Merge branch 'master' into VK_EXT_pageable_device_local_memory

Tencent · Apr 1, 2024 · 70c6f49 · 70c6f49
2 parents 091c609 + 6872cca
commit 70c6f49
Show file tree

Hide file tree

Showing 703 changed files with 16,246 additions and 6,919 deletions.
diff --git a/.ci/pnnx.yml b/.ci/pnnx.yml
@@ -52,6 +52,10 @@ jobs:
             torchvision-version: 0.16.0
             torchvision-cache-key: '0_16_0'
 
+          - torch-version: 2.2.1
+            torchvision-version: 0.17.1
+            torchvision-cache-key: '0_17_1'
+
     runs-on:
       pool-name: docker
       container:
@@ -122,6 +126,7 @@ jobs:
     - name: test
       run: |
         export PYTHONUSERBASE=${{ci.workspace}}/torch-${{matrix.torch-version}}
+        export LD_LIBRARY_PATH=${{ci.workspace}}/torchvision-${{matrix.torchvision-version}}-install/lib
         export OMP_NUM_THREADS=1
         export MKL_NUM_THREADS=1
         export MKL_ENABLE_INSTRUCTIONS=SSE4_2
@@ -131,8 +136,9 @@ jobs:
     - name: python-pnnx
       run: |
         export PYTHONUSERBASE=${{ci.workspace}}/torch-${{matrix.torch-version}}
+        export LD_LIBRARY_PATH=${{ci.workspace}}/torchvision-${{matrix.torchvision-version}}-install/lib
         export PNNX_WHEEL_WITHOUT_BUILD=ON
-        cd tools/pnnx
-        cp build/src/pnnx python/pnnx/
+        cd tools/pnnx/python
+        cp ../build/src/pnnx pnnx/
         python3 setup.py install --user
-        pytest python/tests/
+        pytest tests
diff --git a/.github/workflows/android-armv7-cpu.yml b/.github/workflows/android-armv7-cpu.yml
@@ -33,12 +33,12 @@ jobs:
       run: |
         mkdir build && cd build
         cmake -DCMAKE_TOOLCHAIN_FILE=$ANDROID_NDK_LATEST_HOME/build/cmake/android.toolchain.cmake -DANDROID_ABI="armeabi-v7a" -DANDROID_ARM_NEON=ON -DANDROID_PLATFORM=android-14 ..
-        cmake --build . -j 2
+        cmake --build . -j $(nproc)
     - name: build-shared
       run: |
         mkdir build-shared && cd build-shared
         cmake -DCMAKE_TOOLCHAIN_FILE=$ANDROID_NDK_LATEST_HOME/build/cmake/android.toolchain.cmake -DANDROID_ABI="armeabi-v7a" -DANDROID_ARM_NEON=ON -DANDROID_PLATFORM=android-14 -DNCNN_SHARED_LIB=ON -DNCNN_ENABLE_LTO=OFF ..
-        cmake --build . -j 2
+        cmake --build . -j $(nproc)
 
