Merge 2998ad82a1 into 5b8cfa920b

.github/workflows/ccpp.yml

@@ -7,10 +7,10 @@ on:
     branches: [ master ]
 
 permissions:
-  contents: read # to fetch code (actions/checkout)
+  contents: write # to fetch code (actions/checkout),and release
 
 jobs:
-  job:
+  build:
     name: ${{ matrix.name }}-build-and-test
     runs-on: ${{ matrix.os }}
     strategy:
@@ -93,12 +93,12 @@ jobs:
     - name: Set Windows specific CMake arguments
       if: contains(matrix.name, 'windows')
       id: windows_extra_cmake_args
-      run: echo "::set-output name=args::-DASSIMP_BUILD_ASSIMP_TOOLS=1 -DASSIMP_BUILD_ASSIMP_VIEW=1 -DASSIMP_BUILD_ZLIB=1"
+      run: echo ":set-output name=args::=-DASSIMP_BUILD_ASSIMP_TOOLS=1 -DASSIMP_BUILD_ASSIMP_VIEW=1" >> $GITHUB_OUTPUT
     
     - name: Set Hunter specific CMake arguments
       if: contains(matrix.name, 'hunter')
       id: hunter_extra_cmake_args
-      run: echo "::set-output name=args::-DBUILD_SHARED_LIBS=OFF -DASSIMP_HUNTER_ENABLED=ON -DCMAKE_TOOLCHAIN_FILE=${{ github.workspace }}/cmake/polly/${{ matrix.toolchain }}.cmake"
+      run: echo "args=-DBUILD_SHARED_LIBS=OFF -DASSIMP_HUNTER_ENABLED=ON -DCMAKE_TOOLCHAIN_FILE=${{ github.workspace }}/cmake/polly/${{ matrix.toolchain }}.cmake" >> $GITHUB_OUTPUT
 
     - name: configure and build
       uses: lukka/run-cmake@v3
@@ -115,14 +115,98 @@ jobs:
     - name: Exclude certain tests in Hunter specific builds
       if: contains(matrix.name, 'hunter')
       id: hunter_extra_test_args
-      run: echo "::set-output name=args::--gtest_filter=-utOpenGEXImportExport.Importissue1340_EmptyCameraObject:utColladaZaeImportExport.importBlenFromFileTest"
+      run: echo "args=--gtest_filter=-utOpenGEXImportExport.Importissue1340_EmptyCameraObject:utColladaZaeImportExport.importBlenFromFileTest" >> $GITHUB_OUTPUT
 
     - name: test
       run: cd build/bin && ./unit ${{ steps.hunter_extra_test_args.outputs.args }}
       shell: bash
 
