fix: Copy dependencies for windows

fix: :(

fix

fix

fix: Copy runtime dlls on windows

fix: Add explicit copy command for windows dlls

fix

fix

fix: >:(

fix

fix

fix

fix

Update path on windows

check dll dependancies

fix: Update PATH on win32

ci: Update test.yaml

Author: abetlen

.github/workflows/test.yaml

@@ -81,32 +81,6 @@ jobs:
         run: |
           python -m pytest
 
-  # build-linux-opencl:
-
-  #   runs-on: ubuntu-latest
-
-  #   steps:
-  #     - uses: actions/checkout@v4
-  #       with:
-  #         submodules: "recursive"
-  #     - name: Set up Python 3.8
-  #       uses: actions/setup-python@v5
-  #       with:
-  #         python-version: "3.8"
-  #     - name: Set up OpenCL & CLBlast
-  #       run: |
-  #         wget -O- https://apt.repos.intel.com/intel-gpg-keys/GPG-PUB-KEY-INTEL-SW-PRODUCTS.PUB | gpg --dearmor | sudo tee /usr/share/keyrings/oneapi-archive-keyring.gpg > /dev/null
-  #         echo "deb [signed-by=/usr/share/keyrings/oneapi-archive-keyring.gpg] https://apt.repos.intel.com/oneapi all main" | sudo tee /etc/apt/sources.list.d/oneAPI.list  
-  #         sudo apt-get update
-  #         sudo apt-get install -y --no-install-recommends llvm intel-oneapi-runtime-opencl intel-oneapi-runtime-compilers libclblast-dev
-  #     - name: Install dependencies
-  #       run: |
-  #         python -m pip install --upgrade pip
-  #         CMAKE_ARGS="-DLLAMA_CLBLAST=on" python -m pip install .[all] --verbose
-  #     - name: Test with pytest
-  #       run: |
-  #         python -m pytest
-
 
   build-macos-metal:
 

CMakeLists.txt

@@ -72,11 +72,24 @@ if (LLAMA_BUILD)
     add_subdirectory(vendor/llama.cpp)
     llama_cpp_python_install_target(llama)
     llama_cpp_python_install_target(ggml)
+    
     # Workaround for Windows + CUDA https://github.com/abetlen/llama-cpp-python/issues/563
-    if (WIN32 AND (LLAMA_CUDA OR LLAMA_CUBLAS))
+    if (WIN32)
         install(
             FILES $<TARGET_RUNTIME_DLLS:llama>
-            DESTINATION ${LLAMA_CPP_PYTHON_INSTALL_DIR}
+            DESTINATION ${CMAKE_CURRENT_SOURCE_DIR}/llama_cpp/lib
+        )
+        install(
+            FILES $<TARGET_RUNTIME_DLLS:llama>
+            DESTINATION ${SKBUILD_PLATLIB_DIR}/llama_cpp/lib
+        )
+        install(
+            FILES $<TARGET_RUNTIME_DLLS:ggml>
+            DESTINATION ${CMAKE_CURRENT_SOURCE_DIR}/llama_cpp/lib
+        )
+        install(
+            FILES $<TARGET_RUNTIME_DLLS:ggml>
+            DESTINATION ${SKBUILD_PLATLIB_DIR}/llama_cpp/lib
         )
     endif()
 
@@ -98,5 +111,15 @@ if (LLAMA_BUILD)
             set_target_properties(llava_shared PROPERTIES CUDA_ARCHITECTURES OFF)
         endif()
         llama_cpp_python_install_target(llava_shared)
+        if (WIN32)
+            install(
+                FILES $<TARGET_RUNTIME_DLLS:llava_shared>
+                DESTINATION ${CMAKE_CURRENT_SOURCE_DIR}/llama_cpp/lib
+            )
+            install(
+                FILES $<TARGET_RUNTIME_DLLS:llava_shared>
+                DESTINATION ${SKBUILD_PLATLIB_DIR}/llama_cpp/lib
+            )
+        endif()
     endif()
-endif()
+endif()

llama_cpp/llama_cpp.py

@@ -52,7 +52,12 @@ def _load_shared_library(lib_base_name: str):
         _lib_paths = [_lib.resolve()]
 
     cdll_args = dict()  # type: ignore
+
     # Add the library directory to the DLL search path on Windows (if needed)
+    if sys.platform == "win32":
+        os.add_dll_directory(str(_base_path))
+        os.environ['PATH'] = str(_base_path) + os.pathsep + os.environ['PATH']
+
     if sys.platform == "win32" and sys.version_info >= (3, 8):
         os.add_dll_directory(str(_base_path))
         if "CUDA_PATH" in os.environ: