GPU Compute Capability 3.0 but the minimum required Cuda capability is 3.5

2
votes

I tried to install tensorflow-gpu. When I checked the installation using this command

import tensorflow as tf
tf.test.is_gpu_available()

The result was

 Ignoring visible gpu device (device: 0, name: Quadro K5100M, pci bus id: 0000:01:00.0, compute capability: 3.0) with Cuda compute capability 3.0. The minimum required Cuda capability is 3.5. => False
  1. Operating System = Windows 10
  2. GPU type: Nvidia Quadro k5100m
  3. GPU driver version = 426.78
  4. python = 3.7.7
  5. Cuda Toolkit = 10.1
  6. cuDNN = 7.6.5
  7. TensorRT = 6.0
  8. tensorflow-gpu = 2.1.0

How can I solve this problem?

1
pythonwindowstensorflowanacondagpu

1 Answers

0
votes

The Tensorflow Documentation states

Only the NVIDIA® GPU card with CUDA® Compute Capability 3.5 or higher are supported for Tensorflow-GPU.

As the Error Message suggests, your Graphics Card (Nvidia Quadro k5100m) has Compute Capability 3.0 and the Minimum Version supported is 3.5.

Solution is that you may have to upgrade your Graphics Card to any of these GPUs supported by Tensorflow.

Mentioning the List of Quadro GPUs supported by Tensorflow below for your reference:

enter image description here