diff --git a/.github/workflows/visionos-cpu.yml b/.github/workflows/visionos-cpu.yml
index ac82a4789cc..0a3cfdda7b1 100644
--- a/.github/workflows/visionos-cpu.yml
+++ b/.github/workflows/visionos-cpu.yml
@@ -10,16 +10,16 @@ on:
     - 'src/*'
     - 'src/layer/*'
     - 'src/layer/arm/**'
-  pull_request:
-    branches: [master]
-    paths:
-    - '.github/workflows/visionos-cpu.yml'
-    - 'toolchains/ios.toolchain.cmake'
-    - 'CMakeLists.txt'
-    - 'cmake/**'
-    - 'src/*'
-    - 'src/layer/*'
-    - 'src/layer/arm/**'
+  # pull_request:
+  #   branches: [master]
+  #   paths:
+  #   - '.github/workflows/visionos-cpu.yml'
+  #   - 'toolchains/ios.toolchain.cmake'
+  #   - 'CMakeLists.txt'
+  #   - 'cmake/**'
+  #   - 'src/*'
+  #   - 'src/layer/*'
+  #   - 'src/layer/arm/**'
 concurrency:
   group: visionos-cpu-${{ github.ref }}
   cancel-in-progress: true
diff --git a/.github/workflows/watchos.yml b/.github/workflows/watchos.yml
index 2ff98789124..adfbb079d31 100644
--- a/.github/workflows/watchos.yml
+++ b/.github/workflows/watchos.yml
@@ -38,42 +38,127 @@ jobs:
   build:
     runs-on: macos-13
     env:
-      COMMON_CMAKE_OPTIONS: |
+      OPENMP_CMAKE_OPTIONS: |
+        -DCMAKE_TOOLCHAIN_FILE=../../toolchains/ios.toolchain.cmake \
+        -DDEPLOYMENT_TARGET=$WATCHOS_DEPLOYMENT_TARGET \
+        -DENABLE_BITCODE=$ENABLE_BITCODE \
+        -DENABLE_ARC=$ENABLE_ARC \
+        -DENABLE_VISIBILITY=$ENABLE_VISIBILITY \
+        -DCMAKE_INSTALL_PREFIX=install \
+        -DCMAKE_BUILD_TYPE=Release \
+        -DPERL_EXECUTABLE=/usr/local/bin/perl \
+        -DLIBOMP_ENABLE_SHARED=OFF \
+        -DLIBOMP_OMPT_SUPPORT=OFF \
+        -DLIBOMP_USE_HWLOC=OFF \
+
+      NCNN_CMAKE_OPTIONS: |
         -DCMAKE_TOOLCHAIN_FILE=../toolchains/ios.toolchain.cmake \
         -DDEPLOYMENT_TARGET=$WATCHOS_DEPLOYMENT_TARGET \
         -DENABLE_BITCODE=$ENABLE_BITCODE \
         -DENABLE_ARC=$ENABLE_ARC \
         -DENABLE_VISIBILITY=$ENABLE_VISIBILITY \
-        -DNCNN_OPENMP=OFF -DNCNN_THREADS=OFF \
         -DCMAKE_INSTALL_PREFIX=install \
         -DCMAKE_BUILD_TYPE=Release \
+        -DOpenMP_C_FLAGS="-Xclang -fopenmp" -DOpenMP_CXX_FLAGS="-Xclang -fopenmp" \
+        -DOpenMP_C_LIB_NAMES="libomp" -DOpenMP_CXX_LIB_NAMES="libomp" \
+        -DOpenMP_libomp_LIBRARY="libomp.a" \
 
