We often use gpu in tensorflow, which can save our lots of time. Your computer may have have installed gpu, however, do you know the gpu is suitable for tensorflow in your computer? In this tutorial, we will tell you how to check.
Preliminary
You should be sure that a gpu has been installed in your computer. As to win10, to check a gpu is installed or not, you can read this guide.
Best Practice to Check What GPU is Installed in Windows 10: A Simple Guide
Check your gpu is CUDA-enabled
As to NVIDIA gpu, you can check your gpu is cuda-enabled or not in this page:
https://developer.nvidia.com/cuda-gpus
As to us, our gpu is GeForce GT 625M, which is cuda-enabled.
Check the compute capability of gpu
As to tensorflow, if you plan to use gpu for tensorflow, it need compute capability of gpu is 3.5 or higher. However, as to GeForce GT 625M, the compute capability is 2.1, which means this gpu is not suitable for tensorflow.
As to some popular gpu, there are compute capability list
| GPU | Compute Capability |
| NVIDIA TITAN RTX | 7.5 |
| Geforce RTX 2080 Ti | 7.5 |
| Geforce RTX 2080 | 7.5 |
| GeForce GTX 1080 Ti | 6.1 |