jinxianwei / onnxruntime-gpu-cpp-dummy-demo

simplest demo of onnxruntime gpu in c/cpp

onnxruntime gpu cpp dummy demo

winxos 20240123 验证：windows10 20240624 平台：cuda12.0，onnxruntime-win-x64-gpu-1.17.1 visual studio 说明：只能用vs的MSVC进行编程，不能用mingw进行编译，因为onnxruntime没有对mingw的支持，pytorch的libtorch也放弃了mingw的支持

this is the simplest way to use onnxruntime gpu in cpp

1. env prepare

!!! you need to install the right version of CUDA and CUDNN

you can found from nvidia site

2. create cpp console proj in vs

3. install onnxruntime-gpu

you can install package using vcpkg, or nuget

nuget is more convenient. only need onnxruntime-gpu.

4. clone this repo

5. replace const name in the code

includes:

• model path
• input node name
• output node name
• image path

6. have fun.