Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads.
GTC 20I8
“GTC was the introduction to the future of AI, a protector, a healer, a helper,
a guardian, a visionary, and just a little...
“Press quote.”
—Publication
“Clearly the adoption of GPU computing is growing.”
—ZDNet
Deep Learning Institute
10,000 training instances
Keynote
550K views
Show Floor
150 exhibitors
Inception Awards
$1M to 3 A...
“NVIDIA’s CEO, Jensen Huang, is a consummate
showman and he didn’t disappoint this year.”
—IT Business Edge
Real-time ray tracing has been the dream of computer scientists since it was first described nearly 40 years ago.
NVIDIA s...
The Quadro GV100 with NVIDIA RTX technology is the greatest advance in computer graphics of the past 15 years, since our
i...
For 30 years, the dynamics of Moore’s law held true. But CPU performance scaling has slowed. GPU computing is defining a
n...
TITAN VNVIDIA AI InferenceNVIDIA GPU Cloud
We are advancing GPU computing for deep learning and AI at the speed of light. ...
AI researchers want gigantic GPUs. We launched a breakthrough in deep learning computing with the introduction of NVIDIA
D...
1 DGX-2 | $399K | 10kW
1/8 the Cost 1/60 the Space 1/18 the Power
300 Dual-CPU Servers | $3M | 180 kW
“The More GPUs You B...
There is a Cambrian explosion of neural networks. Since
AlexNet, thousands of new models have emerged. With
hundreds of la...
“NVIDIA Strengthened Its Inference Push by Unveiling TensorRT 4”
—TheStreet
Every hyperscale server — millions — will be a...
Kubernetes on NVIDIA GPUs is here! Hyperscale datacenters are made up of thousands of CPUs, memory modules,
networking and...
DRIVE PX Parker
DRIVE PX 2
DRIVE Xavier
OrinDRIVE Pegasus
Autonomous vehicles will modernize the $10 trillion transportati...
Each year, 10 trillion miles are driven around the world. Test cars can eventually cover millions of miles, an insignifica...
NVIDIA DRIVE Constellation allows cars to drive billions of miles in virtual reality. Constellation consists of two differ...
The next chapter of AI is autonomous machines. We created a robotics platform called NVIDIA Isaac to accelerate the
develo...
Early detection is the most powerful weapon to treat disease. The latest breakthroughs of AI and computational imaging
can...
Billions of smart sensing devices will connect to the internet someday. NVIDIA and Arm announced a partnership to bring
de...
How do you drive a car remotely? Use virtual reality to teleport into the mind of an autonomous machine. Huh? As a final,
...
NEW DGX-2
1ST
2PF COMPUTER
300 SERVERS IN A BOX
NEW TESLA V100 32GB
NEW TENSORRT 4
AND MORE ISAAC
CLARA
ONE ARCHITECTURE
X...
Capping off an exciting week at GTC 2018, Jensen appeared on CNBC’s Mad Money for an interview with famed “NVIDIA”
dog own...
“NVIDIA Shows Why It’s
Leading in AI Mind
Expansion at GTC 2018”
—Forbes
“There’s no question that
NVIDIA remains one step...
GTC 2018: A New AI Era Dawns
Upcoming SlideShare
Loading in …5
×

GTC 2018: A New AI Era Dawns

8,016 views

Published on

At the 2018 GPU Technology Conference in Silicon Valley, NVIDIA CEO Jensen Huang announced the new "double-sized" 32GB Volta GPU; unveiled the NVIDIA DGX-2, the power of 300 servers in a box; showed an expanded inference platform with TensorRT 4 and Kubernetes on NVIDIA GPU; and revealed the NVIDIA GPU Cloud registry with 30 GPU-optimized containers and made it available from more cloud service providers. GTC attendees also got a sneak peek of the latest NVIDIA DRIVE software stack and the next DRIVE AI car computer, "Orin," along with developments in the NVIDIA Isaac platform for robotics and Project Clara, NVIDIA's medical imaging supercomputer.

