GCP NEXT 2016: Day 1 Keynote (Mar 23)

As usual leave it up to Jeff Dean to be an outlier, this[0] is his talk/demo at GCP Next 2016 - I was on edge of my seat as I watched him talk and code things up at the same time. He is awesome. Oh and Gary Bernhardt's talk WAT[1] which happens to be one of my favorite talk of all time. [0] [VIDEO] [1] [LINK]

This narrative gets repeated time and time again, and it really doesn't hold up to even surface debate: - Google TI and Cloud merged. Same teams. Per Urs Holzle in Jan2014 - Just last week Sundar (CEO), Eric Schmidt (chairman), Jeff Dean, Urs Holzle, and Diane were on stage talking about how they spent $10 billion last year investing in infrastructure and how serious Google is about cloud ( [VIDEO] ) - In many cases, publically available technologies are actually replacing internal technologies (BigQuery and Dremel, Dataflow and FlumeJava/Millwheel for example). - Disney, Coca Cola, Snapchat, Spotify, Home Depot are just some of the customers running on Google Cloud. It'd be very foolish to abandon these. - If you watch the day two GCP NEXT keynote, you'll see how Google is the cloud leader in environmental responsibility and open source. We're the good guys :) ( [LINK] ) Happy to debate further!

In March at the GCP NEXT keynote [1], Jeff Dean demos Cloud ML on the GCP. He casually mentions passing in the argument "replicas=20" to get "20 way parallelism in optimizing this particular model". GCE does not currently offer GPU instances. I've never heard the term replicas in the GPU ML discourse. These devices may enable a type of parallelism that we have not seen before. Furthermore, his experiments are apparently using the Criteo dataset, which is a 10GB dataset. Now, I haven't looked into the complexity of the model or to what extent they train it to, but right now that sounds really impressive to me. 1: [VIDEO]