llama.cpp cmake

CMakePresets.json

{
	"version": 4,
	"configurePresets": [
		{
			"name": "default",
			"displayName": "default",
			"binaryDir": "${workspaceRoot}/build/${presetName}",
			"cacheVariables": {
				"CMAKE_INSTALL_PREFIX": "${workspaceRoot}/install/${presetName}",
				"CMAKE_C_COMPILER": "/usr/bin/gcc",
				"CMAKE_CXX_COMPILER": "/usr/bin/g++",
				"CMAKE_BUILD_TYPE": "Release",
				"CMAKE_CXX_FLAGS": "-mcpu=native",
				"CMAKE_C_FLAGS": "-mcpu=native",
				"GGML_LLAMAFILE": "OFF"
			}
		},
        {
			"name": "cuda",
			"displayName": "cuda",
			"binaryDir": "${workspaceRoot}/build/${presetName}",
			"cacheVariables": {
				"CMAKE_INSTALL_PREFIX": "${workspaceRoot}/install/${presetName}",
				"CMAKE_C_COMPILER": "/opt/nvidia/hpc_sdk/Linux_aarch64/24.3/compilers/bin/nvcc",
				"CMAKE_CXX_COMPILER": "/opt/nvidia/hpc_sdk/Linux_aarch64/24.3/compilers/bin/nvc++",
				"CMAKE_BUILD_TYPE": "Release",
				"CMAKE_CXX_FLAGS": "-mcpu=native",
				"CMAKE_C_FLAGS": "-mcpu=native",
				"GGML_LLAMAFILE": "OFF",
                "GGML_CUDA": "ON"
			}
		},
		{
			"name": "debug",
			"displayName": "debug",
			"binaryDir": "${sourceDir}/out/build/${presetName}",
			"cacheVariables": {
				"CMAKE_INSTALL_PREFIX": "${sourceDir}/out/install/${presetName}",
				"CMAKE_C_COMPILER": "/usr/bin/aarch64-linux-gnu-gcc",
				"CMAKE_CXX_COMPILER": "/usr/bin/aarch64-linux-gnu-g++",
				"CMAKE_BUILD_TYPE": "Debug",
				"GGML_LLAMAFILE": "OFF"	
			}
		}
	],
	"buildPresets": [
		{
		  "name": "default",
		  "configurePreset": "default",
		  "jobs": 8
		}
	  ]
}