Merge branch 'master' into fix_decoding

* master: (119 commits) cann : add Ascend NPU support (ggerganov#2336) whisper : fix compile warning (#0) sync : ggml ggml : add CANN backend (llama/0) scripts : sync cann ci : disable ruby workflow (#0) ci : try to fix FreeBSD (#0) build : fix aarch64 (#0) talk-llama : sync llama.cpp sync : ggml ggml-backend : fix async copy from CPU (llama/8897) Updated SYCL device filtering (llama/8901) CUDA/HIP: fix tests/test-backend-ops (llama/8896) CUDA: fix padding logic for FP16/FP32 (llama/8884) ggml : add epsilon as a parameter for group_norm (llama/8818) ggml : fix overflows in elu function (llama/8866) ggml : reading the runtime sve config of the cpu (llama/8709) Fix conversion of unnormalized BF16->BF16 weights (llama/7843) Fixing wrong VDR iq4nl value (llama/8812) ggml-cuda: Adding support for unified memory (llama/8035) ...
bygreencn · Aug 12, 2024 · 6f54482 · 6f54482
2 parents 3e301b9 + 2ca527c
commit 6f54482
Show file tree

Hide file tree

Showing 426 changed files with 203,008 additions and 28,809 deletions.
diff --git a/.devops/cublas.Dockerfile b/.devops/cublas.Dockerfile
@@ -21,7 +21,7 @@ COPY . .
 # Set nvcc architecture
 ENV CUDA_DOCKER_ARCH=${CUDA_DOCKER_ARCH}
 # Enable cuBLAS
-ENV WHISPER_CUBLAS=1
+ENV GGML_CUDA=1
 
 RUN make
 

diff --git a/.devops/main-cuda.Dockerfile b/.devops/main-cuda.Dockerfile
@@ -14,7 +14,7 @@ ARG CUDA_DOCKER_ARCH=all
 # Set nvcc architecture
 ENV CUDA_DOCKER_ARCH=${CUDA_DOCKER_ARCH}
 # Enable cuBLAS
-ENV WHISPER_CUBLAS=1
+ENV GGML_CUDA=1
 
 RUN apt-get update && \
     apt-get install -y build-essential \

diff --git a/.github/workflows/bindings-ruby.yml → .github/workflows/bindings-ruby.yml.disabled b/.github/workflows/bindings-ruby.yml → .github/workflows/bindings-ruby.yml.disabled
@@ -1,3 +1,4 @@
+# TODO: fix this workflow file, disabled for now
 name: Bindings Tests (Ruby)
 on:
   push:

diff --git a/.github/workflows/build.yml b/.github/workflows/build.yml
@@ -59,7 +59,7 @@ jobs:
         uses: cross-platform-actions/[email protected]
         with:
           operating_system: freebsd
-          version: '13.2'
+          version: '13.3'
           run: |
             sudo pkg update
             sudo pkg install -y gmake sdl2
@@ -101,7 +101,10 @@ jobs:
       fail-fast: false
       matrix:
         build: [Debug, Release]
-        arch: [linux/amd64, linux/arm64, linux/arm/v7, linux/ppc64le]
+        #arch: [linux/amd64, linux/arm64, linux/arm/v7, linux/ppc64le]
+        # TODO: arm/v7 disabled due to clang bug
+        #       https://github.com/ggerganov/whisper.cpp/actions/runs/9657764109/job/26637633042?pr=2256#step:4:1990
+        arch: [linux/amd64, linux/arm64, linux/ppc64le]
 
     steps:
       - name: Clone
@@ -197,7 +200,7 @@ jobs:
           source /opt/intel/oneapi/setvars.sh
           mkdir build
           cd build
-          cmake -DWHISPER_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx ..
+          cmake -DGGML_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx ..
           cmake --build . --config Release -j $(nproc)
 
   ubuntu-22-cmake-sycl-fp16:
@@ -247,7 +250,7 @@ jobs:
           source /opt/intel/oneapi/setvars.sh
           mkdir build
           cd build
-          cmake -DWHISPER_SYCL_F16=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx ..
+          cmake -DGGML_SYCL_F16=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx ..
           cmake --build . --config Release -j $(nproc)
 
   windows-msys2:
@@ -289,7 +292,7 @@ jobs:
       - name: Build using make w/ OpenBLAS
         shell: msys2 {0}
         run: |
-            make WHISPER_OPENBLAS=1 -j $(nproc)
+            make GGML_OPENBLAS=1 -j $(nproc)
 