     - uses: actions/upload-artifact@v3
-      if: matrix.name == 'windows-msvc'
+      if: contains(matrix.name,'windows')
+      with:
+        name: 'assimp-bins-${{ matrix.name }}'
+        path: build/bin/assimp*.exe
+        
+    - uses: marvinpinto/action-automatic-releases@latest
+      if: contains(matrix.name, 'windows-msvc-hunter')
+      with:
+        repo_token: "${{ secrets.GITHUB_TOKEN }}"
+        automatic_release_tag: "master"
+        prerelease: true
+        title: "AutoRelease"
+        files: |
+          build/bin/assimp*.exe
+
+  create-release:
+    needs: [build]
+    runs-on: ubuntu-latest
+    if: startsWith(github.ref, 'refs/tags/')
+    steps:
+      - id: create-release
+        uses: actions/create-release@v1
+        env:
+          GITHUB_TOKEN: '${{secrets.GITHUB_TOKEN}}'
+        with:
+          tag_name: '${{github.ref}}'
+          release_name: 'Release ${{github.ref}}'
+          draft: false
+          prerelease: true
+      - run: |
+          echo '${{steps.create-release.outputs.upload_url}}' > release_upload_url.txt
+      - uses: actions/upload-artifact@v1
+        with:
+          name: create-release
+          path: release_upload_url.txt
+
+  upload-release:
+    strategy:
+      matrix:
+        name: [ubuntu-latest-g++, macos-latest-clang++, windows-latest-cl.exe, ubuntu-latest-clang++, ubuntu-gcc-hunter, macos-clang-hunter, windows-msvc-hunter]
+        # For Windows msvc, for Linux and macOS let's use the clang compiler, use gcc for Linux.
+        include:
+          - name: windows-latest-cl.exe
+            os: windows-latest
+            cxx: cl.exe
+            cc: cl.exe
+          - name: ubuntu-latest-clang++
+            os: ubuntu-latest
+            cxx: clang++
+            cc: clang
+          - name: macos-latest-clang++
+            os: macos-latest
+            cxx: clang++
+            cc: clang
+          - name: ubuntu-latest-g++
+            os: ubuntu-latest
+            cxx: g++
+            cc: gcc
+          - name: ubuntu-gcc-hunter
+            os: ubuntu-latest
+            toolchain: ninja-gcc-cxx17-fpic
+          - name: macos-clang-hunter
+            os: macos-latest
+            toolchain: ninja-clang-cxx17-fpic
+          - name: windows-msvc-hunter
+            os: windows-latest
+            toolchain: ninja-vs-win64-cxx17
+            
+    needs: [create-release]
+    runs-on: ubuntu-latest
+    if: startsWith(github.ref, 'refs/tags/')
+    steps:
+      - uses: softprops/action-gh-release@v1
+        with:
+          name: create-release
+      - id: upload-url
+        run: |
+          echo "url=$(cat create-release/release_upload_url.txt)" >> $GITHUB_OUTPUT
+      - uses: actions/download-artifact@v1
         with:
           name: 'assimp-bins-${{ matrix.name }}-${{ github.sha }}'
-        path: build/bin
+      - uses: actions/upload-release-asset@v1
+        env:
+          GITHUB_TOKEN: '${{secrets.GITHUB_TOKEN}}'
+        with:
+          files: |
+            *.zip
+ 

code/PostProcessing/ImproveCacheLocality.cpp

@@ -3,9 +3,7 @@
 Open Asset Import Library (assimp)
 ---------------------------------------------------------------------------
 
-Copyright (c) 2006-2022, assimp team
+Copyright (c) 2006-2023, assimp team
-
-
 
 All rights reserved.
 
@@ -59,12 +57,12 @@ OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 #include <stdio.h>
 #include <stack>
 
-using namespace Assimp;
+namespace Assimp {
 
 // ------------------------------------------------------------------------------------------------
 // Constructor to be privately used by Importer
-ImproveCacheLocalityProcess::ImproveCacheLocalityProcess()
+ImproveCacheLocalityProcess::ImproveCacheLocalityProcess() :
-: mConfigCacheDepth(PP_ICL_PTCACHE_SIZE) {
+        mConfigCacheDepth(PP_ICL_PTCACHE_SIZE) {
     // empty
 }
 
@@ -110,36 +108,13 @@ void ImproveCacheLocalityProcess::Execute( aiScene* pScene) {
 }
 
 // ------------------------------------------------------------------------------------------------
-// Improves the cache coherency of a specific mesh
+static ai_real calculateInputACMR(aiMesh *pMesh, const aiFace *const pcEnd,
-ai_real ImproveCacheLocalityProcess::ProcessMesh( aiMesh* pMesh, unsigned int meshNum) {
+        unsigned int configCacheDepth, unsigned int meshNum) {
-    // TODO: rewrite this to use std::vector or boost::shared_array
+    ai_real fACMR = 0.0f;
-    ai_assert(nullptr != pMesh);
+    unsigned int *piFIFOStack = new unsigned int[configCacheDepth];
-
+    memset(piFIFOStack, 0xff, configCacheDepth * sizeof(unsigned int));
-    // Check whether the input data is valid
-    // - there must be vertices and faces
-    // - all faces must be triangulated or we can't operate on them
-    if (!pMesh->HasFaces() || !pMesh->HasPositions())
-        return static_cast<ai_real>(0.f);
-
-    if (pMesh->mPrimitiveTypes != aiPrimitiveType_TRIANGLE) {
-        ASSIMP_LOG_ERROR("This algorithm works on triangle meshes only");
-        return static_cast<ai_real>(0.f);
-    }
-
-    if(pMesh->mNumVertices <= mConfigCacheDepth) {
-        return static_cast<ai_real>(0.f);
-    }
-
-    ai_real fACMR = 3.f;
-    const aiFace* const pcEnd = pMesh->mFaces+pMesh->mNumFaces;
-
-    // Input ACMR is for logging purposes only
-    if (!DefaultLogger::isNullLogger())     {
-
-        unsigned int* piFIFOStack = new unsigned int[mConfigCacheDepth];
-        memset(piFIFOStack,0xff,mConfigCacheDepth*sizeof(unsigned int));
     unsigned int *piCur = piFIFOStack;
-        const unsigned int* const piCurEnd = piFIFOStack + mConfigCacheDepth;
+    const unsigned int *const piCurEnd = piFIFOStack + configCacheDepth;
 
