Investigate timing in multi-GPU example

[agitter]

#24 adds a multi-GPU PyTorch example that demonstrates how to use Distributed Data Parallel training. However, training with multiple GPUs does not speed up training in the example. See #24 (comment)

It would be worthwhile to monitor the training more closely, for instance the GPU utilization, to understand why this is the case.

[agitter]

Additional testing described in #24 (comment) shows the GPU utilization is high for two different types of GPUs. The lack of speedup could be related to the relatively small convolutional neural network model.