3DCNN type model tensorrt inference dynamic batch support correlation

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

Description

When I quantize a 3DCNN type model with tensorrt7,2&8.2, and use dynamic batching, the output is all 0

And it prompts an error: [TensorRT] ERROR: Parameter check failed at: engine.cpp::enqueue::451, condition: bindings != nullptr

I would like to know if tensorrt currently supports dynamic batch for 3DCNN type networks?
if support, how do i do it, thanks!

Environment

TensorRT Version: 7.2&8.2
GPU Type: 3080ti

#2

Hi,
Below link might help you with your query, Kindly check below link for all 3d support layers:

https://docs.nvidia.com/deeplearning/tensorrt/archives/tensorrt-722/developer-guide/index.html#layers

Thanks!

#4

Hi,

Could you please check if you forgot to call TensorRT: nvinfer1::IExecutionContext Class Reference.

Thank you.