     - name: ndk-r16b
       run: |
@@ -48,9 +48,9 @@ jobs:
       run: |
         mkdir build-noneon && cd build-noneon
         cmake -DCMAKE_TOOLCHAIN_FILE=$GITHUB_WORKSPACE/android-ndk-r16b/build/cmake/android.toolchain.cmake -DANDROID_ABI="armeabi-v7a" -DANDROID_ARM_NEON=OFF -DANDROID_PLATFORM=android-14 ..
-        cmake --build . -j 2
+        cmake --build . -j $(nproc)
     - name: build-noneon-shared
       run: |
         mkdir build-noneon-shared && cd build-noneon-shared
         cmake -DCMAKE_TOOLCHAIN_FILE=$GITHUB_WORKSPACE/android-ndk-r16b/build/cmake/android.toolchain.cmake -DANDROID_ABI="armeabi-v7a" -DANDROID_ARM_NEON=OFF -DANDROID_PLATFORM=android-14 -DNCNN_SHARED_LIB=ON -DNCNN_ENABLE_LTO=OFF ..
-        cmake --build . -j 2
+        cmake --build . -j $(nproc)
diff --git a/.github/workflows/android-armv7-gpu.yml b/.github/workflows/android-armv7-gpu.yml
@@ -37,9 +37,9 @@ jobs:
       run: |
         mkdir build && cd build
         cmake -DCMAKE_TOOLCHAIN_FILE=$ANDROID_NDK_LATEST_HOME/build/cmake/android.toolchain.cmake -DANDROID_ABI="armeabi-v7a" -DANDROID_ARM_NEON=ON -DANDROID_PLATFORM=android-21 -DNCNN_VULKAN=ON ..
-        cmake --build . -j 2
+        cmake --build . -j $(nproc)
     - name: build-shared
       run: |
         mkdir build-shared && cd build-shared
         cmake -DCMAKE_TOOLCHAIN_FILE=$ANDROID_NDK_LATEST_HOME/build/cmake/android.toolchain.cmake -DANDROID_ABI="armeabi-v7a" -DANDROID_ARM_NEON=ON -DANDROID_PLATFORM=android-21 -DNCNN_VULKAN=ON -DNCNN_SHARED_LIB=ON -DNCNN_ENABLE_LTO=OFF ..
-        cmake --build . -j 2
+        cmake --build . -j $(nproc)
diff --git a/.github/workflows/android-armv8-cpu.yml b/.github/workflows/android-armv8-cpu.yml
@@ -33,9 +33,9 @@ jobs:
       run: |
         mkdir build && cd build
         cmake -DCMAKE_TOOLCHAIN_FILE=$ANDROID_NDK_LATEST_HOME/build/cmake/android.toolchain.cmake -DANDROID_ABI="arm64-v8a" -DANDROID_PLATFORM=android-21 ..
-        cmake --build . -j 2
+        cmake --build . -j $(nproc)
     - name: build-shared
       run: |
         mkdir build-shared && cd build-shared
         cmake -DCMAKE_TOOLCHAIN_FILE=$ANDROID_NDK_LATEST_HOME/build/cmake/android.toolchain.cmake -DANDROID_ABI="arm64-v8a" -DANDROID_PLATFORM=android-21 -DNCNN_SHARED_LIB=ON ..
-        cmake --build . -j 2
+        cmake --build . -j $(nproc)
diff --git a/.github/workflows/android-armv8-gpu.yml b/.github/workflows/android-armv8-gpu.yml
@@ -37,27 +37,27 @@ jobs:
       run: |
         mkdir build && cd build
         cmake -DCMAKE_TOOLCHAIN_FILE=$ANDROID_NDK_LATEST_HOME/build/cmake/android.toolchain.cmake -DANDROID_ABI="arm64-v8a" -DANDROID_PLATFORM=android-21 -DNCNN_VULKAN=ON ..
-        cmake --build . -j 2
+        cmake --build . -j $(nproc)
     - name: build-shared
       run: |
         mkdir build-shared && cd build-shared
         cmake -DCMAKE_TOOLCHAIN_FILE=$ANDROID_NDK_LATEST_HOME/build/cmake/android.toolchain.cmake -DANDROID_ABI="arm64-v8a" -DANDROID_PLATFORM=android-21 -DNCNN_VULKAN=ON -DNCNN_SHARED_LIB=ON ..
-        cmake --build . -j 2
+        cmake --build . -j $(nproc)
     - name: build-termux
       run: |
         mkdir build-termux && cd build-termux
         cmake -DCMAKE_TOOLCHAIN_FILE=$ANDROID_NDK_LATEST_HOME/build/cmake/android.toolchain.cmake -DANDROID_ABI="arm64-v8a" -DANDROID_PLATFORM=android-21 -DNCNN_VULKAN=ON -DNCNN_PLATFORM_API=OFF ..
-        cmake --build . -j 2
+        cmake --build . -j $(nproc)
     - name: build-android-29
       run: |
         mkdir build-android-29 && cd build-android-29
         cmake -DCMAKE_TOOLCHAIN_FILE=$ANDROID_NDK_LATEST_HOME/build/cmake/android.toolchain.cmake -DANDROID_ABI="arm64-v8a" -DANDROID_PLATFORM=android-29 -DNCNN_VULKAN=ON ..
-        cmake --build . -j 2
+        cmake --build . -j $(nproc)
     - name: build-android-29-shared
       run: |
         mkdir build-android-29-shared && cd build-android-29-shared
         cmake -DCMAKE_TOOLCHAIN_FILE=$ANDROID_NDK_LATEST_HOME/build/cmake/android.toolchain.cmake -DANDROID_ABI="arm64-v8a" -DANDROID_PLATFORM=android-29 -DNCNN_VULKAN=ON -DNCNN_SHARED_LIB=ON ..
-        cmake --build . -j 2
+        cmake --build . -j $(nproc)
 
