[Build] [Linux] [.NET] Microsoft.ML.OnnxRuntime.Gpu keep publishing unnecessary DLL large files for linux-x64

[jiaxuwu2021]

Describe the issue

Nuget package Microsoft.ML.OnnxRuntime.Gpu 1.20.1 keep publishing DLL files for linux-x64, especially onnxruntime_providers_cuda.dll which is very large (>600MB), these DLL files are unnecessary to linux

Please fix to stop publishing these onnxruntime*.dll files for linux-x64 when dotnet build/dotnet publish, otherwise the final binaries size will be double (both cuda.so and cuda.dll >600MB)

---

One possible clue: it seems that this file Microsoft.ML.OnnxRuntime.Gpu.Windows.props is trying to copy DLL files for all x64 platforms including linux-x64.

Related issue: microsoft/onnxruntime-genai#1135

Urgency

No response

Target platform

linux-x64

Build script

build linux-x64 under linux command line:
dotnet publish test.csproj --runtime linux-x64

• test.csproj

<Project Sdk="Microsoft.NET.Sdk.Web">

  <PropertyGroup>
    <OutputType>Exe</OutputType>
    <TargetFramework>net8</TargetFramework>
    <ImplicitUsings>enable</ImplicitUsings>
    <Nullable>enable</Nullable>
    <PublishAot>true</PublishAot>
    <IsPackable>false</IsPackable>
    <SelfContained>true</SelfContained>
  </PropertyGroup>

  <!-- OnnxRuntime CUDA https://onnxruntime.ai/docs/get-started/with-csharp.html#builds -->
  <ItemGroup>
    <PackageReference Include="Microsoft.ML.OnnxRuntime.Gpu" Version="1.20.1" />
  </ItemGroup>
</Project>

• Program.cs

namespace MyTest;

using Microsoft.ML.OnnxRuntime;

public class Program
{
    public static async Task Main(string[] args)
    {
        const string modelPath = "/home/jiaxuwu/.aitk/models/onnx/cuda/mistral-7b-instruct-v0.2-cuda-int4-rtn-block-32/mistral-7b-instruct-v0.2-cuda-int4-rtn-block-32.onnx";
        using var session = new InferenceSession(modelPath);
    }
}

Error / output

There are many DLL files in the final binaries

$ ll -h bin/Release/net8/linux-x64/publish/onnxruntime*.dll
...  12M Nov 19 23:09 bin/Release/net8/linux-x64/publish/onnxruntime.dll*
... 643M Nov 19 23:10 bin/Release/net8/linux-x64/publish/onnxruntime_providers_cuda.dll*
...  22K Nov 19 23:10 bin/Release/net8/linux-x64/publish/onnxruntime_providers_shared.dll*
... 744K Nov 19 23:09 bin/Release/net8/linux-x64/publish/onnxruntime_providers_tensorrt.dll*

$ ll -h bin/Release/net8/linux-x64/onnxruntime*.dll
...  12M Nov 19 23:09 bin/Release/net8/linux-x64/onnxruntime.dll*
... 643M Nov 19 23:10 bin/Release/net8/linux-x64/onnxruntime_providers_cuda.dll*
...  22K Nov 19 23:10 bin/Release/net8/linux-x64/onnxruntime_providers_shared.dll*
... 744K Nov 19 23:09 bin/Release/net8/linux-x64/onnxruntime_providers_tensorrt.dll*

Visual Studio Version

No response

GCC / Compiler Version

No response

[jiaxuwu2021]

ep:CUDA

[github-actions]

This issue has been automatically marked as stale due to inactivity and will be closed in 30 days if no further activity occurs. If further support is needed, please provide an update and/or more details.