MirrorStrategy doesn't see GPU - INFO:tensorflow:Not using Distribute Coordinator
Opened this issue · 0 comments
Nicolabo commented
Hi,
Thanks for your clear example. I was trying to replicate your solution to see how MirrorStrategy works.
However once I run the code:
distribution = tf.contrib.distribute.MirroredStrategy()
config = tf.estimator.RunConfig(train_distribute= distribution,
model_dir="/tmp/mnist_convnet_model")
I get a warning:
INFO:tensorflow:Initializing RunConfig with distribution strategies.
INFO:tensorflow:Not using Distribute Coordinator.
I turn on GPU on Colab. I also tested it on EC2 instance with several GPUs. Do you know what wrong with this stuff?
Thanks,