft-dev"]' || '[]'}}
+          use-local-cache: 'false'
+          use-github-cache: 'false' #it is not working at the moment https://github.com/Jimver/cuda-toolkit/issues/390
       - name: Build ${{ matrix.buildplat[1] }} wheels
         uses: pypa/[email protected]
         with:

.github/workflows/build_finufft_wheels.yml

@@ -1,6 +1,23 @@
 name: Build and test finufft Python wheels
 
-on: [ push, pull_request ]
+on:
+  # Weekly schedule (Sunday 00:00 UTC)
+  schedule:
+    - cron: "0 0 * * 0"
+
+  # On tag push
+  push:
+    tags:
+      - 'v*'
+  # Manual trigger (with an optional deploy toggle)
+  workflow_dispatch:
+    inputs:
+      deploy:
+        description: "Deploy after build?"
+        type: boolean
+        required: false
+        default: false
+
 
 jobs:
   build_wheels:

.github/workflows/cmake_ci.yml

@@ -3,18 +3,6 @@ name: cmake ci linux macos windows
 on: [push, pull_request]
 
 jobs:
-  prepare:
-    runs-on: ubuntu-22.04
-    outputs:
-      matrix: ${{ steps.generate_matrix.outputs.matrix }}
-    steps:
-      - name: Checkout code
-        uses: actions/checkout@v4
-      - name: Generate matrix
-        id: generate_matrix
-        run: |
-          MATRIX=$(python3 ${{ github.workspace }}/.github/workflows/generate_cmake_matrix.py)
-          echo "matrix=$MATRIX" >> $GITHUB_OUTPUT
   pip-requirements:
     runs-on: ubuntu-22.04
     steps:
@@ -23,7 +11,7 @@ jobs:
       - name: Set up Python
         uses: actions/setup-python@v5
         with:
-          cache : 'pip'
+          cache: 'pip'
           python-version: '3.10'
       - name: Generate requirements.txt
         run: |
@@ -38,142 +26,193 @@ jobs:
         with:
           name: requirements
           path: requirements.txt
+
   cache:
     strategy:
       matrix:
-        os:
-         - ubuntu-22.04
-         - windows-2022
+        os: [ubuntu-22.04, windows-2022]
     runs-on: ${{ matrix.os }}
     steps:
       - name: Checkout code
         uses: actions/checkout@v4
       - name: create cache directory
-        run: |
-          mkdir -p cpm
+        shell: bash
+        run: mkdir -p "cpm"
       - name: Check if cache exists
         id: cache
         uses: actions/cache@v4
         with:
-          key: cpm-cache-00-${{ runner.os == 'windows' && 'windows-' || '' }}${{ hashFiles('CMakeLists.txt', 'cmake/*') }}
+          key: cpm-cache-00-${{ runner.os == 'windows' && 'windows-' || '' }}${{ hashFiles('**/CMakeLists.txt', 'cmake/**') }}
           enableCrossOsArchive: true
           path: cpm
-      - name: Setup Cpp
+      - name: Setup Cpp (on cache miss)
         if: steps.cache.outputs.cache-hit != 'true'
         uses: aminya/setup-cpp@v1
         with:
-          cmake : true
-      - name: Download dependencies in cache linux
+          cmake: true
+      - name: Download dependencies in cache (Linux)
         if: steps.cache.outputs.cache-hit != 'true' && runner.os == 'Linux'
         run: |
-          cmake -S . -B ./build
+          cmake -S . -B ./build -DCPM_SOURCE_CACHE="cpm"
           rm -rf build
-          cmake -S . -B ./build -DFINUFFT_USE_DUCC0=ON
-        env:
-          CPM_SOURCE_CACHE: cpm
-      - name: Download dependencies in cache windows
+          cmake -S . -B ./build -DFINUFFT_USE_DUCC0=ON -DCPM_SOURCE_CACHE="cpm"
+      - name: Download dependencies in cache (Windows)
         if: steps.cache.outputs.cache-hit != 'true' && runner.os != 'Linux'
         run: |
-          cmake -S . -B build
+          cmake -S . -B build -DCPM_SOURCE_CACHE="cpm"
           rm build -r -force
-          cmake -S . -B build -DFINUFFT_USE_DUCC0=ON
-        env:
-          CPM_SOURCE_CACHE: cpm
+          cmake -S . -B build -DFINUFFT_USE_DUCC0=ON -DCPM_SOURCE_CACHE="cpm"
       - name: Cache dependencies
+        if: steps.cache.outputs.cache-hit != 'true'
         uses: actions/cache/save@v4
         with:
-          key: cpm-cache-00-${{ runner.os == 'windows' && 'windows-' || '' }}${{ hashFiles('CMakeLists.txt', 'cmake/*') }}
