"Computers powered by deep learning can do tasks that we can't imagine writing software for. Deep Learning isn't just a field or an app. It's way bigger than that. So, our company has gone all in for it."
-Jen-Hsun Huang, NVIDIA CEO & Co-founder
2016 GPU Technology Conference
This is my second year attending the GTC. My first year was eye-opening and exciting, but I was somewhat timid in meeting people and starting discussions. This was because I don't have a technical/coding background. I worried that my questions would be naive and uninformed. However, I've been learning and researching over the last year and this gave me much-needed confidence coming to the 2016 GTC conference.
The GTC experience this year was much broader for me. I started conversations a lot more and felt much clearer in discussing graphics technology with not only professionals, but other members of the press corp as well. I was also able to manage my time better (thanks to NVIDIA's great MyGTC app) which led to attending more sessions and talks. Plus, I spent considerable time on the exhibition floor trying out demos and approaching various companies that interested me.
Keynote Surprise: TensorFlow and Open Source
The main keynote address given by Jen-Hsun Huang, the CEO and Co-founder of NVIDIA, was amazing. Hard to imagine he could top last year's address, but he did. In his charismatic, but understated/humorous style, Mr. Huang did more than just announce new NVIDIA products and technology, he shared the ideas behind the technology as well. In particular, his focus on Deep Learning laid the foundation for all of the technologies discussed in his keynote.
You can read my full report on Jen-Hsun Huang's GTC keynote address here, or watch it for yourself (recommended) at the NVIDIA on-demand page. Note that NVIDIA also makes available the other two keynote addresses along with featured talks at the same page. There are pdf's of the talks available for download as well.
"What's cool is you open sourced it," Jen-Hsun says. "That made it possible for us to start adopting TensorFlow for DGX-1 right away. "We took open source, and adopted it for DGX-1."
In a surprising moment, Mr. Huang invited Rajat Monga, TensorFlow technical lead and manager at Google, onstage to discuss his Open-Source Deep Learning software, TensorFlow. I say surprising, because I can't think of another major corporate keynote that praised the idea of Open Source as an aide to their technology (the new DGX-1, a supercomputer build solely for Deep Learning). By emphasizing Rajat's Open Source Deep Learning software in his Keynote address, Mr. Huang validated the increasing importance of the Open Source movement thereby giving it a significant role in developing both for-profit and free graphics technology and research. And while there perhaps more important announcements in Mr. Huang's keynote, his embrace of Open Source will open doors with other tech companies around the world leading to a much broader collaboration between the corporate tech world and the Open Source movement. Something that is long overdue.
VR is for Real
Prior to the 2016, I had only experienced one VR demo: the early Oculus Rift at Siggraph 2015. And although the experience was remarkable, the demos at the GTC's VR Village were light-years ahead of that technology.
The VR field of vision is wider, making the experience more life-like; there was literally no latency in the 360 degree worlds I visited and the VR programming/media is much more subtle and fun.
Epic's Bullet Train, the amazing VR shooter, is going to blow gamers away with it's "you are there" experience and simplified controls. I'm also excited about the possiblity of nature documentaries, like the BBC's "Life on Earth", after going through Solfar Studio's Everest VR. Designed as a game-like experience, it seemed to me that the truly remarkable re-creation of Everest lends itself better to nature documentary/learning that the "can I make it to the top" game.
But the real revelation for me came with the demo of Google's Tilt Brush VR art creation application. It was enormous fun to be able to paint in 3 dimensions. And watching others get the hang of it and suddenly have their creative imagination take off was a remarkable experience.
Google's Tilt Brush VR art application in action
My Personal Highlight: Automomous Drones
Of course, the keynote address by Jen-Hsun Huang was the high-point of the 2016 GTC, as it should be. And I had many inspiring experiences at the GTC (the conversation with XX and XX from NVIDIA being one), but the one event that has stayed with me was Dor Abuhasira's talk Enabling Autonomous Drones with Real Time Computer Vision Applications.
The reason this talk has stayed with me is that, first, it was an inspiring talk that got me to thinking about the applications of Deep Learning and autonomous machines in a way I hadn't imagined before. Plus, Dor's story of developing his idea for an autonomous drone for use in Isreal's giant wind turbine industry, shows how NVIDIA's Deep Learning technology combined with ingenuity on a small scale can make a huge impact on the future of autonomous machines. Corporations and large companies spend millions on research and development: Dor bought a build-it-yourself drone kit and with determination and intelligence pushed intelligent machine technology forward and created a company (Percepto) that is inspiring.
Modeled along the lines of the great Siggraph conferences (although not derivative of them), NVIDIA's GPU Technology Conference has doubled in size over the last two years. And the increasing importance of the graphics processor/chip in enabling remarkable progress in the areas of deep learning, autonomous machines and virtual reality are making the GTC more than just an essential event; it has become a necessary one. The technologies the GTC is emphasizing with their presentations, keynotes, tutorials, labs and meeting pods, will will profoundly affect the future of every aspect of our lives from medical care to the spaces we live in. NVIDIA knows this which is why the GTC is the most forward-looking conference in the world today.
My own experience of the GPU Technology Conference is a small one. Literally hundreds of conversations, pod chats and meetings took place over the 4 day conference. And the result of those interactions are likely to influence the way we live our lives in the near future. NVIDIA has committed itself as a company to go "all in" on technologies that are profoundly influential and important for our future and not just because they stand to make a lot of money doing it. They also make the commitment because it's a good thing to do for everyone.
I believe NVIDIA is a model for a new type of 21st-century corporation. The company, under the inspiring leadership of Jen-Hsun Huang and his team, have managed to balance the classic "profit is the bottom-line" philosophy with the very moral idea that a company should do good while making that profit.
Thank you, NVIDIA
If there was one word I could use to describe NVIDIA's producing and organization of the GTC it would be "classy". Everything I came in contact with from the people serving coffee to the division heads, there was a sense that a good deal of thought had gone in to how things were organized and presented. That takes the worry and stress out of a conference. And belive me that is no easy accomplishment. Incredible amounts of effort went in to making the GTC look effortless and classy.
Particular thanks go to Gail Laguna and Sandeep Gupte, both of NVIDIA, for absorbing conversations and graceful help in negotiating such a large conference.
There's a great moment by moment account of Mr. Huang's keynote address can be found at the NVIDIA blog. Of course, all of the keynotes and many presentations (including my favorite on Autonomous Drones are available on demand at http://on-demand-gtc.gputechconf.com/gtcnew/on-demand-gtc.php
The main GTC website has a wealth of information and links. NVIDIA's YouTube channel is also an easy way to find out more about the technologies behind the conference and learn about new NVIDIA graphics hardware and VR announcements (see embed below).
I also recommend the daily wrap-up videos produced by NVIDIA. The overall GTC video wrap-up by Greg Estes is superb.