Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
Markus Weber
SC16 - November 2016
SUPERCHARGE DEEP LEARNING WITH DGX-1
2
GE Revolution — The GPU choice when it really matters
The processor of #1 U.S. supercomputer and 9 of 10 of world’s most energy-efficient supercomputers
DGX-1: World’s 1st Deep Learning Supercomputer — The deep learning platform for AI researchers worldwide
100M NVIDIA GeForce Gamers — The world’s largest gaming platform
Pioneering AI computing for self-driving cars
NVIDIA Pioneered GPU Computing | Founded 1993 | $7B | 9,500 Employees
The visualization platform of every car company and movie studio
3
GPU Computing
NVIDIA Computing for the Most Demanding Users
Computing Human Imagination
Computing Human Intelligence
4
DEEP LEARNING — A NEW COMPUTING MODEL
“Software that writes software”
“little girl is eating
piece of cake"
LEARNING
ALGORITHM
“millions of trillions
of FLOPS”
5
72%
74%
84%
88%
93%
96%
2010 2011 2012 2013 2014 2015
“SUPERHUMAN” RESULTS SPARK HYPERSCALE ADOPTION
Deep Learning
ImageNet — Accuracy %
Cloud Services with AI Powered by NVIDIA
Alibaba/Aliyun Amazon Baidu eBay Facebook
Flickr Google iFLYTEK iQIYI JD.com
Orange Periscope Pinterest Qihoo 360 Shazam
Skype Sogou Twitter Yahoo Supermarket Yandex Yelp Hand-coded CV
Human
74% 76%
8
NVIDIA DGX-1 IN ACTION Deep Learning and AI Analytics Users
Pedestrian Detection
Lane Tracking
Fraud/Anomaly Detection
Risk Analysis
Trading algorithms
Face Detection
Video Surveillance
Graph Analytics
Video Search
Speech Recognition
Sentiment Analysis Recommendation
Image Classification
A.I. Research
Speech Processing
Source: Bloomberg
Automotive Financial Services Government/Defense
Healthcare Higher Education/Research A.I. Start-ups
Cancer Cell Detection
Disease Identification
Drug Discovery
9
IDENTIFYING DEEP LEARNING OPPORTUNITIES
• 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?
Data Types and Applications
10
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
11
“FIVE MIRACLES”
16nm FinFET Pascal Architecture CoWoS with HBM2 New AI Algorithms NVLink
12
Instant productivity — plug-and-play, 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
NVIDIA DGX-1 VALUE PROP: SOFTWARE STACK
Fully integrated Deep Learning platform
13
14
DGX-1 VALUE PROP: CONTAINER LAUNCH FLOW Customer data stays on premise
Web Browser
Node Management
User Authentication
Docker Image push/pull
Scheduler UI
HW/SW Metrics
LOCAL LAN
All Application Data
NFS Storage
DIGITS UI
Interactive Sessions
compute.nvidia.com 1. User schedules containers to run
3. User interacts with application
15
USERS OF NVIDIA DGX-1
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
Data Scientists & AI Researchers
Why buy NVIDIA DGX-1?
• Extract actionable insights
• Create new business opportunities
• Turn huge amounts of data into extreme value
CIO, CTO, CMO, Line Of Business (LOB)
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
IT Directors & Managers
16
0X
4X
8X
12X
16X
GeForce® GTX TITAN X GeForce® GTX 1080 Tesla® P100 DIGITS™ DevBox (4X GeForce® GTX Titan X)
Quadro® VCA (8X Quadro®M6000)
DGX-1™ (8X Tesla® P100)
Rela
tive T
rain
ing P
erf
orm
ance
ResNet Inception v3 AlexNet vgg MSR
DGX-1 VALUE PROP: A LEAGUE OF ITS OWN
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
1X
GeForce GTX TITAN X GeForce GTX 1080 Tesla P100 DIGITS DevBox (4X GeForce GTX TITAN X)
Quadro VCA (8X Quadro M6000)
DGX-1 (8X Tesla P100)
17
NVIDIA DGX-1 — $129K
250 NODE HPC SUPERCOMPUTER-IN-A-BOX
# Servers 250
Cost per server $9,000
IB cost per node $1,000
Total value $2.5M
and more… 100X less power,
smaller footprint,
less DC space…
Easier to manage
18
SAMPLE DGX-1 CUSTOMERS
OpenAI
Mass General
NYU
19
NVIDIA DEEP LEARNING EVERYWHERE, EVERY PLATFORM
TITAN X Available via etail in
200+ countries
DGX-1 The AI Supercomputer for
instant productivity
TESLA Servers in every shape and size
CLOUD Everywhere
20
NVIDIA EXPERTISE AT EVERY STEP
Solution Architects Global Network
of Partners Deep Learning
Institute GTC
Conferences
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
Need image
21
DGX-1 — THE ESSENTIAL TOOL OF DEEP LEARNING SCIENTISTS
The platform of AI pioneers
Reduce training time from weeks to days
250 node HPC Supercomputer-in-a-Box
22
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
NVIDIA DGX-1 The Essential Tool of
Deep Learning Scientists