SUPERCHARGE DEEP LEARNING WITH DGX-1 Markus Weber SC16 - November 2016
NVIDIA Pioneered GPU Computing Founded 1993 $7B 9,500 Employees 100M NVIDIA GeForce Gamers The world s largest gaming platform Pioneering AI computing for self-driving cars DGX-1: World s 1 st Deep Learning Supercomputer The deep learning platform for AI researchers worldwide GE Revolution The GPU choice when it really matters The visualization platform of every car company and movie studio The processor of #1 U.S. supercomputer and 9 of 10 of world s most energy-efficient supercomputers 2
NVIDIA Computing for the Most Demanding Users GPU Computing Computing Human Imagination Computing Human Intelligence 3
DEEP LEARNING A NEW COMPUTING MODEL Software that writes software LEARNING ALGORITHM millions of trillions of FLOPS little girl is eating piece of cake" 4
SUPERHUMAN RESULTS SPARK HYPERSCALE ADOPTION Alibaba/Aliyun Amazon Baidu ebay Facebook ImageNet Accuracy % Human 93% 96% Flickr Google iflytek iqiyi JD.com Deep Learning 84% 88% Orange Periscope Pinterest Qihoo 360 Shazam 74% 74% 72% 76% Hand-coded CV 2010 2011 2012 2013 2014 2015 Skype Sogou Twitter Yahoo Supermarket Yandex Yelp Cloud Services with AI Powered by NVIDIA 5
NVIDIA DGX-1 IN ACTION Deep Learning and AI Analytics Users Automotive Financial Services Government/Defense Pedestrian Detection Lane Tracking Fraud/Anomaly Detection Risk Analysis Trading algorithms Face Detection Video Surveillance Graph Analytics Healthcare Higher Education/Research A.I. Start-ups Source: Bloomberg Cancer Cell Detection Image Classification Video Search Disease Identification A.I. Research Speech Recognition Drug Discovery Speech Processing Sentiment Analysis Recommendation 8
IDENTIFYING DEEP LEARNING OPPORTUNITIES Data Types and Applications Data types: Are you dealing with massive amounts of data in the form of images, videos, speech and text? Deep Learning uses deep neural networks to gobble up vast quantities of data, such as images, videos, speech and text, to learn to recognize patterns. Applications: Are you using signal-processing, image-processing or accelerated analytics applications? These applications can benefit from using Deep Learning, which is suited to solve problems like speech recognition and image classification. Are you developing or training deep learning models? 9
NVIDIA DGX-1 AI Supercomputer-in-a-Box 170 TFLOPS 8x Tesla P100 16GB NVLink Hybrid Cube Mesh 2x Xeon 8 TB RAID 0 Quad IB 100Gbps, Dual 10GbE 3U 3200W 10
FIVE MIRACLES Pascal Architecture 16nm FinFET CoWoS with HBM2 NVLink New AI Algorithms 11
NVIDIA DGX-1 VALUE PROP: SOFTWARE STACK Fully integrated Deep Learning platform Instant productivity plug-andplay, supports every AI framework Performance optimized across the entire stack Always up-to-date via the cloud Mixed framework environments containerized Direct access to NVIDIA experts 12
13
DGX-1 VALUE PROP: CONTAINER LAUNCH FLOW Customer data stays on premise LOCAL LAN compute.nvidia.com 1. User schedules containers to run Web Browser All Application Data Node Management User Authentication 3. User interacts with application NFS Storage DIGITS UI Docker Image push/pull Interactive Sessions Scheduler UI HW/SW Metrics 14
USERS OF NVIDIA DGX-1 Data Scientists & AI Researchers Why use NVIDIA DGX-1? Reduce DL training time Analyze and visualize vast amount of data Accelerate deep learning frameworks Design more sophisticated neural networks CIO, CTO, CMO, Line Of Business (LOB) Why buy NVIDIA DGX-1? Extract actionable insights Create new business opportunities Turn huge amounts of data into extreme value IT Directors & Managers Why add NVIDIA DGX-1 into your datacenter? Cut infrastructure footprint by 250x and reduce cost by 20x Reduce power and cooling costs Save installation and configuration time 15
Relative Training Performance DGX-1 VALUE PROP: A LEAGUE OF ITS OWN 16X ResNet Inception v3 AlexNet vgg MSR 12X 8X 4X 1X 0X GeForce GTX TITAN X X GeForce GTX 1080 Tesla P100 DIGITS DevBox (4X (4X GeForce GTX Titan TITAN X) X) Quadro Quadro VCA (8X VCA Quadro (8X Quadro M6000) M6000) DGX-1 (8X DGX-1 Tesla P100) (8X Tesla P100) Caffe on DeepMark. GeForce TITAN X and GTX 1080 system: Intel Core i7-5930k @ 3.5 GHz, 64 GB System Memory Tesla P100 (SXM2) system: Dual CPU server, Intel E5-2698 v4 @ 2.2 GHz, 256 GB System Memory 16
NVIDIA DGX-1 $129K 250 NODE HPC SUPERCOMPUTER-IN-A-BOX # Servers 250 Easier to manage Cost per server $9,000 IB cost per node $1,000 Total value and more $2.5M 100X less power, smaller footprint, less DC space 17
SAMPLE DGX-1 CUSTOMERS OpenAI NYU Mass General 18
NVIDIA DEEP LEARNING EVERYWHERE, EVERY PLATFORM CLOUD Everywhere TITAN X Available via etail in 200+ countries DGX-1 The AI Supercomputer for instant productivity TESLA Servers in every shape and size 19
NVIDIA EXPERTISE AT EVERY STEP Solution Architects Deep Learning Institute GTC Conferences Global Network of Partners Need image 1:1 support Network training setup Network optimization Certified expert instructors Worldwide workshops Online courses Epicenter of industry leaders Onsite training Global reach NVIDIA Partner Network OEMs Startups 20
DGX-1 THE ESSENTIAL TOOL OF DEEP LEARNING SCIENTISTS 250 node HPC Supercomputer-in-a-Box Reduce training time from weeks to days The platform of AI pioneers 21
NVIDIA DGX-1 The Essential Tool of Deep Learning Scientists Deep Learning is a massive opportunity Data Scientist productivity is vital NVIDIA is the choice of the deep learning world DGX-1 is fast, instantly productive 22