Could not find any implementation for node after updating to the trt 8x from trt 7x

[akhoroshev]

Description

Convert via python api (also via trtexec) failed. All works fine with the trt 7.2.3. But after updating to the latest trt 8.2.3.0 i encountered the following error:

[01/28/2022-07:54:29] [TRT] [E] 10: [optimizer.cpp::computeCosts::2011] Error Code 10: Internal Error (Could not find any implementation for node {ForeignNode[113 + (Unnamed Layer* 72) [Shuffle]...(Unnamed Layer* 171) [Shuffle]]}.)

Increasing workpace does not help. My model does not have dynamic inputs.

Environment

TensorRT Version: 8.2.3.0
NVIDIA GPU: v100
NVIDIA Driver Version: 470.57.02
CUDA Version: release 11.4, V11.4.120
CUDNN Version: 8.1.1
Operating System: Red Hat Enterprise Linux Server release 7.9
Python Version (if applicable): 3.6.8

[zerollzeng]

Can you share the onnx model?

[nvpohanh]

closing for now due to >14 days with no response. Please feel free to reopen if the issue still exists. Thanks