NVIDIA GTC deep learning dispatch: Day 1

08 May, 2017
Sam Ponedal
IBM

It’s begun! NVIDIA’s annual GPU Technology Conference, or GTC. Last week we told you about our Top 5 Things to Do, and now we’re here live at the show to bring you the latest chatter from the conference floor, where industry leaders like IBM, NVIDIA, Mellanox, Samsung and more are coming together. For IBM, our developers have been telling us that they are chomping at the bit to dive into deep learning, and IBM made sure to feed their appetite on day one of GTC.

Booth #411 at IBM GTC

Going deep with deep learning

NVIDIA GTC has emerged as one of the premier conferences on deep learning, and day one proved that this year was no different. It started with NVIDIA Director of Developer Marketing Will Ramey asking the very important question: “What problems are you trying to solve?” Deep learning can be applied in a variety of ways, driving a multitude of different outcomes. He highlighted how different deep learning distributions like Caffe or TensorFlow can be used to help progress a series of related outcomes. For example, he described how deep learning can help improve the cancer diagnostic and treatment process.

what problem are you solving?

He also highlighted what an organization needs to get started in deep learning, and why.

Challenges and deep learning needs

Looking at this, it’s easy to see how and why IBMers started developing PowerAI.

IBM Machine Learning and Deep Learning Chief Engineer talks PowerAI

Michael Gschwind, the Chief Architect and Chief Engineer behind PowerAI, was on site to present the latest on IBM’s enterprise-leading deep learning toolkit. PowerAI has been the talk of the show thus far, and the room was packed with people eager to learn the latest on IBM’s simple deep learning on-ramp. In fact, this simplicity was one of three goals that Michael and his team set for themselves when they started thinking about PowerAI.

Goals for deep learning

He discussed how IBM was primed to become a leader in deep learning, having pioneered many of the best practices common in the realm of high-performance computing that could then be applied to deep learning.

three gens of accelerated power systems for deep learning

And of course, how we pioneered CPU-to-GPU NVLink in the creation of our S822LC for HPC server, the best-in-breed server for deep learning. But part of PowerAI’s secret sauce is its deep integration with the open-source deep learning community. Open source communities like OpenPOWER have proven to be dependable sources of innovation, and this was recognized early on. In this vein, Michael detailed how future partnerships will only continue to build the success of PowerAI, bringing in industry leaders like Julia, DL4J, Apache Spark and more.

PowerAI supported by a growing, optimized high-performance ecosystem

Continue your deep learning journey with IBM

For many, day one of GTC marked the start of their deep learning journey, and they learned about how PowerAI is the on-ramp to get them there. For others who have already begun exploring deep learning, they became more convinced that PowerAI is the best guide to help them along. Tomorrow we’ll be bringing you more coverage, including IBM VP of HPC, Analytics and AI Sumit Gupta’s presentation on how PowerAI will continue to be the enterprise deep learning platform of choice for years to come.

Do you want to see how you can get started on deep learning today? Then try it for free with Nimbix here.

Don’t forget to tag @IBMPowerSystems on Twitter and IBM Power Systems on LinkedIn and Facebook to show us how you’re enjoying the IBM experience at GTC!

The post NVIDIA GTC deep learning dispatch: Day 1 appeared first on IBM Systems Blog: In the Making.