conan入门(三十):对腾讯ncnn进行Conan封装

10km

已于 2023-12-29 18:08:47 修改

阅读量302

点赞数

分类专栏： conan deeplearning 文章标签： conan ncnn 腾讯 cmake

于 2023-09-26 14:06:20 首次发布

本文链接：https://blog.csdn.net/10km/article/details/133307079

版权

conan 同时被 2 个专栏收录

39 篇文章 98 订阅

订阅专栏

deeplearning

26 篇文章 1 订阅

订阅专栏

ncnn是腾讯开源的一个为手机端极致优化的高性能神经网络前向计算框架。因为项目中需要用到它，所以我需要对它进行conan封装，以支持我们基于conan管理的项目的引用。
ncnn代码比较成熟，项目结构清晰，对于conan封装的主要工作量就是将CMakeLists.txt脚本的option命令定义的开关在conanfile.py的options中定义为对应的conan 选项开关.

conanfile.py for ncnn 20200916

conanfile.py – 基于 ncnn的20200916版本实现,由于ncnn不同的版本，CMakeLists.txt中定义的option会有所不同，所以，如果你需要实现其他ncnn版本的conan封装，可以在此文件基础上对照对应版本的CMakeLists.txt的option定义对python代码进行增减

from conan import ConanFile
from conan.tools.cmake import CMake, CMakeDeps, CMakeToolchain
from conan.tools.env import VirtualBuildEnv
# from conan.tools.microsoft import is_msvc, is_msvc_static_runtime

class NcnnConan(ConanFile):
    name = "ncnn"
    version = "20200916"
    # Optional metadata
    url = "https://github.com/Tencent/ncnn"
    description = "a high-performance neural network inference computing framework optimized for mobile platforms"
    topics = ("deep learning","ai","ncnn")

    #requires = []
    tool_requires = "cmake/[>=3.15.7]"
    package_type = "static-library"
    # Binary configuration
    settings = "os", "compiler", "build_type", "arch"
    options = {
        "openmp": [True, False],
        "stdio": [True, False],
        "string": [True, False],
        "opencv": [True, False],
        "simplestl": [True, False],
        "threads": [True, False],
        "benchmark": [True, False],
        "pixel": [True, False],
        "pixel_rotate": [True, False],
        "vulkan": [True, False],
        "vulkan_online_spirv": [True, False],
        "system_glslang": [True, False],
        "requant": [True, False],
        "runtime_cpu": [True, False],
        "fPIC": [True, False],
        "build_tests": [True, False],
        "coverage": [True, False],
        "build_benchmark": [True, False],
        # "disable_rtti": [True, False],
        # "build_tools": [True, False],
        # "build_examples": [True, False],
        # "disable_exception": [True, False],
        # "arm82": [True, False],
        # "avx2": [True, False],
        # "build_external": [True, False],
        # "enable_spvremapper": [True, False],
        # "enable_glslang_binaries": [True, False],
        # "enable_hlsl": [True, False],
        # "enable_rtti": [True, False],
        # "enable_exceptions": [True, False],
        # "enable_opt": [True, False],
        # "enable_pch": [True, False],
        # "enable_ctest": [True, False],

        }
    default_options = {
        "openmp": True,
        "stdio": True,
        "string": True,
        "opencv": False,
        "simplestl": False,
        "threads": True,
        "benchmark": False,
        "pixel": True,
        "pixel_rotate": True,
        "vulkan": False,
        "vulkan_online_spirv": True,
        "system_glslang": False,
        "requant": False,
        "runtime_cpu": True,
        "fPIC": True,
        "build_tests": False,
        "coverage": False,
        "build_benchmark": True,
        # "disable_rtti": False,
        # "build_tools": False,
        # "build_examples": False,
        # "disable_exception": False,
        # "arm82": False,
        # "avx2": True,
        # "build_external": False,
        # "enable_spvremapper": False,
        # "enable_glslang_binaries": False,
        # "enable_hlsl": False,
        # "enable_rtti": False,
        # "enable_exceptions": False,
        # "enable_opt": False,
        # "enable_pch": False,
        # "enable_ctest": False,

        }
    # Sources are located in the same place as this recipe, copy them to the recipe
    exports_sources = "CMakeLists.txt",  "*/*"
    def generate(self):
        tc = CMakeToolchain(self)

        tc.variables["NCNN_OPENMP"] = self.options.openmp
        tc.variables["NCNN_STDIO"] = self.options.stdio
        tc.variables["NCNN_STRING"] = self.options.string
        tc.variables["NCNN_INSTALL_SDK"] = True
        tc.variables["NCNN_OPENCV"] = self.options.opencv
        tc.variables["NCNN_SIMPLESTL"] = self.options.simplestl
        tc.variables["NCNN_THREADS"] = self.options.threads
        tc.variables["NCNN_BENCHMARK"] = self.options.benchmark
        tc.variables["NCNN_PIXEL"] = self.options.pixel
        tc.variables["NCNN_PIXEL_ROTATE"] = self.options.pixel_rotate
        tc.variables["NCNN_CMAKE_VERBOSE"] = False
        tc.variables["NCNN_VULKAN"] = self.options.vulkan
        tc.variables["NCNN_VULKAN_ONLINE_SPIRV"] = self.options.vulkan_online_spirv
        tc.variables["NCNN_SYSTEM_GLSLANG"] = self.options.system_glslang
        tc.variables["NCNN_REQUANT"] = self.options.requant
        tc.variables["NCNN_RUNTIME_CPU"] = self.options.runtime_cpu
        tc.variables["NCNN_DISABLE_PIC"] = not self.options.fPIC
        tc.variables["NCNN_BUILD_TESTS"] = self.options.build_tests
        tc.variables["NCNN_COVERAGE"] = self.options.coverage
        tc.variables["NCNN_BUILD_BENCHMARK"] = self.options.build_benchmark
        