     steps:
     - uses: actions/checkout@v4
 
+    - name: cache-openmp
+      id: cache-openmp
+      uses: actions/cache@v4
+      with:
+        path: openmp-install
+        key: openmp-watchos-install-20240401
+    - name: openmp
+      if: steps.cache-openmp.outputs.cache-hit != 'true'
+      run: |
+        wget https://github.com/llvm/llvm-project/releases/download/llvmorg-18.1.2/openmp-18.1.2.src.tar.xz
+        tar -xf openmp-18.1.2.src.tar.xz
+        cd openmp-18.1.2.src
+        sed -i'' -e '/.size __kmp_unnamed_critical_addr/d' runtime/src/z_Linux_asm.S
+        sed -i'' -e 's/__kmp_unnamed_critical_addr/___kmp_unnamed_critical_addr/g' runtime/src/z_Linux_asm.S
+    - name: openmp-armv7k
+      if: steps.cache-openmp.outputs.cache-hit != 'true'
+      run: |
+        cd openmp-18.1.2.src
+        mkdir -p build-armv7k && cd build-armv7k
+        cmake ${{ env.OPENMP_CMAKE_OPTIONS }} -DPLATFORM=WATCHOS -DARCHS="armv7k" ..
+        cmake --build . -j 4
+        cmake --build . --target install
+    - name: openmp-arm64_32
+      if: steps.cache-openmp.outputs.cache-hit != 'true'
+      run: |
+        cd openmp-18.1.2.src
+        mkdir -p build-arm64_32 && cd build-arm64_32
+        cmake ${{ env.OPENMP_CMAKE_OPTIONS }} -DPLATFORM=WATCHOS -DARCHS="arm64_32" ..
+        cmake --build . -j 4
+        cmake --build . --target install
+    - name: openmp-simulator-x86_64
+      if: steps.cache-openmp.outputs.cache-hit != 'true'
+      run: |
+        cd openmp-18.1.2.src
+        mkdir -p build-simulator-x86_64 && cd build-simulator-x86_64
+        cmake ${{ env.OPENMP_CMAKE_OPTIONS }} -DPLATFORM=SIMULATOR_WATCHOS -DARCHS="x86_64" ..
+        cmake --build . -j 4
+        cmake --build . --target install
+    - name: openmp-simulator-arm64
+      if: steps.cache-openmp.outputs.cache-hit != 'true'
+      run: |
+        cd openmp-18.1.2.src
+        mkdir -p build-simulator-arm64 && cd build-simulator-arm64
+        cmake ${{ env.OPENMP_CMAKE_OPTIONS }} -DPLATFORM=SIMULATOR_WATCHOS -DARCHS="arm64" ..
+        cmake --build . -j 4
+        cmake --build . --target install
+    - name: openmp-merge-fat-library
+      if: steps.cache-openmp.outputs.cache-hit != 'true'
+      run: |
+        mkdir -p $GITHUB_WORKSPACE/openmp-install
+        mkdir -p $GITHUB_WORKSPACE/openmp-install/watchos
+        mkdir -p $GITHUB_WORKSPACE/openmp-install/watchos-simulator
+
+        cp -a openmp-18.1.2.src/build-arm64_32/install/include $GITHUB_WORKSPACE/openmp-install/watchos
+        mkdir -p $GITHUB_WORKSPACE/openmp-install/watchos/lib
+        lipo -create \
+            openmp-18.1.2.src/build-armv7k/install/lib/libomp.a \
+            openmp-18.1.2.src/build-arm64_32/install/lib/libomp.a \
+            -o $GITHUB_WORKSPACE/openmp-install/watchos/lib/libomp.a
+
+        cp -a openmp-18.1.2.src/build-simulator-x86_64/install/include $GITHUB_WORKSPACE/openmp-install/watchos-simulator
+        mkdir -p $GITHUB_WORKSPACE/openmp-install/watchos-simulator/lib
+        lipo -create \
+            openmp-18.1.2.src/build-simulator-x86_64/install/lib/libomp.a \
+            openmp-18.1.2.src/build-simulator-arm64/install/lib/libomp.a \
+            -o $GITHUB_WORKSPACE/openmp-install/watchos-simulator/lib/libomp.a
+
+    - name: install-openmp
+      run: |
+        sudo cp $GITHUB_WORKSPACE/openmp-install/watchos/include/* $DEVELOPER_DIR/Platforms/WatchOS.platform/Developer/SDKs/WatchOS.sdk/usr/include
+        sudo cp $GITHUB_WORKSPACE/openmp-install/watchos/lib/libomp.a $DEVELOPER_DIR/Platforms/WatchOS.platform/Developer/SDKs/WatchOS.sdk/usr/lib
+
+        sudo cp $GITHUB_WORKSPACE/openmp-install/watchos-simulator/include/* $DEVELOPER_DIR/Platforms/WatchSimulator.platform/Developer/SDKs/WatchSimulator.sdk/usr/include
+        sudo cp $GITHUB_WORKSPACE/openmp-install/watchos-simulator/lib/libomp.a $DEVELOPER_DIR/Platforms/WatchSimulator.platform/Developer/SDKs/WatchSimulator.sdk/usr/lib
+
     - name: armv7k
       run: |
         mkdir build-armv7k && cd build-armv7k
-        cmake ${{ env.COMMON_CMAKE_OPTIONS }} -DPLATFORM=WATCHOS -DARCHS="armv7k" ..
+        cmake ${{ env.NCNN_CMAKE_OPTIONS }} -DPLATFORM=WATCHOS -DARCHS="armv7k" ..
         cmake --build . -j 4
     - name: arm64_32
       run: |
         mkdir build-arm64_32 && cd build-arm64_32
-        cmake ${{ env.COMMON_CMAKE_OPTIONS }} -DPLATFORM=WATCHOS -DARCHS="arm64_32" ..
+        cmake ${{ env.NCNN_CMAKE_OPTIONS }} -DPLATFORM=WATCHOS -DARCHS="arm64_32" ..
         cmake --build . -j 4
 
-    - name: simulator-i386
-      run: |
-        mkdir build-simulator-i386 && cd build-simulator-i386
-        cmake ${{ env.COMMON_CMAKE_OPTIONS }} -DPLATFORM=SIMULATOR_WATCHOS -DARCHS="i386" ..
-        cmake --build . -j 4
     - name: simulator-x86_64
       run: |
         mkdir build-simulator-x86_64 && cd build-simulator-x86_64
-        cmake ${{ env.COMMON_CMAKE_OPTIONS }} -DPLATFORM=SIMULATOR_WATCHOS -DARCHS="x86_64" ..
+        cmake ${{ env.NCNN_CMAKE_OPTIONS }} -DPLATFORM=SIMULATOR_WATCHOS -DARCHS="x86_64" ..
         cmake --build . -j 4
     - name: simulator-arm64
       run: |
         mkdir build-simulator-arm64 && cd build-simulator-arm64
-        cmake ${{ env.COMMON_CMAKE_OPTIONS }} -DPLATFORM=SIMULATOR_WATCHOS -DARCHS="arm64" ..
+        cmake ${{ env.NCNN_CMAKE_OPTIONS }} -DPLATFORM=SIMULATOR_WATCHOS -DARCHS="arm64" ..
         cmake --build . -j 4