IPluginV2 called during build phase and building failed

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

I rewrote my old plugins from IPlugin class to IPluginV2 class so I can upgrade my tensorRT7 to tensorRT8.

When I was running the usual pipeline (file parsing → engine building → inference), the process ended during building.

After a bit more investigation, I found that some custom plugins were called during building which did not happen in tensorRT7.

Can this be the reason why building failed? Is this normal for IPluginV2? Where do I start to debug?

mainboard.log.INFO.20221017-160406.3087681 (1.9 MB)

Jetpack 5.0
TRT: 8.4.0.9
GPU: Orin
CUDA: 11.4.166

2

Hi,

Looks like this post is a duplicate of [pluginV2Runner.cpp::execute::265] Error Code 2: Internal Error (Assertion status == kSTATUS_SCUESS failed. ).

Thank you.

5

Did you ever figure this out? I’m trying to do the same thing with the same error.

6

Instead of using c++ API, it’s easier to convert model to onnx format first. I got around this issue by that.

1 Like
7

so you converted the caffe model into onnx first and then loaded it into tensorRT? I am also using apollo.