I gave a talk at the Spark Summit Europe last week, where i went into detail on this server and how we can scale out deep learning training on Tensorflow with i

I gave a talk at the Spark Summit Europe last week, where i went into detail on this server and how we can scale out deep learning training on Tensorflow with it and AllReduce (by Uber): https://www.slideshare.net/secret/A7b9rAsLaipg6 https://www.slideshare.net/secret/A7b9rAsLaipg6 TLDR; You can scale out distributed Tensorflow training to tens/hundreds of GPUs with AllReduce on machines like this one, not just on the DGX-1.