Published in: Technology

GTC 2018: A New AI Era Dawns

  1. 1. GTC 20I8
  2. “GTC was the introduction to the future of AI, a protector, a healer, a helper, a guardian, a visionary, and just a little slice of amazing.” —IT Business Edge
  3. “Press quote.” —Publication “Clearly the adoption of GPU computing is growing.” —ZDNet
  4. Deep Learning Institute 10,000 training instances Keynote 550K views Show Floor 150 exhibitors Inception Awards $1M to 3 AI startups “If the excitement at GTC this week is any indication, the market will need lots and lots of GPUs in the years to come.” —Datanami GTC is ground zero of the GPU computing movement. NVIDIA pioneered this computing model for those doing groundbreaking work that cannot be done without a supercharged computer. GTC 2018 was a four-day event with 8,300 registered attendees who were offered more than 600 technical sessions. More than 200 reporters and 80 industry analysts experienced first-hand NVIDIA’s lineup of announcements.
  5. “NVIDIA’s CEO, Jensen Huang, is a consummate showman and he didn’t disappoint this year.” —IT Business Edge
  6. Real-time ray tracing has been the dream of computer scientists since it was first described nearly 40 years ago. NVIDIA stunned the audience with its new NVIDIA RTX technology — a platform for real-time ray tracing. “Ray Tracing and Giant GPUs Electrify GTC 2018” —Electronic Design
  7. The Quadro GV100 with NVIDIA RTX technology is the greatest advance in computer graphics of the past 15 years, since our introduction of the programmable shader. NVIDIA RTX is the culmination of 10 years of research, combining a new GPU architecture, algorithms, and deep learning in a way that no one else can. NVIDIA has reinvented computer graphics, again. “NVIDIA Is Really Throwing Down the Hammer with the Quadro GV100” —Hot Hardware
  8. For 30 years, the dynamics of Moore’s law held true. But CPU performance scaling has slowed. GPU computing is defining a new, supercharged law. It starts with a highly specialized parallel processor called the GPU and continues through system design, system software, algorithms, and all the way through optimized applications. The world is jumping on board. “NVIDIA Is So Far Ahead of the Curve” —The Inquirer 370PF 20182013 Total GPU FLOPS of Top 50 Systems 15X in 5 Yrs 8M 20182013 820,000 20182013 CUDA Downloads 5X in 5 Yrs GPU Developers 10X in 5 Yrs Measured performance of Amber, CHROMA, GTC, LAMMPS, MILC, NAMD, Quantum Espresso, SPECFEM3D 1 10 100 2013 2018 GPU–Accelerated Computing CPU Moore’s Law
  9. TITAN VNVIDIA AI InferenceNVIDIA GPU Cloud We are advancing GPU computing for deep learning and AI at the speed of light. We create the entire stack, and make it easily available in every computer, datacenter, and cloud. We supercharged NVIDIA AI with a new “double-sized” 32GB Volta GPU; announced the NVIDIA DGX-2, the power of 300 servers in a box; expanded our inference platform with TensorRT 4 and Kubernetes on NVIDIA GPU; and we built out the NVIDIA GPU Cloud registry with 30 GPU-optimized containers and made it available from more cloud service providers. “Creating Powerful System-level Solutions Will Give It an Edge Against Rivals Who Have Merely Developed a Good Chip” —TheStreet 30 GPU-Optimized Containers NEW TensorRT 4, TensorFlow, Kaldi, ONNX, WinML Out of stock!NGC Now on AWS, GCP, AliCloud, Oracle NEW 32GB NEW with V100 32GB NEW DGX-2 Tesla V100 DGX Systems Every Cloud
  10. AI researchers want gigantic GPUs. We launched a breakthrough in deep learning computing with the introduction of NVIDIA DGX-2, the first single server capable of delivering two petaflops of computational power. DGX-2 features NVSwitch, a revolutionary GPU interconnect fabric which enables its 16 Tesla V100 GPUs to simultaneously communicate at a record speed of 2.4 terabytes per second. Programming DGX-2 is like programming “the largest GPU in the world.” “NVIDIA Gave a Look Inside Its DGX-2, the Star of This Year’s GTC” —EE Times