   android-aarch64-gpu-ndk-r16b:
     runs-on: ubuntu-20.04
@@ -72,4 +72,4 @@ jobs:
     - name: configure
       run: mkdir build && cd build && cmake -DCMAKE_TOOLCHAIN_FILE=$GITHUB_WORKSPACE/android-ndk-r16b/build/cmake/android.toolchain.cmake -DANDROID_ABI="arm64-v8a" -DANDROID_PLATFORM=android-21 -DNCNN_VULKAN=ON ..
     - name: build
-      run: cmake --build build -j 2
+      run: cmake --build build -j $(nproc)
diff --git a/.github/workflows/android-x64-cpu.yml b/.github/workflows/android-x64-cpu.yml
@@ -33,9 +33,9 @@ jobs:
       run: |
         mkdir build && cd build
         cmake -DCMAKE_TOOLCHAIN_FILE=$ANDROID_NDK_LATEST_HOME/build/cmake/android.toolchain.cmake -DANDROID_ABI="x86_64" -DANDROID_PLATFORM=android-21 ..
-        cmake --build . -j 2
+        cmake --build . -j $(nproc)
     - name: build-shared
       run: |
         mkdir build-shared && cd build-shared
         cmake -DCMAKE_TOOLCHAIN_FILE=$ANDROID_NDK_LATEST_HOME/build/cmake/android.toolchain.cmake -DANDROID_ABI="x86_64" -DANDROID_PLATFORM=android-21 -DNCNN_SHARED_LIB=ON ..
-        cmake --build . -j 2
+        cmake --build . -j $(nproc)
diff --git a/.github/workflows/android-x64-gpu.yml b/.github/workflows/android-x64-gpu.yml
@@ -37,9 +37,9 @@ jobs:
       run: |
         mkdir build && cd build
         cmake -DCMAKE_TOOLCHAIN_FILE=$ANDROID_NDK_LATEST_HOME/build/cmake/android.toolchain.cmake -DANDROID_ABI="x86_64" -DANDROID_PLATFORM=android-21 -DNCNN_VULKAN=ON ..
-        cmake --build . -j 2
+        cmake --build . -j $(nproc)
     - name: build-shared
       run: |
         mkdir build-shared && cd build-shared
         cmake -DCMAKE_TOOLCHAIN_FILE=$ANDROID_NDK_LATEST_HOME/build/cmake/android.toolchain.cmake -DANDROID_ABI="x86_64" -DANDROID_PLATFORM=android-21 -DNCNN_VULKAN=ON -DNCNN_SHARED_LIB=ON ..
-        cmake --build . -j 2
+        cmake --build . -j $(nproc)
diff --git a/.github/workflows/android-x86-cpu.yml b/.github/workflows/android-x86-cpu.yml
@@ -33,9 +33,9 @@ jobs:
       run: |
         mkdir build && cd build
         cmake -DCMAKE_TOOLCHAIN_FILE=$ANDROID_NDK_LATEST_HOME/build/cmake/android.toolchain.cmake -DANDROID_ABI="x86"  -DANDROID_PLATFORM=android-14 ..
-        cmake --build . -j 2
+        cmake --build . -j $(nproc)
     - name: build-shared
       run: |
         mkdir build-shared && cd build-shared
         cmake -DCMAKE_TOOLCHAIN_FILE=$ANDROID_NDK_LATEST_HOME/build/cmake/android.toolchain.cmake -DANDROID_ABI="x86"  -DANDROID_PLATFORM=android-14 -DNCNN_SHARED_LIB=ON ..
-        cmake --build . -j 2
+        cmake --build . -j $(nproc)
diff --git a/.github/workflows/android-x86-gpu.yml b/.github/workflows/android-x86-gpu.yml
@@ -37,9 +37,9 @@ jobs:
       run: |
         mkdir build && cd build
         cmake -DCMAKE_TOOLCHAIN_FILE=$ANDROID_NDK_LATEST_HOME/build/cmake/android.toolchain.cmake -DANDROID_ABI="x86" -DANDROID_PLATFORM=android-21 -DNCNN_VULKAN=ON ..
-        cmake --build . -j 2
+        cmake --build . -j $(nproc)
     - name: build-shared
       run: |
         mkdir build-shared && cd build-shared
         cmake -DCMAKE_TOOLCHAIN_FILE=$ANDROID_NDK_LATEST_HOME/build/cmake/android.toolchain.cmake -DANDROID_ABI="x86" -DANDROID_PLATFORM=android-21 -DNCNN_VULKAN=ON -DNCNN_SHARED_LIB=ON ..
-        cmake --build . -j 2
+        cmake --build . -j $(nproc)
diff --git a/.github/workflows/code-format.yml b/.github/workflows/code-format.yml
@@ -19,7 +19,7 @@ jobs:
 
