win10系统上LibTorch的安装和使用（cuda10.1版本）

*Major*

已于 2022-09-11 00:05:38 修改

阅读量6.3k

点赞数 8

文章标签：深度学习 pytorch c++

于 2020-12-03 17:38:40 首次发布

本文链接：https://blog.csdn.net/qq_41375318/article/details/110556350

版权

https://www.bilibili.com/video/BV14v411b733#reply3775162098

vs2019中配置使用cuda问题

win10系统上LibTorch的安装和使用（cuda10.1版本）

链接：libtorch_cuda10.1_win10
提取码：1234

链接：cuda10.1 + cudnn10.1
提取码：1234

$L ib T orc h 的安装和使用$

https://pytorch.org/get-started/locally/
LibTorch就在torch的安装界面，Package选择LibTorch就行
在这里插入图片描述

首先，VS2019，编译的时候记得选x64，坑爹玩意，害我晚上11.09才搞完，一晚上没了。

在这里插入图片描述

— 在这里插入图片描述

E:\BaiduNetdiskDownload\libtorch\include
E:\BaiduNetdiskDownload\libtorch\include\torch\csrc\api\include

在这里插入图片描述

E:\BaiduNetdiskDownload\libtorch\lib

在这里插入图片描述

配置libtorch GPU VS2019,添加的lib依赖

GPU版

torch.lib
torch_cuda.lib
caffe2_detectron_ops_gpu.lib
caffe2_module_test_dynamic.lib
torch_cpu.lib
c10_cuda.lib
caffe2_nvrtc.lib
mkldnn.lib
c10.lib
dnnl.lib
libprotoc.lib
libprotobuf.lib
libprotobuf-lite.lib
fbgemm.lib
asmjit.lib
cpuinfo.lib
clog.lib

1.7.0-GPU

asmjit.lib
c10.lib
c10d.lib
caffe2_module_test_dynamic.lib
caffe2_detectron_ops_gpu.lib
caffe2_nvrtc.lib
clog.lib
cpuinfo.lib
dnnl.lib
fbgemm.lib
gloo.lib
gloo_cuda.lib
mkldnn.lib
torch.lib
torch_cpu.lib
torch_cuda.lib
c10_cuda.lib
libprotoc.lib
libprotobuf.lib
libprotobuf-lite.lib

deps-debug-1.12.1+cu113

asmjit.lib
c10.lib
c10_cuda.lib
caffe2_nvrtc.lib
clog.lib
cpuinfo.lib
dnnl.lib
fbgemm.lib
kineto.lib
XNNPACK.lib
torch.lib
torch_cpu.lib
torch_cuda.lib
torch_cuda_cpp.lib
torch_cuda_cu.lib
libprotocd.lib
libprotobufd.lib
libprotobuf-lited.lib
pthreadpool.lib

CPU版

asmjit.lib
c10.lib
c10d.lib
caffe2_detectron_ops.lib
caffe2_module_test_dynamic.lib
clog.lib
cpuinfo.lib
dnnl.lib
fbgemm.lib
gloo.lib
libprotobufd.lib
libprotobuf-lited.lib
libprotocd.lib
mkldnn.lib
torch.lib
torch_cpu.lib

设置dll文件的环境变量，
在这里插入图片描述

PATH=E:\BaiduNetdiskDownload\libtorch\lib;%PATH%

#include <iostream>
#include <torch/torch.h>

int main()
{

    torch::Tensor tensor = torch::rand({ 5,3 });
    std::cout << tensor << std::endl;

    return EXIT_SUCCESS;
}

在这里插入图片描述

#include <torch/script.h>  // One-stop header.
#include <iostream>
#include <memory>

int main() {
  using torch::jit::script::Module;
  Module module =
      torch::jit::load("UNet_model.pt");

  std::cout << "ok\n";
  // Create a vector of inputs.
  std::vector<torch::jit::IValue> inputs;
  inputs.push_back(torch::ones({1, 3, 512, 512}));

  // Execute the model and turn its output into a tensor.
  at::Tensor output = module.forward(inputs).toTensor();

  std::cout << output << '\n';
    
}

#include <torch/script.h>  // One-stop header.
#include <iostream>
#include <memory>

int main() {
	// Deserialize the ScriptModule from a file using torch::jit::load().
	//std::shared_ptr<torch::jit::script::Module> module =
	//          torch::jit::load("E:/HM_DL/torch_test/traced_resnet_model.pt");

	using torch::jit::script::Module;
	Module module =
		torch::jit::load("C:/Users/major/Desktop/AlexNet_model.pt");

	std::cout << "model load successfully\n";
	// Create a vector of inputs.
	std::vector<torch::jit::IValue> inputs;
	inputs.push_back(torch::ones({ 1, 3, 224, 224 }));

	// Execute the model and turn its output into a tensor.
	at::Tensor output = module.forward(inputs).toTensor();

	std::cout << output << '\n';
	std::cout << "输出结束\n";

		;
}