Any one tell me why need Addinput when build engine

I am using tensorrt API to build network layer
just wonder why need to define a Input Tensor at first

Could anyone tell me? Thanks

Hi,
Please refer to below links related custom plugin implementation and sample:
https://github.com/NVIDIA/TensorRT/tree/master/samples/opensource/sampleOnnxMnistCoordConvAC

Thanks!

just found out a saying in developer guide
“In addition, the build phase also runs layers on dummy data to select the fastest from its kernel catalog and performs weight pre-formatting and memory optimization where appropriate.”

I guess it’s it.

Hi @1054399357,

We believe all DL frameworks need to define input tensor.
For your reference,

Thank you.