GPU compatibility with tensorflow installation

0
votes

New in the field of Deep Learning.

I need to install Tensorflow with GPU support. Before I purchase GPU I need to know like GPU is compatible with Tensorflow or not. In the Tensorflow installation page, with option Tensorflow with GPU below are the software requirements.

The following NVIDIA® software must be installed on your system:

  1. NVIDIA® GPU drivers —CUDA 9.0 requires 384.x or higher. CUDA®
  2. Toolkit —TensorFlow supports CUDA 9.0. CUPTI ships with the CUDA
  3. Toolkit. cuDNN SDK (>= 7.2)
  4. (Optional) NCCL 2.2 for multiple GPU support.
  5. (Optional) TensorRT 4.0 to improve latency and throughput for inference on some models.

If I go with brands available in market other than Nvidia, like Gigabyte, Asus etc lets say AORUS GeForce® GTX 1080 Ti Xtreme Edition 11G , will it allow me to proceed with Tensorflow or not ?

1
tensorflowgpu

1 Answers

0
votes

For compatibility with Nvidia drivers Visit website: https://www.nvidia.com/object/unix.html Click on driver, which you are interested in. In your case it is 390.87 and 410.73 click on SUPPORTED PRODUCTS. If there is your GPU you are ok.

In your case: yes you can use latest version of tensorflow.

Installation can be little tricky, but there are lots of how on the internet to install Nvidia driver, Cuda, cuDNN, NCCL and so on