     // count the number of cache misses
     unsigned int iCacheMisses = 0;
@@ -174,21 +149,53 @@ ai_real ImproveCacheLocalityProcess::ProcessMesh( aiMesh* pMesh, unsigned int me
         ASSIMP_LOG_WARN(szBuff);
         return static_cast<ai_real>(0.f);
     }
+    return fACMR;
+}
+
+// ------------------------------------------------------------------------------------------------
+// Improves the cache coherency of a specific mesh
+ai_real ImproveCacheLocalityProcess::ProcessMesh(aiMesh *pMesh, unsigned int meshNum) {
+    // TODO: rewrite this to use std::vector or boost::shared_array
+    ai_assert(nullptr != pMesh);
+
+    // Check whether the input data is valid
+    // - there must be vertices and faces
+    // - all faces must be triangulated or we can't operate on them
+    if (!pMesh->HasFaces() || !pMesh->HasPositions())
+        return static_cast<ai_real>(0.f);
+
+    if (pMesh->mPrimitiveTypes != aiPrimitiveType_TRIANGLE) {
+        ASSIMP_LOG_ERROR("This algorithm works on triangle meshes only");
+        return static_cast<ai_real>(0.f);
+    }
+
+    if (pMesh->mNumVertices <= mConfigCacheDepth) {
+        return static_cast<ai_real>(0.f);
+    }
+
+    ai_real fACMR = 3.f;
+    const aiFace *const pcEnd = pMesh->mFaces + pMesh->mNumFaces;
+
+    // Input ACMR is for logging purposes only
+    if (!DefaultLogger::isNullLogger()) {
+        fACMR = calculateInputACMR(pMesh, pcEnd, mConfigCacheDepth, meshNum);
     }
 
     // first we need to build a vertex-triangle adjacency list
     VertexTriangleAdjacency adj(pMesh->mFaces, pMesh->mNumFaces, pMesh->mNumVertices, true);
 
     // build a list to store per-vertex caching time stamps
-    unsigned int* const piCachingStamps = new unsigned int[pMesh->mNumVertices];
+    std::vector<unsigned int> piCachingStamps;
-    memset(piCachingStamps,0x0,pMesh->mNumVertices*sizeof(unsigned int));
+    piCachingStamps.resize(pMesh->mNumVertices);
+    memset(&piCachingStamps[0], 0x0, pMesh->mNumVertices * sizeof(unsigned int));
 
     // allocate an empty output index buffer. We store the output indices in one large array.
     // Since the number of triangles won't change the input faces can be reused. This is how
     // we save thousands of redundant mini allocations for aiFace::mIndices
     const unsigned int iIdxCnt = pMesh->mNumFaces * 3;
-    unsigned int* const piIBOutput = new unsigned int[iIdxCnt];
+    std::vector<unsigned int> piIBOutput;
-    unsigned int* piCSIter = piIBOutput;
+    piIBOutput.resize(iIdxCnt);
+    std::vector<unsigned int>::iterator piCSIter = piIBOutput.begin();
 
     // allocate the flag array to hold the information
     // whether a face has already been emitted or not
@@ -202,7 +209,8 @@ ai_real ImproveCacheLocalityProcess::ProcessMesh( aiMesh* pMesh, unsigned int me
     const std::vector<unsigned int> piNumTriPtrNoModify(piNumTriPtr, piNumTriPtr + pMesh->mNumVertices);
 
