Feat: Adding Linear Algebra Dot operation support

.github/workflows/build_wheels.yml

@@ -8,9 +8,11 @@ on:
       - "quaddtype-v*"
     paths:
       - "quaddtype/**"
+      - ".github/workflows/**"
   pull_request:
     paths:
       - "quaddtype/**"
+      - ".github/workflows/**"
   workflow_dispatch:
 
 jobs:
@@ -19,12 +21,19 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v3
+        with:
+          submodules: recursive
 
       - name: Set up Python
         uses: actions/setup-python@v4
         with:
           python-version: ">=3.10.0"
 
+      - name: Verify QuadBLAS submodule
+        run: |
+          ls -la quaddtype/numpy_quaddtype/QBLAS/
+          ls -la quaddtype/numpy_quaddtype/QBLAS/include/quadblas/
+
       - name: Install cibuildwheel
         run: pip install cibuildwheel==2.20.0
 
@@ -34,16 +43,23 @@ jobs:
           CIBW_MANYLINUX_X86_64_IMAGE: manylinux_2_28
           CIBW_BUILD_VERBOSITY: "3"
           CIBW_BEFORE_ALL: |
+            yum update -y
+            yum install -y cmake gcc gcc-c++ make git pkgconfig
+            # Install SLEEF in container
             git clone --branch 3.8 https://github.com/shibatch/sleef.git
             cd sleef
-            cmake -S . -B build -DSLEEF_BUILD_QUAD:BOOL=ON -DSLEEF_BUILD_SHARED_LIBS:BOOL=ON -DCMAKE_POSITION_INDEPENDENT_CODE=ON
+            cmake -S . -B build \
+              -DSLEEF_BUILD_QUAD:BOOL=ON \
+              -DSLEEF_BUILD_SHARED_LIBS:BOOL=ON \
+              -DCMAKE_POSITION_INDEPENDENT_CODE=ON
             cmake --build build/ --clean-first -j
             cmake --install build --prefix /usr/local
           CIBW_ENVIRONMENT: >
-            CFLAGS="-I/usr/local/include $CFLAGS"
-            CXXFLAGS="-I/usr/local/include $CXXFLAGS"
-            LDFLAGS="-L/usr/local/lib64 $LDFLAGS"
-            LD_LIBRARY_PATH="/usr/local/lib64:$LD_LIBRARY_PATH"
+            CFLAGS="-I/usr/local/include -I{project}/numpy_quaddtype/QBLAS/include $CFLAGS"
+            CXXFLAGS="-I/usr/local/include -I{project}/numpy_quaddtype/QBLAS/include -fext-numeric-literals $CXXFLAGS"
+            LDFLAGS="-L/usr/local/lib64 -L/usr/local/lib -Wl,-rpath,/usr/local/lib64 -Wl,-rpath,/usr/local/lib -fopenmp $LDFLAGS"
+            LD_LIBRARY_PATH="/usr/local/lib64:/usr/local/lib:$LD_LIBRARY_PATH"
+            PKG_CONFIG_PATH="/usr/local/lib64/pkgconfig:/usr/local/lib/pkgconfig:$PKG_CONFIG_PATH"
           CIBW_REPAIR_WHEEL_COMMAND: |
             auditwheel repair -w {dest_dir} --plat manylinux_2_28_x86_64 {wheel}
           CIBW_TEST_COMMAND: |
@@ -68,46 +84,59 @@ jobs:
 
     steps:
       - uses: actions/checkout@v3
+        with:
+          submodules: recursive
 
       - name: Set up Python
         uses: actions/setup-python@v4
         with:
           python-version: "3.10"
 