     - name: cache-clang-format
       id: cache-clang-format
-      uses: actions/cache@v3
+      uses: actions/cache@v4
       with:
         path: clang-format-install
         key: clang-format-install-4

diff --git a/.github/workflows/ios-arm64-gpu.yml b/.github/workflows/ios-arm64-gpu.yml
@@ -44,7 +44,7 @@ jobs:
 
     - name: cache-openmp
       id: cache-openmp
-      uses: actions/cache@v3
+      uses: actions/cache@v4
       with:
         path: openmp-install
         key: openmp-ios-install-20230504

diff --git a/.github/workflows/ios-cpu.yml b/.github/workflows/ios-cpu.yml
@@ -40,7 +40,7 @@ jobs:
 
     - name: cache-openmp
       id: cache-openmp
-      uses: actions/cache@v3
+      uses: actions/cache@v4
       with:
         path: openmp-install
         key: openmp-ios-install-20230504

diff --git a/.github/workflows/ios-simulator-gpu.yml b/.github/workflows/ios-simulator-gpu.yml
@@ -44,7 +44,7 @@ jobs:
 
     - name: cache-openmp
       id: cache-openmp
-      uses: actions/cache@v3
+      uses: actions/cache@v4
       with:
         path: openmp-install
         key: openmp-ios-simulator-install-20230504

diff --git a/.github/workflows/ios-simulator.yml b/.github/workflows/ios-simulator.yml
@@ -42,7 +42,7 @@ jobs:
 
     - name: cache-openmp
       id: cache-openmp
-      uses: actions/cache@v3
+      uses: actions/cache@v4
       with:
         path: openmp-install
         key: openmp-ios-simulator-install-20230504

diff --git a/.github/workflows/linux-aarch64-cpu-gcc.yml b/.github/workflows/linux-aarch64-cpu-gcc.yml
@@ -36,7 +36,7 @@ jobs:
 
     - name: cache-qemu
       id: cache-qemu
-      uses: actions/cache@v3
+      uses: actions/cache@v4
       with:
         path: qemu-install
         key: qemu-aarch64-install-20220502-ubuntu-2004-2
@@ -57,7 +57,7 @@ jobs:
       run: |
         cd qemu
         ./configure --prefix=$GITHUB_WORKSPACE/qemu-install --target-list=aarch64-linux-user --disable-system
-        make -j2
+        make -j$(nproc)
         make install
 
