M1 Apple silicon GPU for node-js

Hello guys, I’m newbie in ML and looking for a way to use tensorflow-metal within node-js. I’ve tried tensorflow-metal with python and it works like a charm. I can’t find any info about it integration with tfjs-node-gpu package. Is there plans to implement this feature?

#help_request #apple_silicon #node-js #gpu

1 Like

Hi, @Elma_Ranin

Welcome to TensorFlow Forum!

I apologize for the delayed response and as far my current understanding TensorFlow.js with tfjs-node-gpu this combination brings TensorFlow.js with WebGL support to Node.js enabling GPU acceleration for web machine learning models while it doesn’t directly access the M1 GPU’s full Metal capabilities it provides WebGL-based GPU acceleration for compatible models.This approach is generally easier to use but might not deliver the same performance.

TensorFlow.js Node.js GPU support heavily relies on the underlying TensorFlow C++ library which in turn relies on TensorFlow support for Apple Silicon. If TensorFlow adds support for M1 chips and GPU acceleration, TensorFlow.js might inherit this support in the future and as per our official documentation
There is currently no official GPU support for MacOS.

Just let me know, If I have missed something. Thank you for your understanding and patience.