port rvv intrinsic 1.0+

.ci/test-coverage.yml

@@ -840,149 +840,6 @@ jobs:
     - name: codecov
       run: ./codecov -t ${{settings.CODECOV_TOKEN.access_token}} -C ${{ ci.sha }} -B ${{ ci.head_ref }} -f build/lcov.info
 
-  linux-gcc-riscv64-rvv:
-    name: linux-gcc-riscv64-rvv
-    strategy:
-      matrix:
-        OPENMP: ['OFF', 'ON']
-
-    runs-on:
-      pool-name: docker
-      container:
-        image: bkci/ci:ubuntu
-    steps:
-    - name: checkout
-      checkout: self
-      with:
-        strategy: FRESH_CHECKOUT
-        enableSubmodule: false
-        enableGitLfs: false
-
-    - name: install-deps
-      run: |
-        apt-get update
-        apt-get install -y lcov libcapstone4 libglib2.0-0
-        curl https://uploader.codecov.io/verification.gpg | gpg --no-default-keyring --keyring trustedkeys.gpg --import
-        curl -Os https://uploader.codecov.io/latest/linux/codecov
-        curl -Os https://uploader.codecov.io/latest/linux/codecov.SHA256SUM
-        curl -Os https://uploader.codecov.io/latest/linux/codecov.SHA256SUM.sig
-        gpgv codecov.SHA256SUM.sig codecov.SHA256SUM
-        shasum -a 256 -c codecov.SHA256SUM
-        chmod +x codecov
-
-    - name: cache-qemu
-      id: cache-qemu
-      uses: cache@1.*
-      with:
-        cachePaths: qemu-install
-        cacheKey: qemu-riscv64-install-20230624-1
-
-    - name: checkout-qemu
-      if: steps.cache-qemu.outputs.cacheHit != 'true'
-      checkout: https://github.com/qemu/qemu.git
-      with:
-        pullType: COMMIT_ID
-        refName: b455ce4c2f300c8ba47cba7232dd03261368a4cb
-        localPath: qemu
-        enableSubmodule: false
-        enableGitLfs: false
-
-    - name: qemu
-      if: steps.cache-qemu.outputs.cacheHit != 'true'
-      run: |
-        echo 'deb-src http://mirrors.cloud.tencent.com/debian bullseye main' | tee -a /etc/apt/sources.list
-        echo 'deb-src http://mirrors.cloud.tencent.com/debian bullseye-updates main' | tee -a /etc/apt/sources.list
-        apt-get update
-        apt-get build-dep -y qemu
-        apt-get install -y python3-pip
-        python3 -m pip install --upgrade pip
-        apt-get remove -y python3-setuptools
-        pip3 install -U setuptools
-        cd qemu
-        wget https://raw.githubusercontent.com/nihui/ncnn-assets/master/qemu-patches/0007-linux-user-Expose-risc-v-V-isa-bit-in-get_elf_hwcap.patch
-        patch -p1 -i 0007-linux-user-Expose-risc-v-V-isa-bit-in-get_elf_hwcap.patch
-        ./configure --prefix=${{ci.workspace}}/qemu-install --target-list=riscv64-linux-user --disable-system
-        make -j$(nproc)
-        make install
-
-    - name: cache-rv64gcv
-      id: cache-rv64gcv
-      uses: cache@1.*
-      with:
-        cachePaths: rv64gcv-install
-        cacheKey: rv64gcv-linux-install-20221029-1
-
-    - name: checkout-riscv-gnu-toolchain
-      if: steps.cache-rv64gcv.outputs.cacheHit != 'true'
-      checkout: https://github.com/riscv/riscv-gnu-toolchain.git
-      with:
-        pullType: COMMIT_ID
-        refName: da01ba455ce3802ffa84fdca3a089079996dbfc3
-        localPath: riscv-gnu-toolchain
-        enableSubmodule: false
-        enableGitLfs: false
-
-    - name: riscv-gnu-toolchain
-      if: steps.cache-rv64gcv.outputs.cacheHit != 'true'
-      run: |
-        apt-get update
-        apt-get install -y autoconf automake autotools-dev curl python3 libmpc-dev libmpfr-dev libgmp-dev gawk build-essential bison flex texinfo gperf libtool patchutils bc zlib1g-dev libexpat-dev device-tree-compiler
