No CuDNN support for this activation type

AI & Data Science Deep Learning (Training & Inference) TensorRT
1

Description

I converted a keras model to onnx.
When converting .onnx model to tensorrt, I get the following error:

[04/01/2021-09:00:49] [V] [TRT] --------------- Timing Runner: 2-layer MLP: dense_1 + Add23 -> dense_2 + Add10 (CudnnMLPFC)
    [04/01/2021-09:00:49] [F] [TRT] Assertion failed: No CuDNN support for this activation type
    ../rtExt/cuda/cudaMLPFCRunner.cpp:35
    Aborting...
    [04/01/2021-09:00:49] [V] [TRT] Builder timing cache: created 388 entries, 858 hit(s)
    [04/01/2021-09:00:49] [E] [TRT] ../rtExt/cuda/cudaMLPFCRunner.cpp (35) - Assertion Error in activationTRTToCUDNN: 0 (No CuDNN support for this activation type)
    [04/01/2021-09:00:49] [E] Engine creation failed
    [04/01/2021-09:00:49] [E] Engine set up failed

The model has elu activation but I have converted other models containing elu without any difficulty.

Environment

latest docker image nvcr.io/nvidia/tensorrt:21.03-py3

TensorRT Version: 7.2.2.3
GPU Type: GTX 1050
Nvidia Driver Version: 460.39
CUDA Version: 11.2
CUDNN Version:
Operating System + Version:
Python Version (if applicable):
TensorFlow Version (if applicable):
PyTorch Version (if applicable):
Baremetal or Container (if container which image + tag):

Relevant Files

Steps To Reproduce

trtexec --explicitBatch --onnx=model.onnx --batch=1 --saveEngine=model.trt

3

Hi @mankaran32 ,
Can you please share your ONNX model and script so that we can try it from our end?

Thanks!

4

Hi, you can find the model here

and I use

trtexec --explicitBatch --onnx=model.onnx --batch=1 --saveEngine=model.trt

to make the engine

5

Hi @mankaran32,

Thank you for sharing the model and command.
We could reproduce the error.

Please allow us some time to work on this.

Thank you.