       - name: Build using CMake
         shell: msys2 {0}
@@ -305,7 +308,7 @@ jobs:
       - name: Build using CMake w/ OpenBLAS
         shell: msys2 {0}
         run: |
-            cmake -B build -DWHISPER_OPENBLAS=ON
+            cmake -B build -DGGML_OPENBLAS=ON
             cmake --build build --config ${{ matrix.build }} -j $(nproc)
 
   windows:
@@ -381,12 +384,9 @@ jobs:
           - arch: Win32
             obzip: https://github.com/OpenMathLib/OpenBLAS/releases/download/v0.3.25/OpenBLAS-0.3.25-x86.zip
             s2arc: x86
-            clblast: OFF
           - arch: x64
             obzip: https://github.com/OpenMathLib/OpenBLAS/releases/download/v0.3.25/OpenBLAS-0.3.25-x64.zip
             s2arc: x64
-            clblast: ON
-            clver: 1.6.1
           - sdl2: ON
             s2ver: 2.28.5
 
@@ -413,26 +413,13 @@ jobs:
           7z x sdl2.zip
           echo "SDL2_DIR=$env:GITHUB_WORKSPACE/SDL2-${{ matrix.s2ver }}/cmake" >> $env:GITHUB_ENV
 
-      - name: Install OpenCL
-        if: matrix.clblast == 'ON'
-        run: vcpkg.exe --triplet=${{ matrix.arch }}-windows install opencl
-
-      - name: Fetch CLBlast and set CLBlast_DIR
-        if: matrix.clblast == 'ON'
-        run: |
-          C:/msys64/usr/bin/wget.exe -qO clblast.zip https://github.com/CNugteren/CLBlast/releases/download/${{ matrix.clver }}/CLBlast-${{ matrix.clver }}-windows-x64.zip
-          7z x clblast.zip
-          7z x CLBlast-${{ matrix.clver }}-windows-x64.7z
-          echo "CLBlast_DIR=$env:GITHUB_WORKSPACE/CLBlast-${{ matrix.clver }}-windows-x64/lib/cmake/CLBlast" >> $env:GITHUB_ENV
-
       - name: Configure
         run: >
           cmake -S . -B ./build -A ${{ matrix.arch }}
           -DCMAKE_BUILD_TYPE=${{ matrix.build }}
-          -DWHISPER_OPENBLAS=${{ matrix.blas }}
+          -DGGML_OPENBLAS=${{ matrix.blas }}
           -DCMAKE_LIBRARY_PATH="$env:OPENBLAS_PATH/lib"
           -DWHISPER_SDL2=${{ matrix.sdl2 }}
-          -DWHISPER_CLBLAST=${{ matrix.clblast }}
 
       - name: Build
         run: |
@@ -447,15 +434,11 @@ jobs:
         if: matrix.sdl2 == 'ON'
         run: copy "$env:SDL2_DIR/../lib/${{ matrix.s2arc }}/SDL2.dll" build/bin/${{ matrix.build }}
 
-      - name: Copy clblast.dll
-        if: matrix.clblast == 'ON'
-        run: copy "$env:CLBlast_DIR/../../clblast.dll" build/bin/${{ matrix.build }}
-
       - name: Upload binaries
         if: matrix.blas == 'ON' && matrix.sdl2 == 'ON'
         uses: actions/upload-artifact@v4
         with:
-          name: whisper-blas${{ matrix.clblast == 'ON' && '-clblast' || ''}}-bin-${{ matrix.arch }}
+          name: whisper-blas-bin-${{ matrix.arch }}
           path: build/bin/${{ matrix.build }}
 
   windows-cublas:
@@ -498,7 +481,7 @@ jobs:
         run: >
           cmake -S . -B ./build -A ${{ matrix.arch }}
           -DCMAKE_BUILD_TYPE=${{ matrix.build }}
-          -DWHISPER_CUDA=${{ matrix.cublas }}
+          -DGGML_CUDA=${{ matrix.cublas }}
           -DWHISPER_SDL2=${{ matrix.sdl2 }}
 
