CODEWITHSUNDEEP
tensorflowkerasraspberry-pigpuraspberry-pi4
DrRed
DrRed

Reputation: 35

Keras GPU Acceleration on a Raspberry Pi?

I'm fairly sure this is a very stupid question but I can't get it off my brain. So I'm sure that you know that you can use CUDA or ROCm to accelerate learning in TensorFlow/Keras, but I was just wondering if there was any way that a Raspberry Pi 4 with its GPU could help with training?

Upvotes: 0

Views: 544

Answers (2)

Mojtaba Abdi Khassevan
Mojtaba Abdi Khassevan

Reputation: 539

You should train on a computer and test the model. If it worked, save your model weights and structure and deploy it to a RaspberryPie.

Upvotes: 1

rok
rok

Reputation: 2775

I don't know what you mean by "help".. but in the raspberry pi 4 the GPU is a Videocore VI, integrated. It does not support CUDA and you can not use an external GPU (there are not connections dedicated to it). You could only train on CPU but, Raspberry is a resource-limited device, forget about it. You can do inference on raspberry.

Upvotes: 2

Related Questions