Update Makefile

tc-wolf · tc-wolf · commit aff151d2a8cf · 2025-03-18T14:14:40.000-05:00
- Needed for building the llama-cpp-python server for Mac OS
- Include all libs in `libggml_base_path`
diff --git a/Makefile b/Makefile
@@ -89,16 +89,18 @@ deploy.pyinstaller.mac:
 	@if [ `uname -m` != "arm64" ]; then echo "Must be on aarch64"; exit 1; fi
 	@if [ `uname` != "Darwin" ]; then echo "Must be on MacOS"; exit 1; fi
 	@echo "Building and installing with proper env vars for aarch64-specific ops"
-	CMAKE_ARGS="-DGGML_METAL=off -DGGML_LLAMAFILE=OFF -DGGML_BLAS=OFF -DCMAKE_BUILD_TYPE=Release" python3 -m pip install -v -e .[server,dev]
+
+	# This still builds with metal support (I think b/c GGML_NATIVE=ON). Not an
+	# issue since can still run Q4_0 models w/ repacking support on CPU if `-ngl 0`.
+	CMAKE_ARGS="-DGGML_METAL=OFF -DGGML_LLAMAFILE=OFF -DGGML_BLAS=OFF \
+	-DGGML_NATIVE=ON -DGGML_CPU_AARCH64=ON \
+	-DCMAKE_BUILD_TYPE=Release" python3 -m pip install -v -e .[server,dev]
 	@server_path=$$(python -c 'import llama_cpp.server; print(llama_cpp.server.__file__)' | sed s/init/main/) ; \
 	echo "Server path: $$server_path" ; \
-	libllama_path=$$(python -c 'import llama_cpp.llama_cpp; print(llama_cpp.llama_cpp._load_shared_library("llama")._name)') ; \
-	libggml_path=$$(python -c 'import llama_cpp.llama_cpp; print(llama_cpp.llama_cpp._load_shared_library("ggml")._name)') ; \
-	echo "libllama path: $$libllama_path" ; \
-	echo "libggml path: $$libggml_path" ; \
+	base_path=$$(python -c 'from llama_cpp._ggml import libggml_base_path; print(str(libggml_base_path))') ; \
+	echo "Base path: $$base_path" ; \
 	pyinstaller -DF $$server_path \
-	--add-data $$libllama_path:llama_cpp/lib \
-	--add-data $$libggml_path:llama_cpp/lib \
+	--add-data $$base_path:llama_cpp/lib \
 	-n llama-cpp-py-server
 
 test:
