Merge pull request #150 from auto-differentiation-dev/feature/perform…

…ance-improvements Performance Enhancements
auto-differentiation · Nov 29, 2024 · bbee306 · bbee306
2 parents aaf5626 + 9570b12
commit bbee306
Show file tree

Hide file tree

Showing 32 changed files with 1,509 additions and 410 deletions.
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
@@ -12,13 +12,16 @@ jobs:
         compiler: [msvc, clang]
         config: [Release, Debug]
         standard: ["11", "17"]
+        memory: ["default", "lowmem"]
         include:
           - config: Debug
             sccache: "false"
             cmake_extra: ""
           - config: Release
             sccache: "true"
-            cmake_extra: -DCMAKE_CXX_COMPILER_LAUNCHER=sccache  
+            cmake_extra: -DCMAKE_CXX_COMPILER_LAUNCHER=sccache
+          - memory: "lowmem"
+            cmake_mem_flag: -DXAD_REDUCED_MEMORY=ON
           - compiler: clang
             cxx: C:/Program Files/Microsoft Visual Studio/2022/Enterprise/VC/Tools/Llvm/x64/bin/clang-cl.exe
             cc: C:/Program Files/Microsoft Visual Studio/2022/Enterprise/VC/Tools/Llvm/x64/bin/clang-cl.exe
@@ -47,6 +50,12 @@ jobs:
             toolset: "14.2"
           - compiler: clang
             toolset: "14.0"
+          - config: Release
+            memory: lowmem
+          - compiler: clang
+            memory: lowmem
+          - compiler: clang
+            config: Debug
     runs-on: ${{ matrix.os }}
     steps:
       - uses: actions/checkout@v3
@@ -57,7 +66,7 @@ jobs:
         if: ${{ matrix.sccache == 'true' }}
         uses: hendrikmuhs/[email protected]
         with:
-          key: windows-${{ matrix.compiler }}-${{ matrix.toolset }}-${{ matrix.config }}
+          key: windows-${{ matrix.compiler }}-${{ matrix.toolset }}-${{ matrix.config }}-${{ matrix.memory }}
           variant: sccache
       - name: configure
         shell: cmd
@@ -71,6 +80,7 @@ jobs:
              -DCMAKE_C_COMPILER="${{ matrix.cc }}" ^
              -DCMAKE_CXX_STANDARD="${{ matrix.standard }}" ^
              ${{ matrix.cmake_extra }} ^
+             ${{ matrix.cmake_mem_flag }} ^
              -DCMAKE_BUILD_TYPE=${{ matrix.config }} 
       - name: build
         shell: cmd
@@ -86,9 +96,9 @@ jobs:
           ctest --no-compress-output --output-on-failure -j3 --output-junit test_results.xml
       - name: upload test results
         if: always()
-        uses: actions/upload-artifact@v3
+        uses: actions/upload-artifact@v4
         with:
-          name: Test Results (windows ${{ matrix.compiler }}-${{ matrix.toolset }}-${{ matrix.config }})
+          name: Test Results (windows ${{ matrix.compiler }}-${{ matrix.toolset }}-${{ matrix.config }}-${{ matrix.memory }}-${{ matrix.standard }})
           path: build/test_results.xml
       - name: install
         shell: cmd
@@ -125,6 +135,7 @@ jobs:
         config: [Release, Debug]
         reuse_slots: [OFF, ON]
         standard: ["11", "17"]
+        memory: ["default", "lowmem"]
         include:
           - compiler: clang-11
             cxx: clang++-11
@@ -146,9 +157,21 @@ jobs:
             cxx: clang++-15
             cc: clang-15
             os: ubuntu-22.04
+          - memory: lowmem
+            cmake_mem_flag: -DXAD_REDUCED_MEMORY=ON
         exclude:
           - config: Debug
             reuse_slots: ON
+          - config: Release
+            memory: lowmem
+          - standard: "17"
+            memory: lowmem
+          - compiler: clang-11
+            config: Debug
+          - compiler: clang-12
+            config: Debug
+          - compiler: clang-13
+            config: Debug
     runs-on: ${{ matrix.os }}
     steps:
       - uses: actions/checkout@v3
@@ -163,7 +186,7 @@ jobs:
       - name: ccache
         uses: hendrikmuhs/[email protected]
         with:
-          key: linux-${{ matrix.compiler }}-${{ matrix.config }}-${{ matrix.reuse_slots }}
+          key: linux-${{ matrix.compiler }}-${{ matrix.config }}-${{ matrix.reuse_slots }}-${{ matrix.memory }}
       - name: configure
         run: |
           mkdir build
@@ -175,6 +198,7 @@ jobs:
             -DCMAKE_CXX_STANDARD=${{ matrix.standard }} \
             -DCMAKE_C_COMPILER=${{ matrix.cc }} \
             -DXAD_TAPE_REUSE_SLOTS=${{ matrix.reuse_slots }} \
+            ${{ matrix.cmake_mem_flag }} \
             -DCMAKE_INSTALL_PREFIX=${{ github.workspace }}/install
       - name: build
         run: |
@@ -186,9 +210,9 @@ jobs:
           ctest --no-compress-output --output-on-failure --parallel $(($(nproc) + 2)) --output-junit test_results.xml
       - name: upload test results
         if: always()
-        uses: actions/upload-artifact@v3
+        uses: actions/upload-artifact@v4
         with:
-          name: Test Results (linux ${{ matrix.compiler }}-${{ matrix.config }}-${{ matrix.reuse_slots }})
+          name: Test Results (linux ${{ matrix.compiler }}-${{ matrix.config }}-${{ matrix.reuse_slots }}-${{ matrix.memory }}-${{ matrix.standard }})
           path: build/test_results.xml
       - name: install
         run: |
@@ -212,34 +236,42 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        compiler: 
-          - gcc5
-          - gcc6 
-          - gcc7
-          - gcc8
-          - gcc9
-          - gcc10
-          - gcc11
+        version: ["7", "8", "9", "10", "11", "12", "13", "14"]
         config: [Release, Debug]
         reuse_slots: [OFF, ON]
         standard: [c++11, c++17]
+        memory: ["default", "lowmem"]
         exclude:
           - config: Debug
             reuse_slots: ON
-          - compiler: gcc5
-            standard: c++17
-          - compiler: gcc6
-            standard: c++17
-          - compiler: gcc7
+          - version: "7"
             standard: c++17
+          - config: Release
+            memory: lowmem
+          - standard: c++17
+            config: lowmem
+          - version: "7"
+            config: Debug
+          - version: "8"
+            config: Debug
+          - version: "9"
+            config: Debug
+          - version: "10"
+            config: Debug
+          - version: "12"
+            config: Debug
+          - version: "13"
+            config: Debug
         include:
+          - memory: lowmem
+            cmake_mem_flag: -DXAD_REDUCED_MEMORY=ON
           - config: Debug
-            compiler: gcc11
+            version: "11"
             coverage: true
             coverage_cxx_flags: "-fprofile-arcs -ftest-coverage" 
             coverage_ld_flags: "-lgcov"
             standard: c++11
-          - compiler: gcc11
+          - version: "11"
             config: Debug    # also with reusing slots, for coverage
             reuse_slots: ON
             coverage: true
@@ -248,48 +280,33 @@ jobs:
             standard: c++11
     runs-on: ubuntu-latest
     container: 
-      image: conanio/${{ matrix.compiler }}
-      options: --user root
-    env:
-      ACTIONS_RUNNER_FORCED_INTERNAL_NODE_VERSION: node16
-      ACTIONS_RUNNER_FORCE_ACTIONS_NODE_VERSION: node16
+      image: ghcr.io/foonathan/gcc:${{ matrix.version }}
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - name: Setup tools
         run: |