-        cd riscv-gnu-toolchain
-        git submodule update --init --recursive --depth 1 glibc
-        git submodule update --init --recursive --depth 1 newlib
-        #git submodule update --init --recursive --depth 1 riscv-binutils
-        #git submodule update --init --recursive --depth 1 riscv-gcc
-        git submodule update --init --recursive --depth 1 riscv-dejagnu
-        git submodule update --init --recursive --depth 1 riscv-gdb
-        rm -rf riscv-binutils
-        git clone -b binutils-2_39-branch https://sourceware.org/git/binutils-gdb.git riscv-binutils
-        rm -rf riscv-gcc
-        git clone -b riscv-gcc-rvv-next https://github.com/riscv-collab/riscv-gcc.git riscv-gcc
-        cd riscv-gcc
-        git checkout 8a0c1b106f01c455a8fb478cfe52d859a69020fd
-        cd ..
-        sed -i '/__OBSOLETE_MATH/d' newlib/newlib/libm/common/math_errf.c
-        ./configure --prefix=${{ci.workspace}}/rv64gcv-install --with-arch=rv64gcv_zfh
-        make linux -j$(nproc)
-        find ${{ci.workspace}}/rv64gcv-install -type f | xargs -i strip -g {} || true
-
-    - name: build
-      run: |
-        export RISCV_ROOT_PATH=${{ci.workspace}}/rv64gcv-install
-        mkdir build && cd build
-        cmake -DCMAKE_TOOLCHAIN_FILE=../toolchains/riscv64-unknown-linux-gnu.toolchain.cmake -DCMAKE_BUILD_TYPE=debug -DCMAKE_C_FLAGS="-O1" -DCMAKE_CXX_FLAGS="-O1" -DNCNN_COVERAGE=ON -DNCNN_RUNTIME_CPU=OFF -DNCNN_RVV=ON -DNCNN_OPENMP=${{matrix.OPENMP}} -DNCNN_BUILD_TOOLS=OFF -DNCNN_BUILD_EXAMPLES=OFF -DNCNN_BUILD_TESTS=ON ..
-        cmake --build . -j 4
-    - name: test-vlen128
-      run: |
-        export PATH=${{ci.workspace}}/qemu-install/bin:$PATH
-        cd build
-        TESTS_EXECUTABLE_LOADER=qemu-riscv64 TESTS_EXECUTABLE_LOADER_ARGUMENTS="-cpu;rv64,v=true,Zfh=true,x-zvfh=true,vlen=128,elen=64,vext_spec=v1.0;-L;${{ci.workspace}}/rv64gcv-install/sysroot" ctest --output-on-failure -j 4
-    - name: lcov-collect-vlen128
-      run: |
-        cd build
-        lcov --gcov-tool ${{ci.workspace}}/rv64gcv-install/bin/riscv64-unknown-linux-gnu-gcov -d ./src -c -o lcov.info
-        lcov -r lcov.info '/usr/*' -o lcov.info
-        lcov -r lcov.info '*/build/*' -o lcov.info
-        lcov -r lcov.info '*/rv64gcv-install/*' -o lcov.info
-        lcov --list lcov.info
-    - name: codecov-vlen128
-      run: ./codecov -t ${{settings.CODECOV_TOKEN.access_token}} -C ${{ ci.sha }} -B ${{ ci.head_ref }} -f build/lcov.info
-    - name: test-vlen256
-      run: |
-        export PATH=${{ci.workspace}}/qemu-install/bin:$PATH
-        cd build
-        TESTS_EXECUTABLE_LOADER=qemu-riscv64 TESTS_EXECUTABLE_LOADER_ARGUMENTS="-cpu;rv64,v=true,Zfh=true,x-zvfh=true,vlen=256,elen=64,vext_spec=v1.0;-L;${{ci.workspace}}/rv64gcv-install/sysroot" ctest --output-on-failure -j 4
-    - name: lcov-collect-vlen256
-      run: |
-        cd build
-        lcov --gcov-tool ${{ci.workspace}}/rv64gcv-install/bin/riscv64-unknown-linux-gnu-gcov -d ./src -c -o lcov.info
-        lcov -r lcov.info '/usr/*' -o lcov.info
-        lcov -r lcov.info '*/build/*' -o lcov.info
-        lcov -r lcov.info '*/rv64gcv-install/*' -o lcov.info
-        lcov --list lcov.info
-    - name: codecov-vlen256
-      run: ./codecov -t ${{settings.CODECOV_TOKEN.access_token}} -C ${{ ci.sha }} -B ${{ ci.head_ref }} -f build/lcov.info
-
   linux-gcc-loongarch64:
     name: linux-gcc-loongarch64
     strategy:

