quantum icon indicating copy to clipboard operation
quantum copied to clipboard

Does `tensorflow-quantum` supports multiple GPU?

Open Shuhul24 opened this issue 2 years ago • 1 comments

I am using TFQ for executing of quantum neural networks, but as I need it to be faster, I am trying to run it on multiple GPUs. Has tensorflow-quantum team did something regarding parallel computing for simulations of quantum circuits? I have looked into this issue after googling and following are the issues links that I came up with: https://stackoverflow.com/questions/73250190/does-tensorflow-quantum-support-gpu-and-if-so-how-do-i-make-it-use-mine https://github.com/tensorflow/quantum/issues/599 https://github.com/tensorflow/quantum/issues/193 Some of the issues on these are open? Is there any update regarding this that we can compute parallely using tf.distribute.MirroredStrategy()?

Also adding the link of this issue (https://github.com/tensorflow/tensorflow/issues/52988) as tensorflow==2.7.0 is one of the requisites for tensorflow-quantum.

Shuhul24 avatar Dec 08 '22 11:12 Shuhul24

I don't think GPU ended up being added at all (see the open PR: https://github.com/tensorflow/quantum/pull/687). That being said, if you want to accelerate your code across multiple nodes, you can do that. Here is a detailed tutorial on it: https://github.com/tensorflow/quantum/tree/research/qcnn_multiworker

lockwo avatar Dec 08 '22 16:12 lockwo