When attempting to initialize a new TensorRT model (from caffe model file), I get the following error:
I0611 13:43:01.790940 10628 rt_model_component.cc:23] --------------- Timing C(5)
E0611 13:43:01.794281 10628 rt_model_component.cc:29] cudnnLayerUtils.cpp (355) - Cuda Error in findFastestTactic: 74
E0611 13:43:01.794409 10628 rt_model_component.cc:29] resources.cpp (264) - Cuda Error in ~ScopedCudaEvent: 74
terminate called after throwing an instance of ‘nvinfer1::CudaError’
what(): std::exception
Aborted (core dumped)
I am using TensorRT 3.0.1. The layer itself is below.
layer {
name: “C”
type: “Eltwise”
bottom: “A”
bottom: “B”
top: “C”
eltwise_param {
operation: PROD
}
}
I’m opened to the possibility that this is actually a bug in my implementation, but I find it hard to understand how I would be able to cause an alignment error in the internal implementation. Any help would be appreciated.
Also please let me know if this is the appropriate place to file this issue. I would prefer to file to the bug tracker directly if possible.
Best,
Matt