Hands-On ROS for Robotics Programming

Achieving better performance using the GPU

Alternatively, you could use the GPU version of TensorFlow to take advantage of this hardware on your laptop. The GPU (short for Graphical Processing Unit) card of your laptop is primarily used to power the display output on the screen. Therefore, it is very good at image processing.

As the kinds of calculations we need to do in ML are very similar (that is, floating-point, vector, and matrix operations), you can speed up the training and usage of your ML models by using the GPU for calculations instead of the CPU.

By using the GPU, you may achieve at least a factor of 10 in speed calculation for using the CPU, even in the case of the cheapest GPU cards. Hence, the choice of GPU is worth it. The ...