+      - name: Install dependencies
+        run: |
+          brew install cmake libomp git
+
       - name: Install SLEEF
         env:
-          MACOSX_DEPLOYMENT_TARGET: "11.0"
+          MACOSX_DEPLOYMENT_TARGET: ${{ matrix.os == 'macos-13' && '13.0' || '14.0' }}
         run: |
           git clone --branch 3.8 https://github.com/shibatch/sleef.git
           cd sleef
           cmake -S . -B build \
             -DSLEEF_BUILD_QUAD:BOOL=ON \
             -DSLEEF_BUILD_SHARED_LIBS:BOOL=ON \
             -DCMAKE_POSITION_INDEPENDENT_CODE=ON \
-            -DCMAKE_OSX_DEPLOYMENT_TARGET=11.0 \
+            -DCMAKE_OSX_DEPLOYMENT_TARGET=${{ matrix.os == 'macos-13' && '13.0' || '14.0' }} \
             -DCMAKE_INSTALL_RPATH="@loader_path/../lib" \
             -DCMAKE_BUILD_WITH_INSTALL_RPATH=ON
           cmake --build build/ --clean-first -j
           sudo cmake --install build --prefix /usr/local
+
+      - name: Verify QuadBLAS submodule
+        run: |
+          ls -la quaddtype/numpy_quaddtype/QBLAS/
+          ls -la quaddtype/numpy_quaddtype/QBLAS/include/quadblas/
+
       - name: Install cibuildwheel
         run: pip install cibuildwheel==2.20.0
 
       - name: Build wheels
         env:
           CIBW_BUILD: "cp310-* cp311-* cp312-*"
           CIBW_ARCHS_MACOS: ${{ matrix.os == 'macos-13' && 'x86_64' || 'arm64' }}
-          CIBW_BUILD_VERBOSITY: "1"
+          CIBW_BUILD_VERBOSITY: "3"
           CIBW_ENVIRONMENT: >
-            MACOSX_DEPLOYMENT_TARGET="11.0"
+            MACOSX_DEPLOYMENT_TARGET="${{ matrix.os == 'macos-13' && '13.0' || '14.0' }}"
             DYLD_LIBRARY_PATH="/usr/local/lib:$DYLD_LIBRARY_PATH"
-            CFLAGS="-I/usr/local/include $CFLAGS"
-            CXXFLAGS="-I/usr/local/include $CXXFLAGS"
+            CFLAGS="-I/usr/local/include -I{project}/numpy_quaddtype/QBLAS/include $CFLAGS"
+            CXXFLAGS="-I/usr/local/include -I{project}/numpy_quaddtype/QBLAS/include $CXXFLAGS"
             LDFLAGS="-L/usr/local/lib $LDFLAGS"
+            PKG_CONFIG_PATH="/usr/local/lib/pkgconfig:$PKG_CONFIG_PATH"
           CIBW_REPAIR_WHEEL_COMMAND: >
             delocate-wheel --require-archs {delocate_archs} -w {dest_dir} -v {wheel}
           CIBW_TEST_COMMAND: |
             pip install {package}[test]
-            pytest {project}/tests
+            pytest -s {project}/tests
           CIBW_TEST_EXTRAS: "test"
         run: |
           python -m cibuildwheel --output-dir wheelhouse
@@ -118,6 +147,7 @@ jobs:
           path: ./quaddtype/wheelhouse/*.whl
           name: wheels-${{ matrix.os }}
 
+  # disabling QBLAS optimization for windows due to incompatibility with MSVC
   build_wheels_windows:
     name: Build wheels on Windows
     runs-on: windows-latest
@@ -127,6 +157,8 @@ jobs:
 
     steps:
       - uses: actions/checkout@v3
+        with:
+          submodules: recursive
 
       - name: Setup MSVC
         uses: ilammy/msvc-dev-cmd@v1
@@ -142,6 +174,12 @@ jobs:
       - name: Install CMake
         uses: lukka/get-cmake@latest
 
+      - name: Verify QuadBLAS submodule
+        shell: pwsh
+        run: |
+          Get-ChildItem quaddtype/numpy_quaddtype/QBLAS/
+          Get-ChildItem quaddtype/numpy_quaddtype/QBLAS/include/quadblas/
+
       - name: Clone and Build SLEEF
         shell: pwsh
         run: |
@@ -151,16 +189,6 @@ jobs:
           cmake --build build --config Release
           cmake --install build --prefix "C:/sleef" --config Release
 
