announcing the nvidia tesla p100 gpu for pcie servers
TRANSCRIPT
14
WORLD’S MOST ADVANCED DATA CENTER GPU FOR HPC
AI: THE NEXT BIG THING FOR HPC
NVIDIA DEEP LEARNING SOFTWARE UPDATES
15
A NEW COMPUTING MODELSomething Big That Will Change the Landscape of HPC
Deep Learning Object DetectionDNN + Data + HPC
Traditional Computer VisionExperts + Time
Deep Learning Achieves “Superhuman” Results
0%10%20%30%40%50%60%70%80%90%
100%
2009 2010 2011 2012 2013 2014 2015 2016
Traditional CVDeep Learning
ImageNet
16
DEEP LEARNING FUELING SCIENCE
Classify Satellite Images for Carbon Monitoring
Analyze Obituaries on the Web for Cancer-related Discoveries
Determine Drug Treatments to Increase Child’s Chance of Survival
NASA AMES
17
ISC KEYNOTE: HPC AND AI
“Investments in computer systems — and I think the bleeding-edge of AI, and deep learning specifically, is shifting to HPC — can cut down the time to run an experiment, and therefore go around the circle, from a week to a day and sometimes even faster.”
— Andrew Ng, Baidu
“…deep learning and cognitively enabled applications are driving large-scale high-performance computing (HPC) projects that are heavier on GPUs. IDC expects major advances and potential large build-outs…”
— IDC
18
K40 K80 + cuDNN1
M40 + cuDNN4
P100 + cuDNN5
0x
10x
20x
30x
40x
50x
60x
70x
BLISTERING PACE
OF INNOVATION
FOR DEEP LEARNING
AlexNet training throughput based on 20 iterations, CPU: 1x E5-2680v3 12 Core 2.5GHz. 128GB System Memory, Ubuntu 14.04
M40 bar: 8x M40 GPUs in a nodeP100: 8x P100 NVLink-enabled
Deep Learning Training PerformanceCaffe AlexNet
2013 2014 2015 2016
Spee
d-up
of
Imag
es/S
ec v
s K4
0 in
201
3
19
WORLD’S MOST ADVANCED DATA CENTER GPU FOR HPC
AI: THE NEXT BIG THING FOR HPC
NVIDIA DEEP LEARNING SOFTWARE UPDATES
20
21
NVIDIA DEEP LEARNING SDKHigh performance GPU-acceleration for deep learning
“We are amazed by the steady stream of improvements made to the NVIDIA Deep Learning SDK and the speedups that they deliver”
— Frédéric Bastien, Team Lead (Theano) MILAdeveloper.nvidia.com/deep-learning-software
Powerful tools and libraries for designing and deploying GPU-accelerated deep learning applications
High performance building blocks for training deep neural networks on NVIDIA GPUs
Accelerated linear algebra subroutines for developing novel deep learning algorithms
Multi-GPU scaling that accelerates training on up to eight GPU
22
Powering the Deep Learning EcosystemNVIDIA SDK Accelerates Every Major Framework
developer.nvidia.com/deep-learning-software
DEEP LEARNING FRAMEWORKS
COMPUTER VISION SPEECH AND AUDIO NATURAL LANGUAGE PROCESSINGObject Detection Voice Recognition Language Translation
Recommendation Engines Sentiment Analysis
Mocha.jl
Image Classification
NVIDIA DEEP LEARNING SDK
NCCLcuDNN cuBLAS GIEcuSPARSE
23
cuDNN 5.1GIEDIGITS 4
WHAT’S NEW IN DEEP LEARNING SOFTWARE
Objection Detection High performance deep learning inference
AUTOMOTIVE
DATA CENTER
EMBEDDED
Improved performance for VGG, ResNet style networks
24
NVIDIA DEEP LEARNING SOFTWARE PLATFORM
AUTOMOTIVE
DATA CENTER
EMBEDDED
NVIDIA DEEP LEARNING SDK
DEVELOP WITH DIGITS DEPLOY WITH GIE
TRAINED NETWORK
TRAININGDATA
TRAINING
DATA MANAGEMENT
MODEL ASSESSMENT
25
OBJECT DETECTIONNew in DIGITS 4
ADVANCED DRIVER ASSISTANCE SYSTEMS (ADAS)
REMOTE SENSING
MEDICAL DIAGNOSTICS
INTELLIGENT VIDEO ANALYTICS
developer.nvidia.com/digits
26
GPU INFERENCE ENGINE (GIE)High-performance deep learning inference for production deployment
developer.nvidia.com/gie
0
1
2
3
4
5
6
7
8
1 8 128
CPU-Only Tesla M4 + GIE
Up to 16x More Inference Perf/Watt
EMBEDDED
Jetson TX1
DATA CENTER
Tesla M4
AUTOMOTIVE
Drive PX
Batch Sizes
GoogLenet, CPU-only vs Tesla M4 + GIE on Single-socket Haswell E5-2698 [email protected] with HT
Imag
es/S
econ
d/W
att