  11. 1 DGX-2 | $399K | 10kW 1/8 the Cost 1/60 the Space 1/18 the Power 300 Dual-CPU Servers | $3M | 180 kW “The More GPUs You Buy, The More You Save” —Jensen Huang TRADITIONAL HYPERSCALE CLUSTER NVIDIA DGX-2 FOR DEEP LEARNING
  12. There is a Cambrian explosion of neural networks. Since AlexNet, thousands of new models have emerged. With hundreds of layers and billions of parameters, their complexity has soared by 500X in just 5 years. The hyperscale datacenters that host them serve billions of people, cost billions to operate, and are among the most complex computers the world has ever made. Maintaining great quality of service while minimizing cost is incredibly difficult. Jensen helps us remember with PLASTER. PROGRAMMABILITY LATENCY ACCURACY SIZE THROUGHPUT RATE OF LEARNING ENERGY EFFICIENCY Convolutional Networks Recurrent Networks Generative Adversarial Networks Reinforcement Learning New Species
  13. “NVIDIA Strengthened Its Inference Push by Unveiling TensorRT 4” —TheStreet Every hyperscale server — millions — will be accelerated for AI someday. The workload is complex — remember PLASTER — and the optimizing compiler technologies are still being invented. We announced TensorRT 4, the latest version of our inference software, and its integration into Google’s popular TensorFlow framework. We announced that Kaldi, the most popular framework for speech recognition, is now optimized for GPUs. NVIDIA’s close collaboration with partners such as Amazon, Facebook, and Microsoft makes it easier for developers to take advantage of GPU acceleration using ONNX and WinML. Hyperscale datacenters can save big money with NVIDIA Inference Acceleration. SPEECH SYNTH DGN, S2S TensorRT 2 INT8 TensorRT 3 Tensor Core TensorRT CNNs TensorRT 4 TensorFlow Integration Kaldi Optimization ASR RNN++ RECOMMENDER MLP-NCF NLP RNN IMAGE / VIDEO CNN 30M HYPERSCALE SERVERS 190X IMAGE / VIDEO ResNet-50 with TensorFlow Integration 50X NLP GNMT 45X RECOMMENDER Neural Collaborative Filtering 36X SPEECH SYNTH WaveNet 60X ASR DeepSpeech 2 DNN All speed-ups are chip-to-chip CPU to GV100. Sept ‘16 Apr ‘17 Sept ‘17 Apr ‘18 ONNX WinML
  14. Kubernetes on NVIDIA GPUs is here! Hyperscale datacenters are made up of thousands of CPUs, memory modules, networking and storage devices, and system software. Orchestration is key. We announced Kubernetes-ONG, software that allows orchestration of resources across multi-cloud GPU clusters. Provision on-prem and cloud GPU resources with a single command — even in the event of a system failure. “NVIDIA Brings Joy by Bringing GPU Acceleration to Kubernetes” —TechCrunch
  15. DRIVE PX Parker DRIVE PX 2 DRIVE Xavier OrinDRIVE Pegasus Autonomous vehicles will modernize the $10 trillion transportation industry — making our roads safer and our cities more efficient. NVIDIA DRIVE is a scalable AI car platform that spans the entire range of autonomous driving, from traffic-jam pilots to robotaxis. More than 370 companies contributing to the AV industry have adopted DRIVE. At GTC, we demonstrated our latest DRIVE software stack and gave a sneak peek at our next DRIVE AI car computer, “Orin.” “NVIDIA Already Has Their Eye on What’s Next … And That Is Orin” —AnandTech
  16. Each year, 10 trillion miles are driven around the world. Test cars can eventually cover millions of miles, an insignificant fraction of all the scenarios. We need to cover billions of miles to create a safe and reliable system. We have to invent something new. “There Are Too Many Situations to Train For and Not Enough Testing Time on the Highways” —VentureBeat
