search

louiskirsch / speechT

An opensource speech-to-text software written in tensorflow
Apache License 2.0
157 stars 36 forks source link

Multiple gpu? #10

Open arpit601 opened 7 years ago

arpit601 commented 7 years ago

How to use multiple gpu for this code ?

arpit601 commented 7 years ago

@timediv

louiskirsch commented 7 years ago

This is currently not supported. Because the training times are rather short, I am doubtful that this is a very important feature. But you may explore how this could be applied and whether it makes a big difference! See https://www.tensorflow.org/tutorials/deep_cnn#training_a_model_using_multiple_gpu_cards

arpit601 commented 7 years ago

I have written a multiple gpu code and I have checked that its running well. I will try to change here also.

sruteesh commented 7 years ago

@timediv I wasn't able to run it on a GPU. The code was written using multithreads. How do I use it on GPU. @arpit601 Can u post the updates you made to run it on multiple GPUs

arpit601 commented 6 years ago

@sruteesh Sorry, I dont have the code with me now. But you can follow CIFAR-10 example of multiple gpu usage https://github.com/tensorflow/models/blob/master/tutorials/image/cifar10/cifar10_multi_gpu_train.py You have to use tower concept from the above link and apply it .