-      - name: Setup build environment
-        shell: pwsh
-        run: |
-          $env:INCLUDE += ";C:\sleef\include"
-          $env:LIB += ";C:\sleef\lib"
-          $env:PATH = "C:\sleef\bin;$env:PATH"
-          echo "INCLUDE=$env:INCLUDE" >> $env:GITHUB_ENV
-          echo "LIB=$env:LIB" >> $env:GITHUB_ENV
-          echo "PATH=$env:PATH" >> $env:GITHUB_ENV
-
       - name: Install build dependencies
         shell: bash -l {0}
         run: |
@@ -177,10 +205,17 @@ jobs:
           MSSdk: "1"
           CIBW_BEFORE_BUILD: |
             pip install meson meson-python ninja numpy
+          CIBW_ENVIRONMENT: >
+            INCLUDE="C:/sleef/include;{project}/numpy_quaddtype/QBLAS/include;$INCLUDE"
+            LIB="C:/sleef/lib;$LIB"
+            PATH="C:/sleef/bin;$PATH"
+            CFLAGS="/IC:/sleef/include /I{project}/numpy_quaddtype/QBLAS/include /DDISABLE_QUADBLAS $CFLAGS"
+            CXXFLAGS="/IC:/sleef/include /I{project}/numpy_quaddtype/QBLAS/include /DDISABLE_QUADBLAS $CXXFLAGS"
+            LDFLAGS="C:/sleef/lib/sleef.lib C:/sleef/lib/sleefquad.lib $LDFLAGS"
           CIBW_REPAIR_WHEEL_COMMAND: 'delvewheel repair -w {dest_dir} {wheel} --add-path C:\sleef\bin'
           CIBW_TEST_COMMAND: |
             pip install {package}[test]
-            python -m pytest -v {project}/test
+            pytest -s {project}/tests
           CIBW_TEST_EXTRAS: test
           CIBW_TEST_FAIL_FAST: 1
         shell: pwsh
@@ -199,56 +234,21 @@ jobs:
     needs: [build_wheels_linux, build_wheels_macos, build_wheels_windows]
     runs-on: ubuntu-latest
     if: startsWith(github.ref, 'refs/tags/quaddtype-v')
-    
+
     environment:
       name: quadtype_release
       url: https://pypi.org/p/numpy-quaddtype
-    
+
     permissions:
-      id-token: write  # IMPORTANT: mandatory for trusted publishing
-    
+      id-token: write # IMPORTANT: mandatory for trusted publishing
+
     steps:
       - name: Download all workflow run artifacts
         uses: actions/download-artifact@v4
         with:
           path: dist
-      
+
       - name: Publish to PyPI
         uses: pypa/gh-action-pypi-publish@release/v1
         with:
           packages-dir: dist/*
-
-  # With the current setup, we are not creating a release on GitHub.
-  # create_release:
-  #   name: Create Release
-  #   needs: [build_wheels_linux, build_wheels_macos, build_wheels_windows]
-  #   runs-on: ubuntu-latest
-  #   if: startsWith(github.ref, 'refs/tags/quaddtype-v')
-
-  #   steps:
-  #     - name: Checkout code
-  #       uses: actions/checkout@v2
-
-  #     - name: Download all workflow run artifacts
-  #       uses: actions/download-artifact@v4
-  #       with:
-  #         path: artifacts
-
-  #     - name: Create Release
-  #       id: create_release
-  #       uses: actions/create-release@v1
-  #       env:
-  #         GITHUB_TOKEN: ${{ secrets.QUADDTYPE_GITHUB_TOKEN }}
-  #       with:
-  #         tag_name: ${{ github.ref }}
-  #         release_name: Release ${{ github.ref }}
-  #         draft: false
-  #         prerelease: false
-
-  #     - name: Upload Release Assets
-  #       uses: softprops/action-gh-release@v1
-  #       if: startsWith(github.ref, 'refs/tags/')
-  #       with:
-  #         files: ./artifacts/**/*.whl
-  #       env:
-  #         GITHUB_TOKEN: ${{ secrets.QUADDTYPE_GITHUB_TOKEN }}