+          key: cpm-cache-00-${{ runner.os == 'windows' && 'windows-' || '' }}${{ hashFiles('**/CMakeLists.txt', 'cmake/**') }}
           enableCrossOsArchive: true
           path: cpm
+
   cmake-ci:
     runs-on: ${{ matrix.os }}
-    needs: [prepare, cache, pip-requirements]
+    needs: [cache, pip-requirements]
+    env:
+      CMAKE_GENERATOR: Ninja
+      CPM_SOURCE_CACHE: cpm
+      CTEST_OUTPUT_ON_FAILURE: "1"
+      SCCACHE_GHA_ENABLED: "true"
+      SCCACHE_GHA_VERSION: "0"
+      CMAKE_C_COMPILER_LAUNCHER: sccache
+      CMAKE_CXX_COMPILER_LAUNCHER: sccache
+      CMAKE_CUDA_COMPILER_LAUNCHER: sccache
     strategy:
       fail-fast: false
-      matrix: ${{ fromJSON(needs.prepare.outputs.matrix) }}
+      matrix:
+        include:
+          - { os: ubuntu-22.04, toolchain: gcc-10 }
+          - { os: ubuntu-22.04, toolchain: gcc-11 }
+          - { os: ubuntu-22.04, toolchain: gcc-12 }
+          - { os: ubuntu-22.04, toolchain: gcc-13 }
+          - { os: ubuntu-22.04, toolchain: clang-16 }
+          - { os: ubuntu-22.04, toolchain: clang-17 }
+          - { os: ubuntu-22.04, toolchain: clang-18 }
+
+          - { os: windows-2022, toolchain: msvc }
+          - { os: windows-2022, toolchain: clang-19 }
+
+          - { os: macos-13, toolchain: llvm }
+          - { os: macos-13, toolchain: gcc-14 }
+          - { os: macos-14, toolchain: llvm }
+          - { os: macos-14, toolchain: gcc-14 }
+          - { os: macos-15, toolchain: llvm }
+#          - { os: macos-15, toolchain: gcc-15 } # it is broken at the moment, but it is not caused by finufft
+
     steps:
       - name: Checkout code
         uses: actions/checkout@v4
+
       - name: Restore Cache
         uses: actions/cache/restore@v4
         with:
-          key: cpm-cache-00-${{ runner.os == 'windows' && 'windows-' || '' }}${{ hashFiles('CMakeLists.txt', 'cmake/*') }}
+          key: cpm-cache-00-${{ runner.os == 'windows' && 'windows-' || '' }}${{ hashFiles('**/CMakeLists.txt', 'cmake/**') }}
           enableCrossOsArchive: true
           path: cpm
+
       - name: Download requirements.txt
         uses: actions/download-artifact@v4
         with:
           name: requirements
+
       - name: Run sccache-cache
         uses: mozilla-actions/[email protected]
-      - name: Set caching env vars
-        run: |
-          echo "SCCACHE_GHA_ENABLED=true" >> $GITHUB_ENV
-          echo "SCCACHE_GHA_VERSION=0" >> $GITHUB_ENV
-      - name: Set up sccache on Windows
-        if: runner.os == 'Windows'
-        run: Add-Content -Path $env:GITHUB_ENV -Value "SCCACHE_GHA_ENABLED=true"
+
       - name: Setup Cpp
-        uses: aminya/setup-cpp@v1.1.1
+        uses: aminya/setup-cpp@v1
         with:
           compiler: ${{ matrix.toolchain }}
-          vcvarsall: ${{ contains(matrix.os, 'windows') }}
+          vcvarsall: true
           cmake: true
           ninja: true
-          vcpkg: false
-          cppcheck: false
-          clangtidy: false
-      - name: Set min macOS version and install fftw
-        if: runner.os == 'macOS'
+
+      - name: Install toolchain + FFTW (macOS)
+        if: startsWith(matrix.os, 'macos-')
         run: |
-          brew install fftw libomp
-      - name: Install fftw
-        if: runner.os == 'linux'
+          export HOMEBREW_NO_ANALYTICS=1 HOMEBREW_NO_INSTALL_CLEANUP=1
+          brew install libomp fftw
+          {
+            echo "LDFLAGS=-L$(brew --prefix libomp)/lib"
+            echo "CPPFLAGS=-I$(brew --prefix libomp)/include"
+            echo "LIBRARY_PATH=$(brew --prefix libomp)/lib"
+            echo "DYLD_LIBRARY_PATH=$(brew --prefix libomp)/lib"
+            echo "CMAKE_PREFIX_PATH=$(brew --prefix libomp)"
+            ver="${{ matrix.os }}"; ver="${ver#macos-}"
+            echo "MACOSX_DEPLOYMENT_TARGET=${ver}.0"
+          } >> "$GITHUB_ENV"
+
+      - name: Install fftw (Linux)
+        if: runner.os == 'Linux'
         run: |
           sudo apt update
           sudo apt install -y libfftw3-dev
-      - name: Configure Cmake
-        run: |
