Running Tensorflow 1.15 model in RTX A5000 GPUS (Ampere architecture)

Description

I am planning to buy Nvidia RTX A5000 GPU for training models. However i am concerned if i will be able to run tensorflow 1.15 model in this GPU.
I have read that Ampere architecture only supports nvidia-driver versions above 450.36.06+ and cuda versions CUDA 11. Since tensorflow 1.15 requires cuda 10, I am not sure if I can run such models.

Ref link: CUDA Compatibility :: GPU Deployment and Management Documentation

My colleague has brought an RTX 3090 (Ampere Technology) and has mentioned he is not able to run Tensorflow 1.15 on this GPU.

Also will using any nvidia docker with cuda 10 help me train these models?
I had created a nvidia-docker environment for tensorflow 1.15, in an RTX 2080 TI(Turing architecture machine) with Nvidia driver 460.91.03 and CUDA Version: 11.2 which works fine.

Thank you in advance.

Environment

Linux OS environment.

Hi,
We recommend you to check the below samples links in case of tf-trt integration issues.
https://docs.nvidia.com/deeplearning/frameworks/tf-trt-user-guide/index.html#samples
https://docs.nvidia.com/deeplearning/tensorrt/quick-start-guide/index.html#framework-integration
https://docs.nvidia.com/deeplearning/frameworks/tf-trt-user-guide/index.html#integrate-ovr
https://docs.nvidia.com/deeplearning/frameworks/tf-trt-user-guide/index.html#usingtftrt

If issue persist, We recommend you to reach out to Tensorflow forum.
Thanks!

Hi,
The issue is not regarding Tensor RT integrations.

I wanted to confirm if the NVIDA RTX A5000 (Ampere) will support Cuda 10.
I have doubt if the new Ampere technology cards will support Cuda 10, since it is mentioned they support for cuda versions above 11.

Cuda 10 is required, since i want to run Tensorflow 1.15 model in this GPU.


Thank you.

Hi @tharun.ts ,
This forum talks about TRT related updates and issues.
I would suggest you to reach out to CUDA Forum for better assistance.

Thanks!