     - name: aarch64-gnu-toolchain
@@ -69,34 +69,34 @@ jobs:
       run: |
         mkdir build && cd build
         cmake -DCMAKE_TOOLCHAIN_FILE=../toolchains/aarch64-linux-gnu.toolchain.cmake -DNCNN_ARM82=OFF -DNCNN_ARM82DOT=OFF -DNCNN_BUILD_TOOLS=OFF -DNCNN_BUILD_EXAMPLES=OFF -DNCNN_BUILD_TESTS=ON ..
-        cmake --build . -j 2
+        cmake --build . -j $(nproc)
     - name: test
       run: |
         export PATH=$GITHUB_WORKSPACE/qemu-install/bin:$PATH
         cd build
-        TESTS_EXECUTABLE_LOADER=qemu-aarch64 TESTS_EXECUTABLE_LOADER_ARGUMENTS="-L;/usr/aarch64-linux-gnu" ctest --output-on-failure -j 2
+        TESTS_EXECUTABLE_LOADER=qemu-aarch64 TESTS_EXECUTABLE_LOADER_ARGUMENTS="-L;/usr/aarch64-linux-gnu" ctest --output-on-failure -j $(nproc)
 
     - name: build-noint8
       run: |
         mkdir build-noint8 && cd build-noint8
         cmake -DCMAKE_TOOLCHAIN_FILE=../toolchains/aarch64-linux-gnu.toolchain.cmake -DNCNN_ARM82=OFF -DNCNN_ARM82DOT=OFF -DNCNN_BUILD_TOOLS=OFF -DNCNN_BUILD_EXAMPLES=OFF -DNCNN_BUILD_TESTS=ON ..
-        cmake --build . -j 2
+        cmake --build . -j $(nproc)
     - name: test-noint8
       run: |
         export PATH=$GITHUB_WORKSPACE/qemu-install/bin:$PATH
         cd build-noint8
-        TESTS_EXECUTABLE_LOADER=qemu-aarch64 TESTS_EXECUTABLE_LOADER_ARGUMENTS="-L;/usr/aarch64-linux-gnu" ctest --output-on-failure -j 2
+        TESTS_EXECUTABLE_LOADER=qemu-aarch64 TESTS_EXECUTABLE_LOADER_ARGUMENTS="-L;/usr/aarch64-linux-gnu" ctest --output-on-failure -j $(nproc)
     
     - name: build-simplestl-simplemath
       run: |
         mkdir build-simplestl-simplemath && cd build-simplestl-simplemath 
         cmake -DCMAKE_TOOLCHAIN_FILE=../toolchains/aarch64-linux-gnu-c.toolchain.cmake -DNCNN_STDIO=ON -DNCNN_STRING=ON -DNCNN_SIMPLESTL=ON -DNCNN_SIMPLEMATH=ON -DNCNN_BUILD_TESTS=ON -DNCNN_BUILD_BENCHMARK=OFF -DNCNN_BUILD_TOOLS=OFF -DNCNN_BUILD_EXAMPLES=OFF ..
-        cmake --build . -j 2
+        cmake --build . -j $(nproc)
     - name: test-simplestl-simplemath
       run: |
         export PATH=$GITHUB_WORKSPACE/qemu-install/bin:$PATH
         cd build-simplestl-simplemath
-        TESTS_EXECUTABLE_LOADER=qemu-aarch64 TESTS_EXECUTABLE_LOADER_ARGUMENTS="-L;/usr/aarch64-linux-gnu" ctest --output-on-failure -j 2
+        TESTS_EXECUTABLE_LOADER=qemu-aarch64 TESTS_EXECUTABLE_LOADER_ARGUMENTS="-L;/usr/aarch64-linux-gnu" ctest --output-on-failure -j $(nproc)
 
   linux-gcc-arm82:
     runs-on: ubuntu-20.04
@@ -105,7 +105,7 @@ jobs:
 
     - name: cache-qemu
       id: cache-qemu
-      uses: actions/cache@v3
+      uses: actions/cache@v4
       with:
         path: qemu-install
         key: qemu-aarch64-install-20220502-ubuntu-2004-2
@@ -126,7 +126,7 @@ jobs:
       run: |
         cd qemu
         ./configure --prefix=$GITHUB_WORKSPACE/qemu-install --target-list=aarch64-linux-user --disable-system
-        make -j2
+        make -j$(nproc)
         make install
 