        #tc.variables["NCNN_DISABLE_RTTI"] = self.options.disable_rtti
        tc.variables["NCNN_BUILD_TOOLS"] = False
        tc.variables["NCNN_BUILD_EXAMPLES"] = False
        # tc.variables["NCNN_DISABLE_EXCEPTION"] = self.options.disable_exception
        # tc.variables["NCNN_ARM82"] = self.options.arm82
        # tc.variables["NCNN_AVX2"] = self.options.avx2
        
        # tc.variables["BUILD_EXTERNAL"] = self.options.build_external
        # tc.variables["ENABLE_SPVREMAPPER"] = self.options.enable_spvremapper
        # tc.variables["ENABLE_GLSLANG_BINARIES"] = self.options.enable_glslang_binaries
        # tc.variables["ENABLE_HLSL"] = self.options.enable_hlsl
        # tc.variables["ENABLE_RTTI"] = self.options.enable_rtti
        # tc.variables["ENABLE_EXCEPTIONS"] = self.options.enable_exceptions
        # tc.variables["ENABLE_OPT"] = self.options.enable_opt
        # tc.variables["ENABLE_PCH"] = self.options.enable_pch
        # tc.variables["ENABLE_CTEST"] = self.options.enable_ctest
        # tc.variables["CMAKE_MODULE_PATH"] = self.build_folder
        # if is_msvc(self):
        #     tc.variables["NCNN_BUILD_WITH_STATIC_CRT"] = is_msvc_static_runtime(self)
        # This policy doesn't matter for us, but avoids a warning
        tc.cache_variables["CMAKE_POLICY_DEFAULT_CMP0074"] = "NEW"
        tc.generate()

        cd = CMakeDeps(self)
        cd.generate()

        env = VirtualBuildEnv(self)
        env.generate()

    def build(self):
        cmake = CMake(self)
        cmake.configure()
        cmake.build()

    def package(self):
        cmake = CMake(self)
        cmake.install()

    def package_info(self):
        self.cpp_info.libs = ["ncnn"]

armv8交叉编译profile

~/.conan/profiles/aarch64-linux-gnu

include(default)
{% set target_host = "aarch64-linux-gnu" %}
# 指定交叉编译器安装位置
{% set compiler_install_prefix = "/opt/gcc-arm-8.2-2018.08-x86_64-aarch64-linux-gnu" %}
# 指定交叉编译器bin位置
{% set compiler_prefix = compiler_install_prefix + "/bin/" %}
# 使用添加在PATH的编译器
[settings]
arch=armv8
build_type=Release
compiler=gcc
compiler.libcxx=libstdc++11
compiler.version=8
os=Linux
#[tool_requires]
[options]
opencv/*:with_ffmpeg=False
opencv/*:with_gtk=False
opencv/*:openexr=False
[conf]
tools.cmake.cmaketoolchain:system_processor=aarch64
tools.build:compiler_executables={"c":"{{ compiler_prefix + target_host + "-gcc" }}","cpp":"{{ compiler_prefix + target_host + "-g++" }}","asm":"{{ compiler_prefix + target_host + "-gcc" }}"}
tools.build:sysroot={{ compiler_install_prefix + "/aarch64-linux-gnu/libc" }}
#tools.build:sysroot=/
tools.system.package_manager:mode=install
tools.system.package_manager:sudo=True
[buildenv]
CHOST={{ target_host }}
AR={{ compiler_prefix + target_host + "-ar" }}
AS={{ compiler_prefix + target_host + "-gcc" }}
RANLIB={{ compiler_prefix + target_host + "-ranlib" }}
CC={{ compiler_prefix + target_host + "-gcc" }}
CXX={{ compiler_prefix + target_host + "-g++" }}
FC={{ compiler_prefix + target_host + "-gfortran" }}
LD={{ compiler_prefix + target_host + "-ld" }}
STRIP={{ compiler_prefix + target_host + "-strip" }}

代码仓库

上面的conanfile.py代码我保存在码云仓库：
https://gitee.com/l0km/ncnn.git conan20200916分支

git clone https://gitee.com/l0km/ncnn.git -b conan20200916

本机编译示例

cd ncnn
conan create .

交叉编译

cd ncnn
conan create . -pr:h aarch64-linux-gnu.jinja -pr:b default

conan系列文章

10km

关注

0
点赞
踩
0

收藏

觉得还不错? 一键收藏
打赏
0
评论
conan入门(三十):对腾讯ncnn进行Conan封装

是腾讯开源的一个为手机端极致优化的高性能神经网络前向计算框架。因为项目中需要用到它，所以我需要对它进行conan封装，以支持我们基于conan管理的项目的引用。ncnn代码比较成熟，项目结构清晰，对于conan封装的主要工作量就是将CMakeLists.txt脚本的option命令定义的开关在conanfile.py的options中定义为对应的conan 选项开关.
复制链接

扫一扫