Bug in TensorRT conv layer

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

Hi,

I spotted a bug in TensorRT 3 / 4 conv algo 14 tactic -2535759802710599445.
The results are diverging from caffe if you set non null padding and this tactic is chosen, leading to invalid results.
The other tactics are OK AFAIK.

Is it possible to disable this very tactic for further testing?

Tested with :

Ubuntu 16.04
GTX 1080
GTX 1080 TI
TensorRT 3.0.4
TensorRT 4 RC
CuDNN 7.1.3
Cuda 8
Driver v. 367.48
Driver v. 381.22

mbd.

2

A few things that you can do to help us help you with this issue:

Please file a bug. You can do this through your developer.nvidia.com account.

Please include a detailed description of the network you are using, ideally in the form of code or a model file (caffe, UFF, ONNX, etc…)