search

microsoft / onnxruntime

ONNX Runtime: cross-platform, high performance ML inferencing and training accelerator
https://onnxruntime.ai
MIT License
14.68k stars 2.93k forks source link

Not support rtx 3000 series #6213

Open Mixelon-tera opened 3 years ago

Mixelon-tera commented 3 years ago

It seems onnxruntime 1.6.0 can not use with RTX 3000 series. when I run the model it shows the error "not support and onnxruntime::CUDAExecutionProvider::GetCapability] CUDA kernel not supported".

Can anyone run onnxruntime with the RTX 3000 series?

wschin commented 3 years ago

Is it a warning or an error? I saw

    if (!force_inside && not_supported) {
      if (not_supported) {
        LOGS_DEFAULT(WARNING) << "CUDA kernel not supported. Fallback to CPU execution provider for Op type: " << node.OpType() << " node name: " << node.Name();
      }
    } else {
      candidates.push_back(node.Index());
    }

in cuda_execution_provider.cc. Could you paste full log of your run?