     // get the largest number of referenced triangles and allocate the "candidate buffer"
-    unsigned int iMaxRefTris = 0; {
+    unsigned int iMaxRefTris = 0;
+    {
         const unsigned int *piCur = adj.mLiveTriangles;
         const unsigned int *const piCurEnd = adj.mLiveTriangles + pMesh->mNumVertices;
         for (; piCur != piCurEnd; ++piCur) {
@@ -210,7 +218,8 @@ ai_real ImproveCacheLocalityProcess::ProcessMesh( aiMesh* pMesh, unsigned int me
         }
     }
     ai_assert(iMaxRefTris > 0);
-    unsigned int* piCandidates = new unsigned int[iMaxRefTris*3];
+    std::vector<unsigned int> piCandidates;
+    piCandidates.resize(iMaxRefTris * 3);
     unsigned int iCacheMisses = 0;
 
     // ...................................................................................
@@ -250,7 +259,7 @@ ai_real ImproveCacheLocalityProcess::ProcessMesh( aiMesh* pMesh, unsigned int me
 
         unsigned int icnt = piNumTriPtrNoModify[ivdx];
         unsigned int *piList = adj.GetAdjacentTriangles(ivdx);
-        unsigned int* piCurCandidate = piCandidates;
+        std::vector<unsigned int>::iterator piCurCandidate = piCandidates.begin();
 
         // get all triangles in the neighborhood
         for (unsigned int tri = 0; tri < icnt; ++tri) {
@@ -261,7 +270,7 @@ ai_real ImproveCacheLocalityProcess::ProcessMesh( aiMesh* pMesh, unsigned int me
 
                 // so iterate through all vertices of the current triangle
                 const aiFace *pcFace = &pMesh->mFaces[fidx];
-                unsigned nind = pcFace->mNumIndices;
+                const unsigned nind = pcFace->mNumIndices;
                 for (unsigned ind = 0; ind < nind; ind++) {
                     unsigned dp = pcFace->mIndices[ind];
 
@@ -297,7 +306,7 @@ ai_real ImproveCacheLocalityProcess::ProcessMesh( aiMesh* pMesh, unsigned int me
         // get next fanning vertex
         ivdx = -1;
         int max_priority = -1;
-        for (unsigned int* piCur = piCandidates;piCur != piCurCandidate;++piCur)    {
+        for (std::vector<unsigned int>::iterator piCur = piCandidates.begin(); piCur != piCurCandidate; ++piCur) {
             const unsigned int dp = *piCur;
 
             // must have live triangles
@@ -345,29 +354,29 @@ ai_real ImproveCacheLocalityProcess::ProcessMesh( aiMesh* pMesh, unsigned int me
     }
     ai_real fACMR2 = 0.0f;
     if (!DefaultLogger::isNullLogger()) {
-        fACMR2 = (float)iCacheMisses / pMesh->mNumFaces;
+        fACMR2 = static_cast<ai_real>(iCacheMisses / pMesh->mNumFaces);
-
+        const ai_real averageACMR = ((fACMR - fACMR2) / fACMR) * 100.f;
         // very intense verbose logging ... prepare for much text if there are many meshes
         if (DefaultLogger::get()->getLogSeverity() == Logger::VERBOSE) {
-            ASSIMP_LOG_VERBOSE_DEBUG("Mesh %u | ACMR in: ", meshNum, " out: ", fACMR, " | ~", fACMR2, ((fACMR - fACMR2) / fACMR) * 100.f);
+            ASSIMP_LOG_VERBOSE_DEBUG("Mesh ", meshNum, "| ACMR in: ", fACMR, " out: ", fACMR2, " | average ACMR ", averageACMR);
         }
-
         fACMR2 *= pMesh->mNumFaces;
     }
+
     // sort the output index buffer back to the input array
-    piCSIter = piIBOutput;
+    piCSIter = piIBOutput.begin();
     for (aiFace *pcFace = pMesh->mFaces; pcFace != pcEnd; ++pcFace) {
         unsigned nind = pcFace->mNumIndices;
         unsigned *ind = pcFace->mIndices;
-        if (nind > 0) ind[0] = *piCSIter++;
+        if (nind > 0)
-        if (nind > 1) ind[1] = *piCSIter++;
+            ind[0] = *piCSIter++;
-        if (nind > 2) ind[2] = *piCSIter++;
+        if (nind > 1)
+            ind[1] = *piCSIter++;
+        if (nind > 2)
+            ind[2] = *piCSIter++;
     }
 
-    // delete temporary storage
-    delete[] piCachingStamps;
-    delete[] piIBOutput;
-    delete[] piCandidates;
-
     return fACMR2;
 }
+
+} // namespace Assimp