-          cat /etc/lsb-release
-          /lib/x86_64-linux-gnu/libc.so.6
-          sudo rm -f /etc/apt/sources.list.d/git-core-ubuntu-ppa-artful.list || true
-          sudo rm -f /etc/apt/sources.list.d/git-core-ubuntu-ppa-xenial.list || true
-          sudo apt-get update
-          sudo apt-get install -y ccache ninja-build lcov
+          apt-get update
+          apt-get install -y ccache ninja-build lcov
       - name: GitHub WS Tmp
         # there is an issue with workspace locations in github inside containers, which this works around
         # see: https://github.com/actions/runner/issues/2058
         run: |
            echo "GITHUB_WORKSPACE=$GITHUB_WORKSPACE" >> $GITHUB_ENV
-      - name: Setup cmake
-        run: |
-          wget https://github.com/Kitware/CMake/releases/download/v3.29.0/cmake-3.29.0-linux-x86_64.tar.gz
-          tar -xzf cmake-3.29.0-linux-x86_64.tar.gz 
-          sudo rm /usr/bin/cmake /usr/share/cmake-* -rf
-          echo $(pwd)/cmake-3.29.0-linux-x86_64/bin >> $GITHUB_PATH
       - name: ccache
         uses: hendrikmuhs/[email protected]
         with:
-          key: linux-${{ matrix.compiler }}-${{ matrix.config }}-${{ matrix.reuse_slots }}
+          key: linux-gcc${{ matrix.version }}-${{ matrix.config }}-${{ matrix.reuse_slots }}-${{ matrix.memory }}-${{ matrix.standard }}
       - name: configure
         run: |
           mkdir build
           cd build
           cmake .. -GNinja \
             -DCMAKE_CXX_COMPILER_LAUNCHER=ccache \
             -DCMAKE_BUILD_TYPE=${{ matrix.config }} \
-            -DCMAKE_CXX_COMPILER=${{ matrix.cxx }} \
-            -DCMAKE_C_COMPILER=${{ matrix.cc }} \
             -DCMAKE_CXX_FLAGS="-std=${{ matrix.standard }} ${{ matrix.coverage_cxx_flags }}" \
             -DCMAKE_EXE_LINKER_FLAGS="${{ matrix.coverage_ld_flags }}" \
             -DXAD_TAPE_REUSE_SLOTS=${{ matrix.reuse_slots }} \
+            ${{ matrix.cmake_mem_flag }} \
             -DCMAKE_INSTALL_PREFIX=${{ env.GITHUB_WORKSPACE }}/install
       - name: build
         run: |
@@ -307,9 +324,9 @@ jobs:
           ctest --no-compress-output --output-on-failure --parallel $(($(nproc) + 2)) --output-junit test_results.xml
       - name: upload test results
         if: always()
-        uses: actions/upload-artifact@v3
+        uses: actions/upload-artifact@v4
         with:
-          name: Test Results (linux ${{ matrix.compiler }}-${{ matrix.config }}-${{ matrix.reuse_slots }})
+          name: Test Results (linux gcc${{ matrix.version }}-${{ matrix.config }}-${{ matrix.reuse_slots }}-${{ matrix.memory }}-${{ matrix.standard }})
           path: ${{ env.GITHUB_WORKSPACE }}/build/test_results.xml
       - name: coverage tests
         if: ${{ matrix.coverage }}
@@ -323,7 +340,7 @@ jobs:
         uses: coverallsapp/github-action@master
         with:
           github-token: ${{ secrets.GITHUB_TOKEN }}
-          flag-name: linux-${{ matrix.compiler }}-${{ matrix.reuse_slots }}
+          flag-name: linux-${{ matrix.compiler }}-${{ matrix.reuse_slots }}-${{ matrix.memory }}
           path-to-lcov: ${{env.GITHUB_WORKSPACE }}/build/coverage/lcov.info
           parallel: true
       - name: install
@@ -340,8 +357,6 @@ jobs:
             -DCMAKE_CXX_COMPILER_LAUNCHER=ccache \
             -DCMAKE_PREFIX_PATH="${{ env.GITHUB_WORKSPACE }}/install"  \
             -DCMAKE_CXX_FLAGS="-std=${{ matrix.standard}}"  \
-            -DCMAKE_CXX_COMPILER=${{ matrix.cxx }} \
-            -DCMAKE_C_COMPILER=${{ matrix.cc }} \
             -DCMAKE_EXE_LINKER_FLAGS="${{ matrix.coverage_ld_flags }}"
           cmake --build .
 
