Onnx to TensorRT conversion

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

Description

The issue I am facing is that when I try to convert one ONNX file, named “AP.onnx,” it is successfully converted to TensorRT format. However, when I attempt to convert another RGB.onnx file, the conversion process fails.To address the issue, I have attached both the RGB.onnx file, AP.onnx,and attached error message as error.txt at the end.
The error says that fp16 is not configured in the builder

Environment

TensorRT Version: 8.4.1
GPU Type: NVIDIA GeForce RTX 3050
Nvidia Driver Version: 532.03
CUDA Version: 11.7
CUDNN Version: 8.5
Operating System + Version: Ubuntu 20.04 in wsl2 on windows 11
Python Version (if applicable): 3.8.10
Error.txt (726 Bytes)
test.py (2.2 KB)

Steps To Reproduce

  1. Run the test.py file using python3

AP.onnx (1.3 MB)
RGB_9d9bb888-6342-466a-8705-cf3e7e99677e_2021-07-08_21-37-00.onnx (12.9 MB)

5

Hi,

We recommend you to use the latest TensorRT version 8.6.
Using the latest version, we could successfully build the engine.

[06/21/2023-12:00:08] [I]
&&&& PASSED TensorRT.trtexec [TensorRT v8601] # trtexec --onnx=RGB.onnx --verbose --workspace=20000

Thank you.