     - name: aarch64-gnu-toolchain
@@ -138,23 +138,23 @@ jobs:
       run: |
         mkdir build && cd build
         cmake -DCMAKE_TOOLCHAIN_FILE=../toolchains/aarch64-linux-gnu.toolchain.cmake -DNCNN_ARM82=ON -DNCNN_ARM82DOT=ON -DNCNN_BUILD_TOOLS=OFF -DNCNN_BUILD_EXAMPLES=OFF -DNCNN_BUILD_TESTS=ON ..
-        cmake --build . -j 2
+        cmake --build . -j $(nproc)
     - name: test
       run: |
         export PATH=$GITHUB_WORKSPACE/qemu-install/bin:$PATH
         cd build
-        TESTS_EXECUTABLE_LOADER=qemu-aarch64 TESTS_EXECUTABLE_LOADER_ARGUMENTS="-L;/usr/aarch64-linux-gnu" ctest --output-on-failure -j 2
+        TESTS_EXECUTABLE_LOADER=qemu-aarch64 TESTS_EXECUTABLE_LOADER_ARGUMENTS="-L;/usr/aarch64-linux-gnu" ctest --output-on-failure -j $(nproc)
 
     - name: build-noint8
       run: |
         mkdir build-noint8 && cd build-noint8
         cmake -DCMAKE_TOOLCHAIN_FILE=../toolchains/aarch64-linux-gnu.toolchain.cmake -DNCNN_ARM82=ON -DNCNN_ARM82DOT=ON -DNCNN_BUILD_TOOLS=OFF -DNCNN_BUILD_EXAMPLES=OFF -DNCNN_BUILD_TESTS=ON -DNCNN_INT8=OFF ..
-        cmake --build . -j 2
+        cmake --build . -j $(nproc)
     - name: test-noint8
       run: |
         export PATH=$GITHUB_WORKSPACE/qemu-install/bin:$PATH
         cd build-noint8
-        TESTS_EXECUTABLE_LOADER=qemu-aarch64 TESTS_EXECUTABLE_LOADER_ARGUMENTS="-L;/usr/aarch64-linux-gnu" ctest --output-on-failure -j 2
+        TESTS_EXECUTABLE_LOADER=qemu-aarch64 TESTS_EXECUTABLE_LOADER_ARGUMENTS="-L;/usr/aarch64-linux-gnu" ctest --output-on-failure -j $(nproc)
 
   linux-gcc-arm86:
     runs-on: ubuntu-22.04
@@ -163,7 +163,7 @@ jobs:
 
     - name: cache-qemu
       id: cache-qemu
-      uses: actions/cache@v3
+      uses: actions/cache@v4
       with:
         path: qemu-install
         key: qemu-aarch64-install-20230717
@@ -184,7 +184,7 @@ jobs:
       run: |
         cd qemu
         ./configure --prefix=$GITHUB_WORKSPACE/qemu-install --target-list=aarch64-linux-user --disable-system
-        make -j2
+        make -j$(nproc)
         make install
 
     - name: aarch64-gnu-toolchain
@@ -196,9 +196,9 @@ jobs:
       run: |
         mkdir build && cd build
         cmake -DCMAKE_TOOLCHAIN_FILE=../toolchains/aarch64-linux-gnu.toolchain.cmake -DNCNN_ARM82=ON -DNCNN_ARM82DOT=ON -DNCNN_BUILD_TOOLS=OFF -DNCNN_BUILD_EXAMPLES=OFF -DNCNN_BUILD_TESTS=ON ..
-        cmake --build . -j 2
+        cmake --build . -j $(nproc)
     - name: test
       run: |
         export PATH=$GITHUB_WORKSPACE/qemu-install/bin:$PATH
         cd build
-        TESTS_EXECUTABLE_LOADER=qemu-aarch64 TESTS_EXECUTABLE_LOADER_ARGUMENTS="-L;/usr/aarch64-linux-gnu" ctest --output-on-failure -j 2
+        TESTS_EXECUTABLE_LOADER=qemu-aarch64 TESTS_EXECUTABLE_LOADER_ARGUMENTS="-L;/usr/aarch64-linux-gnu" ctest --output-on-failure -j $(nproc)