-          cmake -S . -B ./build -G Ninja -DCMAKE_C_COMPILER_LAUNCHER=sccache -DCMAKE_CXX_COMPILER_LAUNCHER=sccache -DCMAKE_BUILD_TYPE:STRING=${{matrix.build_type}} -DFINUFFT_ARCH_FLAGS=${{ matrix.arch_flags }} -DFINUFFT_BUILD_TESTS=ON -DFINUFFT_STATIC_LINKING=${{matrix.finufft_static_linking}} -DFINUFFT_USE_DUCC0=${{ matrix.ducc_fft }} -DFINUFFT_ENABLE_SANITIZERS=${{matrix.build_type == 'Debug'}}
-        env:
-          CPM_SOURCE_CACHE: cpm
-      - name: Build
-        run: |
-          cmake --build ./build --config ${{matrix.build_type}}
-      - name: Test
-        working-directory: ./build
-        run: |
-          ctest -C ${{matrix.build_type}} --output-on-failure
+
       - name: Set up Python
-        if: matrix.finufft_static_linking == 'off'
         uses: actions/setup-python@v5
         with:
           python-version: '3.10'
           cache: 'pip'
           cache-dependency-path: requirements.txt
-      - name: Build Python wheels
-        if: matrix.finufft_static_linking == 'off'
-        env:
-          MACOSX_DEPLOYMENT_TARGET: 13
-          CPM_SOURCE_CACHE: cpm
-        shell: bash
+
+      - name: Install Python requirements
         run: |
           python3 -m pip install --upgrade pip
           python3 -m pip install -r requirements.txt
-          python3 -m pip install \
-            --verbose \
-            -C cmake.define.CMAKE_BUILD_TYPE=${{ matrix.build_type }} \
-            -C cmake.define.FINUFFT_ARCH_FLAGS=${{ matrix.arch_flags }} \
-            -C cmake.define.FINUFFT_USE_DUCC0=${{ matrix.ducc_fft }} \
-            -C cmake.define.CMAKE_CXX_COMPILER_LAUNCHER=sccache \
-            -C cmake.define.CMAKE_C_COMPILER_LAUNCHER=sccache \
-            -C cmake.define.CMAKE_GENERATOR=Ninja \
-            -C cmake.define.CMAKE_GENERATOR_PLATFORM= \
-            python/finufft
-      - name: Test Python package
-        if: matrix.finufft_static_linking == 'off'
+
+      - name: Build and test configurations
+        shell: bash
         run: |
-          python3 -m pytest python/finufft/test
+          if [[ "$RUNNER_OS" == "Windows" ]]; then
+            configs=("build_type=Release ducc_fft=On" "build_type=Debug ducc_fft=On")
+          else
+            configs=("build_type=Release ducc_fft=On" "build_type=Release ducc_fft=Off" \
+                     "build_type=Debug ducc_fft=On"  "build_type=Debug ducc_fft=Off")
+          fi
+
+          if [[ "${{ matrix.toolchain }}" == "msvc" ]]; then
+            arch_flags=("/arch:AVX2" "/arch:SSE2" "native")
+          elif [[ "$RUNNER_OS" == "Linux" ]]; then
+            arch_flags=("-march=x86-64" "-march=x86-64-v2" "-march=x86-64-v3" "-march=native")
+          else
+            arch_flags=("native")
+          fi
+
+          for arch in "${arch_flags[@]}"; do
+            arch_dir=$(echo "$arch" | tr -c 'A-Za-z0-9' '_')
+            for cfg in "${configs[@]}"; do
+              eval "$cfg"
+              build_dir="build_${build_type}_${ducc_fft}_${arch_dir}"
+
+              cmake -E make_directory "$build_dir"
+              cmake -S . -B "$build_dir" \
+                -DCMAKE_BUILD_TYPE=$build_type \
+                -DFINUFFT_ARCH_FLAGS="$arch" \
+                -DFINUFFT_USE_DUCC0=$ducc_fft \
+                -DCMAKE_MSVC_DEBUG_INFORMATION_FORMAT=Embedded \
+                -DCMAKE_POLICY_DEFAULT_CMP0141=NEW
+                # CMake will pick the toolchain set by setup-cpp (and macOS brew block for gcc-14).
+
+              cmake --build "$build_dir" --config "$build_type"
+              (cd "$build_dir" && ctest -C "$build_type")
+
+              python3 -m pip install \
+                --verbose \
+                -C cmake.define.CMAKE_BUILD_TYPE=$build_type \
+                -C cmake.define.FINUFFT_ARCH_FLAGS="$arch" \
+                -C cmake.define.FINUFFT_USE_DUCC0=$ducc_fft \
+                -C cmake.define.CMAKE_MSVC_DEBUG_INFORMATION_FORMAT=Embedded \
+                -C cmake.define.CMAKE_POLICY_DEFAULT_CMP0141=NEW \
+                python/finufft
+
+              python3 -m pytest python/finufft/test
+            done
+          done
+
   cleanup:
     runs-on: ubuntu-22.04
     needs: cmake-ci