  17. NVIDIA DRIVE Constellation allows cars to drive billions of miles in virtual reality. Constellation consists of two different GPU servers. The first simulates the environment and what is detected by the car’s many sensors — cameras, radar, and lidar. The second is the NVIDIA DRIVE Pegasus AI car computer that runs the complete AV software stack and processes the simulated detected data as if it were coming from a real car. “NVIDIA Redefines Autonomous Vehicle Testing with VR Simulation System” —ZDNet
  18. The next chapter of AI is autonomous machines. We created a robotics platform called NVIDIA Isaac to accelerate the development and deployment of robotics across a broad range of industries. The Isaac SDK performs the important functions of robotics — perception, localization, navigation, and manipulation. We also created Isaac Sim, a virtual reality simulator where roboticists can create and train robots. Drop the software created in Isaac Sim into a robot with the Isaac SDK, and an intelligent machine is born. “The Robots Are Coming, Thanks to NVIDIA” —FutureCar Isaac Sim Isaac SDK
  19. Early detection is the most powerful weapon to treat disease. The latest breakthroughs of AI and computational imaging can help, but only if put into the hands of doctors using the 3 million medical instruments built a decade ago. We introduced Project Clara, NVIDIA’s medical imaging supercomputer, to do just that. With Clara, existing instruments will be supercharged with state-of-the-art image reconstruction, object detection and segmentation, and visualization capabilities. “Medical Imaging at the Speed of Light: NVIDIA’s Clara Supercomputer” —ZDNet NVIDIA GPU SERVER CUDA | CUDNN | TENSORRT | OGL | RTX GPU CONTAINERS | VGPU IMAGING & VISUALIZATION APPS
  20. Billions of smart sensing devices will connect to the internet someday. NVIDIA and Arm announced a partnership to bring deep learning inferencing to the wide array of mobile, consumer electronics, and Internet of Things devices. Arm has integrated the NVDLA inference accelerator into its Project Trillium platform for machine learning. The collaboration will make it simple for IoT chip companies to integrate AI into their designs and help put intelligent, affordable products into the hands of billions of consumers. “A New A.I. Era Dawns for Chip Makers” —Barron’s
  21. How do you drive a car remotely? Use virtual reality to teleport into the mind of an autonomous machine. Huh? As a final, mind-bending treat for the GTC audience, Jensen and NVIDIA’s auto team demonstrated “Project Wakanda.” A driver on the keynote stage entered virtual reality through NVIDIA Holodeck to teleport into a car at the back of the San Jose Convention Center. Seeing everything through the surround cameras of the car, he was able to remotely drive the car around a parked truck and safely into a parking space. “NVIDIA Stuns by Driving a Car in Real Life Through Virtual Reality” —Yahoo Finance
  22. NEW DGX-2 1ST 2PF COMPUTER 300 SERVERS IN A BOX NEW TESLA V100 32GB NEW TENSORRT 4 AND MORE ISAAC CLARA ONE ARCHITECTURE XAVIER – PEGASUS - ORIN Kubernetes On NVIDIA GPUs AI AUTO NEW PLATFORMSGRAPHICS QUADRO GV100 NVIDIA RTX DRIVE SIM & CONSTELLATION “What We Saw at GTC Was the Future” —IT Business Edge
  23. Capping off an exciting week at GTC 2018, Jensen appeared on CNBC’s Mad Money for an interview with famed “NVIDIA” dog owner and stock market prognosticator Jim Cramer. The interview covered a broad set of topics, from our recently announced products to Jensen’s point of view on the future of computer graphics, AI, and autonomous cars. “NVIDIA CEO Jensen Huang: Tomorrowland of Technology” —CNBC
  24. “NVIDIA Shows Why It’s Leading in AI Mind Expansion at GTC 2018” —Forbes “There’s no question that NVIDIA remains one step ahead.” —The Inquirer “The excitement throughout the entire session was amazing.” —Electronic Times “The company is about to drop the pedal and hit another gear in the AI race.” —Forbes

×