Published On: Thu, Jan 26th, 2017

IBM adds support for Google’s Tensorflow to the PowerAI appurtenance training framework


PowerAI is IBM’s appurtenance training horizon for companies that use servers formed on a Power processors and NVIDIA’s NVLink high-speed interconnects that concede for information to pass intensely fast between a processor and the GPU that does many of a low training calculations. Today, a association announced that PowerAI now supports Google’s renouned Tensorflow appurtenance training library.

While TensorFlow has usually been accessible for a small over a year, it has fast turn a many renouned open source appurtenance training plan on GitHub. IBM’s PowerAI already supported other frameworks and libraries like CAFFETheano, Torch, cuDNN, and NVIDIA DIGITS, yet Tensorflow support was sorely blank from this lineup.

IBM clearly sees a multiple of PowerAI with Nvidia’s NVLink interface and Pascal P100 GPU accelerators as a approach to compute itself from a foe — and in this case, the foe it is gunning for is clearly Intel (though it’s value observant that Intel and Google also recently teamed up to urge TensorFlow opening on a CPUs).

IBM, of course, isn’t in a business of offered inexpensive servers, so a Power System S822LC for high-performance computing — a one appurtenance that supports PowerAI — doesn’t even underline on a pricing chart. Prices for a blurb counterpart, though, start during only underneath $10,000.

In further to TensorFlow support, IBM also currently remarkable that it now supports Chainer, a horizon for building neural networks.

Featured Image: JuliusKielaitis/Shutterstock

About the Author

Leave a comment

XHTML: You can use these html tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>