search

NVIDIA-AI-IOT / CUDA-PointPillars

A project demonstrating how to use CUDA-PointPillars to deal with cloud points data from lidar.
Apache License 2.0
527 stars 155 forks source link

Support for older TRT 5.1.4 #49

Open lazatsoc opened 2 years ago

lazatsoc commented 2 years ago

Hi, would it be possible to run the model on an older version of TRT/CUDNN/CUDA? We are using the DRIVE AGX with Drive Software 10.0 with TRT 5.1.4. Even the latest Drive SDK does not provide TRT 8.4.0. so it seems like a problem right now. If it can be done, can you please provide some instructions on how to do this? Thanks.

lazatsoc commented 2 years ago

I have managed to compile the demo with TRT 6.5 without too much trouble. The only modification needed is described in https://github.com/NVIDIA-AI-IOT/CUDA-PointPillars/issues/33#issuecomment-1111113025

However, compilation with the even older TRT 5.1 seems a bit trickier. Specifically, following the same procedure produces the following error:

`CUDA-PointPillars/include/pointpillar.h:23:10: fatal error: NvInferRuntime.h: No such file or directory

include "NvInferRuntime.h"`

Indeed, the include directory of TRT 5.1 does not include a NvInferRuntime.h as it is shown here:

https://github.com/NVIDIA/TensorRT/tree/release/5.1/include

But it was introduced in TRT 6.0

https://github.com/NVIDIA/TensorRT/tree/release/6.0/include

Could you please provide a workaround to this issue, if there is one?

Thanks.