how to use tensorrt serving docker image in centos7.4 and nvidia driver 396?

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

hello, I have a question~

there are only three tags of tensorrt serving docker image in [url]https://ngc.nvidia.com/catalog/containers/nvidia%2Ftensorrtserver/tags[/url], which all are for nvidia driver 410 or higher.
My server is centos7.4 and Nvidia driver is NVIDIA-SMI 396.44.

So how can I successfully using tensorrt serving docker image if I do not update my Nvidia driver to 410 or higher. cam you give some advises? thank you very much~

Linux distro and version:

LSB Version:	:core-4.1-amd64:core-4.1-noarch
Distributor ID:	CentOS
Description:	CentOS Linux release 7.4.1708 (Core)
Release:	7.4.1708
Codename:	Core

other envirs:

GPU type: Tesla v100
nvidia driver version: NVIDIA-SMI 396.44
CUDA version: 9.0
CUDNN version: 7.3.0
Python version [if using python]: python2.7
TensorRT version: 5.0.2.6
gcc>5.3/lib64
2

Hello

Depending on the GPU you have, you may want to downgrade to driver 384.x, which will allow the containers to use “compatibility mode”.

Release 384.xx if your hardware is Tesla V100, Tesla P4, Tesla P40, or Tesla P100.

But, in the long term, highly recommend you to update to latest driver.

3

@NVES thanks for your reply, but I can not find the official documents about how to turn on the compatibility mode.
Besides, I successfully install Nvidia driver 410 in my server Centos7.4, so I modify my question.

4

Hello,

compatibility mode is automatically detected and entered by the container entry script. It is not something you can manually trigger.

5

@NVES Wow, I got it. Thank you very much~