       - name: Build ${{ matrix.cuda-toolkit }}
@@ -628,48 +611,49 @@ jobs:
           chmod +x ./gradlew
           ./gradlew assembleRelease
 
-  java:
-    needs: [ 'windows' ]
-    runs-on: windows-latest
-    steps:
-      - uses: actions/checkout@v4
-
-      - name: Install Java
-        uses: actions/setup-java@v4
-        with:
-          distribution: zulu
-          java-version: 20
-
-      - name: Download Windows lib
-        uses: actions/download-artifact@v4
-        with:
-          name: win32-x86-64_whisper.dll
-          path: bindings/java/build/generated/resources/main/win32-x86-64
-
-      - name: Build
-        run: |
-          models\download-ggml-model.cmd tiny.en
-          cd bindings/java
-          chmod +x ./gradlew
-          ./gradlew build
-
-      - name: Upload jar
-        uses: actions/upload-artifact@v4
-        with:
-          name: whispercpp.jar
-          path: bindings/java/build/libs/whispercpp-*.jar
-
-      - name: Publish package
-        if: ${{ github.ref == 'refs/heads/master' }}
-        uses: gradle/[email protected]
-        with:
-          arguments: publish
-          build-root-directory: bindings/java
-        env:
-          MAVEN_USERNAME: ${{ secrets.JIRA_USER }}
-          MAVEN_PASSWORD: ${{ secrets.JIRA_PASS }}
-          PGP_SECRET: ${{ secrets.GPG_PRIVATE_KEY }}
-          PGP_PASSPHRASE: ${{ secrets.GPG_PASSPHRASE }}
+# TODO: disabled because of following fail: https://github.com/ggerganov/whisper.cpp/actions/runs/9686220096/job/26735899598
+#  java:
+#    needs: [ 'windows' ]
+#    runs-on: windows-latest
+#    steps:
+#      - uses: actions/checkout@v4
+#
+#      - name: Install Java
+#        uses: actions/setup-java@v4
+#        with:
+#          distribution: zulu
+#          java-version: 20
+#
+#      - name: Download Windows lib
+#        uses: actions/download-artifact@v4
+#        with:
+#          name: win32-x86-64_whisper.dll
+#          path: bindings/java/build/generated/resources/main/win32-x86-64
+#
+#      - name: Build
+#        run: |
+#          models\download-ggml-model.cmd tiny.en
+#          cd bindings/java
+#          chmod +x ./gradlew
+#          ./gradlew build
+#
+#      - name: Upload jar
+#        uses: actions/upload-artifact@v4
+#        with:
+#          name: whispercpp.jar
+#          path: bindings/java/build/libs/whispercpp-*.jar
+#
+#      - name: Publish package
+#        if: ${{ github.ref == 'refs/heads/master' }}
+#        uses: gradle/[email protected]
+#        with:
+#          arguments: publish
+#          build-root-directory: bindings/java
+#        env:
+#          MAVEN_USERNAME: ${{ secrets.JIRA_USER }}
+#          MAVEN_PASSWORD: ${{ secrets.JIRA_PASS }}
+#          PGP_SECRET: ${{ secrets.GPG_PRIVATE_KEY }}
+#          PGP_PASSPHRASE: ${{ secrets.GPG_PASSPHRASE }}
 
   quantize:
     runs-on: ubuntu-latest

diff --git a/.gitmodules b/.gitmodules
@@ -1,3 +0,0 @@
-[submodule "bindings/ios"]
-	path = bindings/ios
-	url = https://github.com/ggerganov/whisper.spm