NVIDIA GPU Technology Conference: A Green Rookie’s Perspective

Last week, on March 24, 2014, I woke up at an insanely early hour. Ahead of me was a long drive up from the Los Angeles area to San Francisco, where I was to meet up with Techgage’s Big Kahuna Rob for the first time. Although we had obviously been in contact for around a year and a half through emails and chat, we had never actually met face-to-face. In a few hours, though, the both of us would be sharing a car to San Francisco to hang out (including a wonderful lunch at the Fisherman’s Wharf, a long search for a gas station in the city, killing time in both a music store and a Guitar Center, and a yummy dinner), then a trip to nearby San Jose to attend NVIDIA’s GPU Technology Conference.

Although Rob is a veteran of such conferences, the 2014 NVIDIA GTC was my first-ever. As a true greenhorn, I had absolutely no expectations on how the event was going to be. Even before leaving L.A. I had decided to be a sponge, just absorbing as much as I could. I was in San Jose to learn, not just about what goes on in events like GTC, but also about the people in the tech industry.

Tuesday started bright and early for us, with an 8:00AM call for breakfast and press badge pick-up. We met up with Nathan Kirsch of Legit Reviews, one of Rob’s long-standing friends in the industry, and shared some time and conversation before the GTC Keynote Address.

The keynote itself was given by Mr. Jen-Hsun Huang, co-founder, President and CEO of NVIDIA. Huang was singularly impressive, shaking hands with the rows of press and greeting each journalist personally prior to the start of the keynote. And once up on stage, he was just as impressive as he spoke about his company’s capabilities and achievements.

Huang spoke about just some of what to expect from NVIDIA in the coming months and years. He talked about NVLink, which is basically a new design of chip-to-chip communications. He spoke about the company’s next-gen GPU architecture, Pascal. Pascal will take advantage of NVLink, which should grant it a massive boost in bandwidth and energy efficiency (talk about having your cake and being able to eat it, too). Huang then revealed the $3,000 GeForce GTX Titan Z, the most monstrous video card I personally have ever seen (even if I was about forty to fifty feet away from where the emperor of NVIDIA was standing on-stage).

After the Titan Z, Huang then talked up the IRAY VCA (Visual Computing Appliance), a $50,000 device used to make photo-realistic renderings in what is close to real-time. The secret to the IRAY VCA’s incredible performance is eight Kepler-class GPUs (with each GPU endowed with a cool 12GB of VRAM!) adding up to a mammoth 23040 CUDA cores. But that’s not all: The IRAY VCA is a scalable device, which means multiple devices can be interlinked together, thereby boosting performance hundreds of times over.

Afterwards, Huang spoke about GRID, which is essentially a GPU in the cloud. GRID is NVIDIA’s contribution to the Horizon DaaS Platform, a collaboration with VMWare. Finally, Huang talked about making CUDA available for mobile technology. The Tegra K1 will be NVIDIA’s first truly mobile super chip, unifying GPU and Tegra architectures, while the Jetson TK1 (available for a very wallet-friendly $192) will be the world’s tiniest super-computer, per Huang.

The GTC Keynote left me with a particular thought: NVIDIA is far more than just a GPU designer. It is actually at the leading edge of computing technology, inventing new ways to solve all sorts of human problems. The company participates in almost every realm of the human experience: Entertainment, art, engineering and design, and medicine are just a few examples. Its GPU business is the very tip of the iceberg, and, I suspect, comprises just a small portion of the company’s revenue stream.

Rob and I spent the rest of Tuesday attending various lectures and talks highlighting NVIDIA’s various technological partners and clients. The highlight was a lecture on machine learning (which, I readily confess, was a lot more than my feeble mind could process). That evening, we were invited to a great dinner with several NVIDIA GeForce team members (headed by Bryan Del Rizzo) and a number of tech press (including Nathan Kirsch of Legit Reviews, Tim Verry of PCPer, Dmitry Novoselov of Hardware Canucks, Anshel Sag of BSN*, and a couple of gentlemen from eTeknix, GamersNexus, amongst others). A fine way to cap my first technology conference, for sure.

Wednesday was also action-packed. The morning started with a keynote address by the good folks at Pixar. The real-time demos of how the artists at Pixar breathe life into its characters and environments using NVIDIA technology again reinforced the thought that NVIDIA is much more than just a producer of graphics chips. The bottom line of the Pixar keynote is that physics simulation is at the heart of everything NVIDIA does, and everything we (the audience) see in a Pixar film.

Indeed, physics simulation is at the heart of many of NVIDIA’s technologies. A talk about real-time geometric acoustics demonstrates this, where a game developer spoke about the nitty-gritty of audio processing and programming ingame design. Simulated environments affect simulated sound, thereby enhancing a game’s immersiveness.

Rob and I attended two other talks on Wednesday: One was about Android differentiation options on Tegra K1 (an impressive look at Tegra’s capabilities, particularly on the graphics side), while the other was about accelerating research and development using the Titan Supercomputer in Oak Ridge. We also took a test drive in a BMW 750Li super sedan, with a dash display powered by NVIDIA.

In the evening, we attended a gigantic party hosted by NVIDIA. I saw Jen-Hsun Huang in attendance, as well as Dave Altavilla of HotHardware and Scott Wasson of The Tech Report for some tasty food and tastier conversation.

Thursday was when we all broke camp and headed for home. Rob had a ride back to San Francisco, while I had to drive back down to Southern California. I chose to drive down using the Pacific Coast Highway, so at least my trip back home was far more breathtaking and entertaining than my drive north.

I really enjoyed my 2014 GTC experience. I learned so much, not just about NVIDIA and its amazing capabilities as a computing solutions provider, but also about the various people I met. At this point I’d like to thank NVIDIA for a wonderful time, as well as the people I’ve named in this retrospective look at GTC; I’m sure I’ve failed to mention at least a couple of other people by name, but only because my brain isn’t as good at retaining information (i.e., the oversight isn’t deliberate). In closing, I want to thank Rob for planting the seed of meeting up in San Jose for GTC in my head. It was great to have finally met after around eighteen months of constant communication and shared experiences; I feel our friendship, as strong as it was before, is even stronger now.