tensorrt failing inference on gt730 gpu

Tensorrt version: 6.0.1

Platform: windows

I am trying to do inference using tensorrt for a unet model in windows, it works well with gtx 1080, gtx 1050 and p2000 series but for gt 730 series it fails when trying to execute the context after creating the engine. Any help will be much appreciated.

Hi,

Could you please share the error details and reproduction steps? If possible, please share the model & script file as well.

Also, please refer below link to check the support matrix and CUDA compatibility of your GPU:
https://docs.nvidia.com/deeplearning/sdk/tensorrt-archived/tensorrt-700/tensorrt-support-matrix/index.html#hardware-precision-matrix

Which GT 730 series GPU model you are using in this case?
GeForce GT 730 —> 3.5
GeForce GT 730M —> 3.0
GeForce GT 730 DDR3,128bit —> 2.1

Thanks

My gpu is a GT730 3.5. There are no error logs it just returns a False for context execute.

Hi,

Could you please share the model and script file so we can better help?

Also, please provide details on the platforms you are using:
o Linux/Windows version
o GPU type
o Nvidia driver version
o CUDA version
o CUDNN version
o Python version [if using python]
o Tensorflow version
o TensorRT version

Thanks

Hi I can’t provide you the code as its confidential, although I have attached the unet model alongwith.

Windows version: windows 10 pro
Gpu: gt730 3.5
nvidia driver version: 385.54
cuda version: 9.0
cudnn: 7.6.3

I am using cpp and uff for inferencing
tensorflow version: 10.0
tensorrt version: 6.0.1

Link to model: [model link][https://drive.google.com/file/d/1ET_ALJUQcTcC4a_rtjWypaNQRIZIOSiS/view?usp=sharing]

Hi,
Just wanted to check one thing.
Was the engine generated on one GPU and run on another?
Since TRT generated engines are not portable across all GPU types, must be generated specifically for the target GPU.

Thanks

Hi,

The engine file was generated on the gpu.

Is there any update on the problem?

Hi,

Sorry for the late reply, this issue is currently not reproducible at our end.
Could you please share the sample script file to reproduce this issue?

Thanks