@@ -354,7 +369,10 @@ jobs:
         os: [macos-13, macos-latest]
         compiler: [default, clang15]
         standard: [c++11, c++17]
+        memory: ["default", "lowmem"]
         include:
+          - memory: lowmem
+            cmake_mem_flag: -DXAD_REDUCED_MEMORY=ON
           - config: Release
             os: macos-13
             compiler: default
@@ -382,6 +400,10 @@ jobs:
             compiler: clang15
           - os: macos-latest
             compiler: default
+          - config: Release
+            memory: lowmem
+          - standard: "c++17"
+            config: lowmem
     runs-on: ${{ matrix.os }}
     steps:
       - name: setup
@@ -399,7 +421,7 @@ jobs:
       - name: ccache
         uses: hendrikmuhs/[email protected]
         with:
-          key: macos-${{ matrix.config }}
+          key: macos-${{ matrix.config }}-${{ matrix.compiler }}-${{ matrix.standard }}-${{ matrix.memory }}
       - name: configure
         run: |
           mkdir build
@@ -408,6 +430,7 @@ jobs:
             -DCMAKE_CXX_COMPILER_LAUNCHER=ccache \
             -DCMAKE_BUILD_TYPE=${{ matrix.config }} \
             -DCMAKE_CXX_FLAGS="-std=${{matrix.standard }} ${{ matrix.coverage_cxx_flags }}" \
+            ${{ matrix.cmake_mem_flag }} \
             -DCMAKE_EXE_LINKER_FLAGS="${{ matrix.coverage_ld_flags }}" \
             -DCMAKE_INSTALL_PREFIX=${{ github.workspace }}/install \
             ${{ matrix.compiler == 'clang15' && ' -DCMAKE_CXX_COMPILER=$(brew --prefix llvm@15)/bin/clang++ -DCMAKE_C_COMPILER=clang' || '' }}
@@ -427,9 +450,9 @@ jobs:
           ctest --no-compress-output --output-on-failure -j$(sysctl -n hw.ncpu) --output-junit test_results.xml
       - name: upload test results
         if: always()
-        uses: actions/upload-artifact@v3
+        uses: actions/upload-artifact@v4
         with:
-          name: Test Results (macos ${{ matrix.config }})
+          name: Test Results (macos ${{ matrix.config }}-${{ matrix.compiler }}-${{ matrix.standard }}-${{ matrix.memory }}-${{ matrix.standard }})
           path: build/test_results.xml
       - name: coverage tests
         if: ${{ matrix.coverage }}
@@ -443,7 +466,7 @@ jobs:
         uses: coverallsapp/github-action@master
         with:
           github-token: ${{ secrets.GITHUB_TOKEN }}
-          flag-name: macos
+          flag-name: macos-${{ matrix.compiler }}-${{ matrix.standard }}-${{ matrix.memory }}
           path-to-lcov: build/coverage/lcov.info
           parallel: true
       - name: install
@@ -483,7 +506,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Upload
-        uses: actions/upload-artifact@v3
+        uses: actions/upload-artifact@v4
         with:
           name: Event File
           path: ${{ github.event_path }}

