Power servers

IBM at GTC: Experience the power of GPU-accelerated deep learning

Share this post:

It’s the most wonderful time of the year, and no, I’m not talking about Spring. It’s finally time for NVIDIA’s GPU Technology Conference, or GTC. Here at IBM Power Systems we couldn’t be more excited, and as a Diamond sponsor of the event we are looking forward to showcasing all of our latest technology and show attendees how our clients are putting that technology into action.

GTC will feature IBM content at over 15 sessions and from 30 speakers at the event focused on world-shaping trends like deep learning, accelerated computing, parallel computing development, graphics processing units (GPUs) and the cloud. But with such a broad agenda, how do you know what should be on your must-see list? To help, we’ve put together our top five list of can’t-miss sessions and things to do at GTC to get you started!

1. See the future of GPU-accelerated deep learning with IBM PowerAI’s roadmap

Join Sumit Gupta, IBM VP of Analytics, AI and Machine Learning on Tuesday at 4PM in Room 210C for a deep-dive into the future of IBM’s enterprise-leading open-source deep learning toolkit, PowerAI. Sumit will discuss how IBM’s engineers are optimizing the performance of the most popular deep-learning distributions like TensorFlow, Caffe and Torch thanks to the world’s only CPU-to-GPU NVIDIA NVLink interface. He’ll also detail the ease-of-use features being implemented in PowerAI to continue its reputation as the simplest avenue for enterprise deep learning.

2. Hear from PowerAI’s chief engineer

Hear from IBM Chief Engineer for Machine Learning and Deep Learning Michael Gschwind on how he and his team are integrating the leading open-source deep learning distributions into PowerAI. He’ll explain how PowerAI’s performance-focused enhancements to distributions like IBM Caffe were combined with NVIDIA NVLink and four NVIDIA Tesla P100 GPUs on the IBM Power Systems S822LC for HPC platform to capture top marks in AlexNet and VGGNet. Be sure to add Session S7368 to your GTC Session Planner!

3. See how IBM’s T.J. Watson Research Center uses NVIDIA NVLink to create larger models and shatter bottlenecks

A featured speaker on the GTC Accelerated Data Analytics track, IBM T.J. Watson Research Center’s Rajesh Bordawekar will show you how developers and data scientists can apply larger models than ever before all the while avoiding the PCIe data transfer bottleneck with the world’s only CPU-to-GPU NVIDIA NVLink server. Don’t miss Rajesh’s session on Wednesday.

4. Visit the PowerAI Pavilion

Come to our PowerAI Pavilion, located at Booth #411, to see the latest updates for IBM PowerAI, IBM’s toolkit of open-source deep learning distributions for the enterprise. Learn more about how we incorporated distributed TensorFlow in the latest version and what that means to you, straight from the IBM experts. Hear from PowerAI early adopters about how they’re building deep learning solutions on IBM’s NVIDIA NVLink-equipped Power Systems S822LC for HPC to tap into the power of GPU-accelerated deep learning.

5. Start building your deep learning game changer with PowerAI and Nimbix

By now you’re probably pretty excited about seeing what PowerAI and deep learning can do for you at GTC, beginning May 8 in Silicon Valley. Come chat with our cloud partner Nimbix at our booth, #411. They can show you how you can get up and running with PowerAI in just minutes using the Nimbix cloud!

Be sure to share your experiences by tagging #IBMPowerSystems in your social posts, and comment below if you make it to the conference. See you there!

Director of Product Management, Deep Learning, IBM Cognitive Systems

Add Comment
One Comment

Leave a Reply

Your email address will not be published.Required fields are marked *


Scott Soutter

Thanks for putting this all in one place, Adel. Really looking forward to GTC!

More Power Systems stories

5 key reasons to opt for hyperconverged infrastructure

Let’s face it—traditional design for data center infrastructure doesn’t always meet today’s needs. Analytics, transaction processing, high-volume web serving and other data-intensive and mission-critical workloads place new and increased demands on the IT infrastructure. At the same time, specialized IT skills are becoming increasingly challenging to find, and budgets for capital IT expenses continue to […]

Continue reading

Scale massive SAP HANA database capacity in the cloud

Every organization, including yours, wants an edge, whether that advantage is getting to market first, delivering exceptional customer experiences or creating unique operating efficiency. And now you have an easy starting point—SAP S/4HANA on IBM Power Systems servers in the IBM Cloud, a new capability announced at Think 2018. In-memory databases such as SAP HANA […]

Continue reading

IBM POWER9 scale-out servers deliver more memory, better price performance vs. Intel x86

IBM recently began shipping its next-generation Power Systems servers incorporating its newly-designed POWER9 processor. These POWER9-based scale-out systems, including S914, S922, S924, H922, H924 and L922, provide leadership capabilities in non-accelerated compute spaces such as in-memory databases, advanced analytics and cloud environments. POWER9 scale-out servers are cloud-ready and include built-in PowerVM virtualization capabilities, enabling you […]

Continue reading