.github/workflows/android.yml

@@ -9,6 +9,7 @@ on:
     - 'src/*'
     - 'src/layer/*'
     - 'src/layer/arm/**'
+    - 'src/layer/riscv/**'
     - 'src/layer/x86/**'
     - 'src/layer/vulkan/**'
   pull_request:
@@ -20,6 +21,7 @@ on:
     - 'src/*'
     - 'src/layer/*'
     - 'src/layer/arm/**'
+    - 'src/layer/riscv/**'
     - 'src/layer/x86/**'
     - 'src/layer/vulkan/**'
 concurrency:
@@ -64,6 +66,11 @@ jobs:
         mkdir build-x86_64 && cd build-x86_64
         cmake ${{ env.NCNN_CMAKE_OPTIONS }} -DANDROID_ABI="x86_64" ..
         cmake --build . -j $(nproc)
+    - name: riscv64
+      run: |
+        mkdir build-riscv64 && cd build-riscv64
+        cmake ${{ env.NCNN_CMAKE_OPTIONS }} -DANDROID_ABI="riscv64" ..
+        cmake --build . -j $(nproc)
 
     - name: armeabi-v7a-shared
       run: |
@@ -85,6 +92,11 @@ jobs:
         mkdir build-x86_64-shared && cd build-x86_64-shared
         cmake ${{ env.NCNN_CMAKE_OPTIONS }} -DANDROID_ABI="x86_64" -DNCNN_SHARED_LIB=ON ..
         cmake --build . -j $(nproc)
+    - name: riscv64-shared
+      run: |
+        mkdir build-riscv64-shared && cd build-riscv64-shared
+        cmake ${{ env.NCNN_CMAKE_OPTIONS }} -DANDROID_ABI="riscv64" -DNCNN_SHARED_LIB=ON ..
+        cmake --build . -j $(nproc)
 
   ndk-r16b:
     runs-on: ubuntu-latest

.github/workflows/elf-riscv32-cpu-gcc.yml → .github/workflows/elf-riscv32.yml

@@ -1,9 +1,9 @@
-name: elf-riscv32-cpu-gcc
+name: elf-riscv32
 on:
   push:
     branches: [master]
     paths:
-    - '.github/workflows/elf-riscv32-cpu-gcc.yml'
+    - '.github/workflows/elf-riscv32.yml'
     - 'toolchains/riscv32-unknown-elf.toolchain.cmake'
     - 'CMakeLists.txt'
     - 'cmake/**'
@@ -14,7 +14,7 @@ on:
   pull_request:
     branches: [master]
     paths:
-    - '.github/workflows/elf-riscv32-cpu-gcc.yml'
+    - '.github/workflows/elf-riscv32.yml'
     - 'toolchains/riscv32-unknown-elf.toolchain.cmake'
     - 'CMakeLists.txt'
     - 'cmake/**'
@@ -23,7 +23,7 @@ on:
     - 'src/layer/riscv/**'
     - 'tests/**'
 concurrency:
-  group: elf-riscv32-cpu-gcc-${{ github.ref }}
+  group: elf-riscv32-${{ github.ref }}
   cancel-in-progress: true
 permissions:
   contents: read