diff --git a/cmake/XADSetupCompiler.cmake b/cmake/XADSetupCompiler.cmake
@@ -158,6 +158,11 @@ function(xad_add_library name type)
         set_target_properties(${name} PROPERTIES 
             MSVC_RUNTIME_LIBRARY "MultiThreaded$<$<CONFIG:Debug>:Debug>DLL")
     endif()
+    if(MSVC)
+        # in MSVC/Debug, we used checked iterators, and MSVC displays a deprecation warning
+        # without this flag
+        target_compile_definitions(${name} PUBLIC "$<$<CONFIG:DEBUG>:_SILENCE_STDEXT_ARR_ITERS_DEPRECATION_WARNING>")
+    endif()
 endfunction()
 
 

diff --git a/cmake/XADSetupOptions.cmake b/cmake/XADSetupOptions.cmake
@@ -63,10 +63,15 @@ endif()
 # Tape options: these end up in Config.hpp, a cmake-generated file
 option(XAD_TAPE_REUSE_SLOTS "Reuse slots in tape that have become free (slower, less memory)" OFF)
 option(XAD_NO_THREADLOCAL "Disable thread-local tape - only for single-threaded tape use" OFF)
-if (MSVC AND MSVC_VERSION GREATER_EQUAL 1920)
-    option(XAD_USE_STRONG_INLINE "Use forced inlining for higher preformance, at a higher compile time cost" OFF)
+if (MSVC AND MSVC_VERSION GREATER_EQUAL 1910)
+    option(XAD_USE_STRONG_INLINE "Use forced inlining for higher performance, at a higher compile time cost" OFF)
 else()
-    # in VS 2015 and 2017, without strong inlining, some long expressions in release mode get miscompiled
+    # in VS 2015, without strong inlining, some long expressions in release mode get miscompiled
     set(XAD_USE_STRONG_INLINE ON CACHE BOOL "Use forced inlining for higher preformance, at a higher compile time cost" FORCE)
 endif()
 option(XAD_ALLOW_INT_CONVERSION "Add real->int conversion operator, potentially missing to track dependencies" ON)
+option(XAD_REDUCED_MEMORY "Reduce memory required for tape, at a slight performance cost" OFF)
+
+if(XAD_REDUCED_MEMORY)
+    message(STATUS "Using reduced memory for tape storage at a slight performance cost")
+endif()
diff --git a/cmake/XADSetupTesting.cmake b/cmake/XADSetupTesting.cmake
@@ -38,18 +38,14 @@ if (NOT TARGET GTest::gmock_main)
     if(CMAKE_COMPILER_IS_GNUCXX AND CMAKE_CXX_COMPILER_VERSION VERSION_LESS 5.0)
         set(_gtest_tag release-1.10.0)
     else()
-        set(_gtest_tag release-1.11.0)
+        set(_gtest_tag v1.15.2)
     endif()
     set(INSTALL_GTEST OFF CACHE BOOL "" FORCE)
     FetchContent_Declare(googletest
                      GIT_REPOSITORY      https://github.com/google/googletest.git
                      GIT_TAG             ${_gtest_tag}
     )
-    FetchContent_GetProperties(googletest)
-    if(NOT googletest_POPULATED)
-        FetchContent_Populate(googletest)
-        add_subdirectory(${googletest_SOURCE_DIR} ${googletest_BINARY_DIR} EXCLUDE_FROM_ALL)
-    endif()
+    FetchContent_MakeAvailable(googletest)
 
     set_target_properties(gtest gtest_main gmock gmock_main PROPERTIES 
         FOLDER "test/gtest")

diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt
@@ -47,6 +47,8 @@ set(public_headers
     XAD/Literals.hpp
     XAD/Macros.hpp
     XAD/MathFunctions.hpp
+    XAD/OperationsContainer.hpp
+    XAD/OperationsContainerPaired.hpp
     XAD/ReusableRange.hpp
     XAD/StdCompatibility.hpp
     XAD/Tape.hpp