.gitmodules

@@ -0,0 +1,3 @@
+[submodule "quaddtype/numpy_quaddtype/QBLAS"]
+	path = quaddtype/numpy_quaddtype/QBLAS
+	url = https://github.com/SwayamInSync/QBLAS

quaddtype/README.md

@@ -53,7 +53,10 @@ source temp/bin/activate
 # Install the package
 pip install meson-python numpy pytest
 
-export LDFLAGS="-Wl,-rpath,$SLEEF_DIR/lib"
+export LDFLAGS="-Wl,-rpath,$SLEEF_DIR/lib -fopenmp -latomic -lpthread"
+export CFLAGS="-fPIC"
+export CXXFLAGS="-fPIC"
+
 python -m pip install . -v --no-build-isolation -Cbuilddir=build -C'compile-args=-v'
 
 # Run the tests

quaddtype/meson.build

@@ -23,9 +23,17 @@ incdir_numpy = run_command(py,
   check : true
 ).stdout().strip()
 
+# Add OpenMP dependency (optional, for threading)
+openmp_dep = dependency('openmp', required: false)
+dependencies = [sleef_dep, py_dep]
+if openmp_dep.found()
+    dependencies += openmp_dep
+endif
+
 includes = include_directories(
     [
         incdir_numpy,
+        'numpy_quaddtype/QBLAS/include',
         'numpy_quaddtype/src',
     ]
 )
@@ -45,7 +53,9 @@ srcs = [
     'numpy_quaddtype/src/umath.h',
     'numpy_quaddtype/src/umath.cpp',
     'numpy_quaddtype/src/dragon4.h',
-    'numpy_quaddtype/src/dragon4.c'
+    'numpy_quaddtype/src/dragon4.c',
+    'numpy_quaddtype/src/quadblas_interface.h',
+    'numpy_quaddtype/src/quadblas_interface.cpp'
 ]
 
 py.install_sources(
@@ -60,7 +70,7 @@ py.extension_module('_quaddtype_main',
   srcs,
   link_args: is_windows ? ['/DEFAULTLIB:sleef', '/DEFAULTLIB:sleefquad'] : ['-lsleef', '-lsleefquad'],
   link_language: 'cpp',
-  dependencies: [sleef_dep, py_dep],
+  dependencies: dependencies,
   install: true,
   subdir: 'numpy_quaddtype',
   include_directories: includes

quaddtype/numpy_quaddtype/__init__.py

@@ -2,13 +2,22 @@
     QuadPrecision,
     QuadPrecDType,
     is_longdouble_128,
-    get_sleef_constant
+    get_sleef_constant,
+    qblas_dot as dot,
+    set_num_threads,
+    get_num_threads,
+    get_quadblas_version
 )
 
+import multiprocessing
+
 __all__ = [
     'QuadPrecision', 'QuadPrecDType', 'SleefQuadPrecision', 'LongDoubleQuadPrecision',
-    'SleefQuadPrecDType', 'LongDoubleQuadPrecDType', 'is_longdouble_128', 'pi', 'e', 
-    'log2e', 'log10e', 'ln2', 'ln10', 'max_value', 'min_value', 'epsilon'
+    'SleefQuadPrecDType', 'LongDoubleQuadPrecDType', 'is_longdouble_128', 
+    # Constants
+    'pi', 'e', 'log2e', 'log10e', 'ln2', 'ln10', 'max_value', 'min_value', 'epsilon', 
+    # QuadBLAS related functions
+    'dot', 'set_num_threads', 'get_num_threads', 'get_quadblas_version'
 ]
 
 def SleefQuadPrecision(value):
@@ -31,4 +40,8 @@ def LongDoubleQuadPrecDType():
 ln10 = get_sleef_constant("ln10")
 max_value = get_sleef_constant("quad_max")
 min_value = get_sleef_constant("quad_min")
-epsilon = get_sleef_constant("epsilon")
+epsilon = get_sleef_constant("epsilon")
+
+num_cores = multiprocessing.cpu_count()
+# set default number of threads for QuadBLAS
+set_num_threads(num_cores)