.github/workflows/elf-riscv64-cpu-gcc.yml → .github/workflows/elf-riscv64.yml

@@ -1,9 +1,9 @@
-name: elf-riscv64-cpu-gcc
+name: elf-riscv64
 on:
   push:
     branches: [master]
     paths:
-    - '.github/workflows/elf-riscv64-cpu-gcc.yml'
+    - '.github/workflows/elf-riscv64.yml'
     - 'toolchains/riscv64-unknown-elf.toolchain.cmake'
     - 'CMakeLists.txt'
     - 'cmake/**'
@@ -14,7 +14,7 @@ on:
   pull_request:
     branches: [master]
     paths:
-    - '.github/workflows/elf-riscv64-cpu-gcc.yml'
+    - '.github/workflows/elf-riscv64.yml'
     - 'toolchains/riscv64-unknown-elf.toolchain.cmake'
     - 'CMakeLists.txt'
     - 'cmake/**'
@@ -23,7 +23,7 @@ on:
     - 'src/layer/riscv/**'
     - 'tests/**'
 concurrency:
-  group: elf-riscv64-cpu-gcc-${{ github.ref }}
+  group: elf-riscv64-${{ github.ref }}
   cancel-in-progress: true
 permissions:
   contents: read

.github/workflows/linux-aarch64-cpu-gcc.yml → .github/workflows/linux-aarch64.yml

@@ -1,9 +1,9 @@
-name: linux-aarch64-cpu-gcc
+name: linux-aarch64
 on:
   push:
     branches: [master]
     paths:
-    - '.github/workflows/linux-aarch64-cpu-gcc.yml'
+    - '.github/workflows/linux-aarch64.yml'
     - 'toolchains/aarch64-linux-gnu.toolchain.cmake'
     - 'CMakeLists.txt'
     - 'cmake/**'
@@ -14,7 +14,7 @@ on:
   pull_request:
     branches: [master]
     paths:
-    - '.github/workflows/linux-aarch64-cpu-gcc.yml'
+    - '.github/workflows/linux-aarch64.yml'
     - 'toolchains/aarch64-linux-gnu.toolchain.cmake'
     - 'CMakeLists.txt'
     - 'cmake/**'
@@ -23,13 +23,13 @@ on:
     - 'src/layer/arm/**'
     - 'tests/**'
 concurrency:
-  group: linux-aarch64-cpu-gcc-${{ github.ref }}
+  group: linux-aarch64-${{ github.ref }}
   cancel-in-progress: true
 permissions:
   contents: read
 
 jobs:
-  linux-gcc:
+  gcc-aarch64:
     runs-on: ubuntu-20.04
     steps:
     - uses: actions/checkout@v4
@@ -98,7 +98,7 @@ jobs:
         cd build-simplestl-simplemath
         TESTS_EXECUTABLE_LOADER=qemu-aarch64 TESTS_EXECUTABLE_LOADER_ARGUMENTS="-L;/usr/aarch64-linux-gnu" ctest --output-on-failure -j $(nproc)
 
-  linux-gcc-arm82:
+  gcc-arm82:
     runs-on: ubuntu-20.04
     steps:
     - uses: actions/checkout@v4
@@ -156,7 +156,7 @@ jobs:
         cd build-noint8
         TESTS_EXECUTABLE_LOADER=qemu-aarch64 TESTS_EXECUTABLE_LOADER_ARGUMENTS="-L;/usr/aarch64-linux-gnu" ctest --output-on-failure -j $(nproc)
 
-  linux-gcc-arm86:
+  gcc-arm86:
     runs-on: ubuntu-22.04
     steps:
     - uses: actions/checkout@v4

