cudnn error when using Tesla T4

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

platforms
CentOS Linux release 7.5.1804 (Core)
GPU type:Tesla T4
Nvidia driver version:418.67
CUDA version:9.0.176
CUDNN version:7.6.3.30
Python version: None
Tensorflow version:None
TensorRT version:TensorRT-6.0.1.5.CentOS-7.6.x86_64-gnu.cuda-9.0.cudnn7.6

1.when I convert caffe model to TensorRT model on Tesla T4 using trtexec,an error about cudnn occured：
“ERROR: …/rtSafe/cuda/cudaConvolutionRunner.cpp (303) - Cudnn Error in execute: 8 (CUDNN_STATUS_EXECUTION_FAILED)”

  1. I test the same version of TensorRT on Tesla T4,that’s OK.

please help me,thanks.

2

Hi,

Could you please share your script and model file so we can better help?

Meanwhile, please try CUDA10 and CUDA10 versions of CuDNN etc.
Please refer to cuDNN support matrix:
https://docs.nvidia.com/deeplearning/sdk/cudnn-support-matrix/index.html#cudnn-cuda-hardware-versions

Thanks