NVIDIA announced at a VLSI conference last year that it had designed a test multi-chip solution for DNN computations. The company is explaining the technology today at Hot Chips, with the idea that what they've created could be a stepping stone for future monetizable products.
A slightly different talk from Hot Chips this time: here's a neural network processor aimed at IoT devices based in 16nm. The slides look interesting, so we're covering this...5 by Ian Cutress on 8/20/2018
This week, Google announced Cloud TPU beta availability on the Google Cloud Platform (GCP), accessible through their Compute Engine infrastructure-as-a-service. Using the second generation of Google’s tensor processing units...8 by Nate Oh on 2/15/2018
Another Hot Chips talk, now talking Google TPU.30 by Ian Cutress on 8/22/2017