The 10th edition of the $9.71-billion NVIDIA Corporation’s annual GPU Technology Conference (GTC 2018) for GPU developers opened on Tuesday to an audience of 8,500 where its Founder, President and CEO, Jensen Huang unveiled a series of advances to its deep learning computing platform.
For over two hours, Huang took the audience through some “amazing graphics, amazing science, amazing AI and amazing robots.”
Introducing NVIDIA RTX technology that runs on a Quadro GV100 processor, he said: “This technology is the most important advance in computer graphics in 15 years as we can now bring real-time ray tracing to the market. Virtually everyone is adopting it.”
Elaborating on its relevance, he said, the gaming industry that makes 400 games a year, uses ray-tracing to render entire games in advance. The film industry uses it in the 500 movies made each year where each frame is rendered multiple times. Today, one billion images are rendered each year and that could go up 10X as more real-time rendering is reduced by Quadro to one-fifth the cost, one seventh the space and one-17th the power.
Virtualised data centre
Stating that the work NVIDIA does in modern medical imaging is one of the things, he is proudest of, Huang says, there are around 3 million medical instruments installed, but only one lakh sold each year. It would take 30 years to update everything. To avoid this, the company has an initiative - Project Clara – a virtualised data centre, remoted, multi-modality, multi-user. “It’s a medical imaging supercomputer. It’s possible now for us to virtually update every system that’s out there.”
He announced key advancements to the NVIDIA platform — a 2X memory boost to NVIDIA Tesla V100, the most powerful data centre graphic processing unit that accelerates AI, high performance computing, graphics and a new GPU interconnect fabric called NVSwitch that enables up to 16 Tesla V100 GPUs to simultaneously communicate at a record speed of 2.4 terabytes per second.
Deep learning computing
Another major breakthrough in deep learning computing was announced with - DGX-2, the first single server capable of delivering two petaflops of computational power with the processing power of 300 servers occupying 15 racks of data centre space, while being 60 times smaller and 18 times more power efficient.
“We are enhancing our platform’s performance at a pace far exceeding Moore’s law, enabling breakthroughs that will help revolutionise healthcare, transportation, science exploration and countless other areas” said Huang.
The updates to the company’s deep learning and HPC software stack are available at no charge to its developer community, which now totals 8.2 lakh registered users, compared with about 4.8 lakh a year ago.
Huang highlighted research projects of NVIDIA’s 200-person research team spread across 11 locations worldwide, that are focused on pushing the boundaries of technology in machine learning, computer vision, self-driving cars, robotics, graphics, computer architecture, programming systems etc.
He invited the audience into NVIDIA’s Holodeck VR (virtual reality) environment to showcase how VR has the ability to teleport us into a new world, where an autonomous vehicle can be assisted.
“In a self-driving car, humans are the backup system. But what about an autonomous machine with no operators? It could be a tractor with no driver. So, how do we create a backup system?” said Huang. The audience watched a man sitting on stage, who is inside the Holodeck VR environment projected on screen and who in VR form takes over the car in VR which is translated into controlling a real car that’s on the street.
Robotics platform
Keeping with the trend of autonomous machines that are revolutionising every industry today, NVIDIA released the Isaac Robotics Platform, which will develop various capabilities for robots to navigate in the Isaac simulated lab.
NVIDIA created GeForce, the world’s largest gaming platform and has been leading visual computing since its invention of the GPU in 1999. However, the creation of its Tesla GPU platform in 2006 opened up the parallel-processing capabilities of the GPU to general purpose computing. “One of best decisions we made was to make our computing general purpose” said Huang.
Today, GPU computing is the most pervasive, energy efficient path forward for HPC and data centres and powers the fastest supercomputers in the US and Europe.
The writer is in San Jose at the invitation of NVIDIA