.github/workflows/linux-arm-cpu-gcc.yml → .github/workflows/linux-arm.yml

@@ -1,9 +1,9 @@
-name: linux-arm-cpu-gcc
+name: linux-arm
 on:
   push:
     branches: [master]
     paths:
-    - '.github/workflows/linux-arm-cpu-gcc.yml'
+    - '.github/workflows/linux-arm.yml'
     - 'toolchains/arm-linux-gnueabi.toolchain.cmake'
     - 'toolchains/arm-linux-gnueabihf.toolchain.cmake'
     - 'CMakeLists.txt'
@@ -15,7 +15,7 @@ on:
   pull_request:
     branches: [master]
     paths:
-    - '.github/workflows/linux-arm-cpu-gcc.yml'
+    - '.github/workflows/linux-arm.yml'
     - 'toolchains/arm-linux-gnueabi.toolchain.cmake'
     - 'toolchains/arm-linux-gnueabihf.toolchain.cmake'
     - 'CMakeLists.txt'
@@ -25,13 +25,13 @@ on:
     - 'src/layer/arm/**'
     - 'tests/**'
 concurrency:
-  group: linux-arm-cpu-gcc-${{ github.ref }}
+  group: linux-arm-${{ github.ref }}
   cancel-in-progress: true
 permissions:
   contents: read
 
 jobs:
-  linux-gcc-arm:
+  gcc-arm:
     runs-on: ubuntu-20.04
     steps:
     - uses: actions/checkout@v4
@@ -89,7 +89,7 @@ jobs:
         cd build-noint8
         TESTS_EXECUTABLE_LOADER=qemu-arm TESTS_EXECUTABLE_LOADER_ARGUMENTS="-L;/usr/arm-linux-gnueabi" ctest --output-on-failure -j $(nproc)
 
-  linux-gcc-armhf:
+  gcc-armhf:
     runs-on: ubuntu-20.04
     steps:
     - uses: actions/checkout@v4
@@ -147,7 +147,7 @@ jobs:
         cd build-noint8
         TESTS_EXECUTABLE_LOADER=qemu-arm TESTS_EXECUTABLE_LOADER_ARGUMENTS="-L;/usr/arm-linux-gnueabihf" ctest --output-on-failure -j $(nproc)
 
-  linux-gcc-armhf-vfpv3-d16:
+  gcc-armhf-vfpv3-d16:
     runs-on: ubuntu-20.04
     steps:
     - uses: actions/checkout@v4

.github/workflows/linux-loongarch64-cpu-gcc.yml → .github/workflows/linux-loongarch64.yml

@@ -1,9 +1,9 @@
-name: linux-loongarch64-cpu-gcc
+name: linux-loongarch64
 on:
   push:
     branches: [master]
     paths:
-    - '.github/workflows/linux-loongarch64-cpu-gcc.yml'
+    - '.github/workflows/linux-loongarch64.yml'
     - 'toolchains/loongarch64-linux-gnu.toolchain.cmake'
     - 'toolchains/loongarch64-unknown-linux-gnu.toolchain.cmake'
     - 'CMakeLists.txt'
@@ -15,7 +15,7 @@ on:
   pull_request:
     branches: [master]
     paths:
-    - '.github/workflows/linux-loongarch64-cpu-gcc.yml'
+    - '.github/workflows/linux-loongarch64.yml'
     - 'toolchains/loongarch64-linux-gnu.toolchain.cmake'
     - 'toolchains/loongarch64-unknown-linux-gnu.toolchain.cmake'
     - 'CMakeLists.txt'
@@ -25,13 +25,13 @@ on:
     - 'src/layer/loongarch/**'
     - 'tests/**'
 concurrency:
-  group: linux-loongarch64-cpu-gcc-${{ github.ref }}
+  group: linux-loongarch64-${{ github.ref }}
   cancel-in-progress: true
 permissions:
   contents: read
 
 jobs:
-  linux-gcc-loongarch64:
+  gcc-loongarch64:
     runs-on: [self-hosted, linux, centos]
 
     steps: