TensorFrames should be able to run tasks concurrently on separate GPUs, if they are available. This should be possible with the latest TensorFlow java API:
https://www.tensorflow.org/api_docs/java/reference/org/tensorflow/Session.html#Session(org.tensorflow.Graph, byte[])
Assuming that all the GPUs are available for running tasks, they can be locked onto specific sessions in python by using the following code:
config = tf.ConfigProto()
config.gpu_options.visible_device_list = "0" # Or different values
session = tf.Session(config=config)
TensorFrames should be able to run tasks concurrently on separate GPUs, if they are available. This should be possible with the latest TensorFlow java API:
https://www.tensorflow.org/api_docs/java/reference/org/tensorflow/Session.html#Session(org.tensorflow.Graph, byte[])
Assuming that all the GPUs are available for running tasks, they can be locked onto specific sessions in python by using the following code: