SlideShare a Scribd company logo

H2O World 2017 Keynote - Jim McHugh, VP & GM of Data Center, NVIDIA

•
•
Sri Ambati
Sri Ambati

Presented at #H2OWorld 2017 in Mountain View, CA. Enjoy the recording: https://youtu.be/NyaJ7uDroww. Learn more about H2O.ai: https://www.h2o.ai/. Follow @h2oai: https://www.twitter.com/h2oai.

1 of 23
Download to read offline
Jim McHugh | NVIDIA VP and GM of Data Center | December 2017 NVIDIA GPU ACCELERATED COMPUTING
2 NVIDIA GPU Computing Gaming VR AI & HPC Self-Driving Cars
3 GPU COMPUTING AT THE HEART OF AI Big Bang of Modern AI 103 105 107 1.5X per year 40 Years of CPU Trend Data Single-threaded perf GPU-Computing perf 1.5X per year 1.1X per year 1000X by 2025 Performance Beyond Moore’s Law Original data up to the year 2010 collected and plotted by M. Horowitz, 
 F. Labonte, O. Shacham, K. Olukotun, L. Hammond, and C. Batten New plot and data collected for 2010-2015 by K. Rupp 1980 1990 2000 2010 2020
4 RISE OF NVIDIA GPU COMPUTING Global GTC Attendees 10X in 5 Years GPU Developers 16X in 5 Years CUDA Downloads 6X in 5 Years 2012 2017 2M670,00022,000 2012 2017 2012 2017
5 ML & DL CHALLENGES Slow Processes Model Complexity Resource Constraints Hours to build GBT 30+ Days Data Transformation Weeks Feature Engineering Months Scoring Pipelines <100Grandmasters in the world
6 REVOLUTIONARY 
 AI PERFORMANCE Performance up to 100 CPUs 21 billion transistors – 5120 CUDA cores New Tensor Core architecture inspired 
 by the demands of deep learning Volta is the Most Advanced Data Center GPU Ever Built
7 MAXIMIZING PERFORMANCE ON VOLTA GPU Generational Training Scaling 0 3 6 9 12 K80 V100 Tensor Core ResNet-152 Training, 8x K80 (16 GPUs total) compared with 8x V100 NVLink GPUs using NVIDIA 17.10 containers Greater Than 10x Performance K80 vs. V100
8 NVIDIA TENSORRT 
 10X BETTER DATA CENTER TCO 160 CPU Servers 45,000 Images / Second 65 KWatts
9 NVIDIA TENSORRT 
 10X BETTER DATA CENTER TCO 1 NVIDIA HGX with 8 Tesla V100 GPUs 45,000 Images / Second 3 KWatts 1/6 the Cost | 1/20 the Power 4 Racks in a Box
10 NVIDIA VOLTA IN EVERY CLOUD, 
 EVERY DATACENTER
11 NVIDIA DGX FAMILY AI Workstation DGX Station
 The Personal AI Supercomputer DGX-1
 The Essential Instrument for AI Research Cloud-Scale AIAI Datacenter NVIDIA GPU Cloud Cloud Service with the Highest Deep Learning Efficiency
12
13 THREE STEPS TO NVIDIA GPU CLOUD www.nvidia.com/ngc SIGN UP To get an NGC account, 
 go to: www.nvidia.com/ngcsignup DEPLOY IMAGE On Amazon EC2, choose a P3 instance and deploy the NVIDIA Volta Deep Learning AMI for NGC PULL CONTAINER Pick your desired framework and pull the container into your instance 1 2 3
14 ANNOUNCING H2O DRIVERLESS AI 
 ON NVIDIA GPU CLOUD Up to 40X Faster
15
16 H2O4GPU PERFORMANCE GLM XGBoost K-Means 40x10x5x
17 NVIDIA GPU ACCELERATED 
 DRIVERLESS AI ADOPTION Downloads Activation Installs 1250+ 300+600+
18 COLLABORATION THROUGH OPEN SOURCE gpuopenanalytics.com GPU DATA FRAME | APACHE ARROW
19 ML ACROSS INDUSTRIES Finance Healthcare Telco
20 NVIDIA + H2O CommunityEngineering GPU DATA FRAME | APACHE ARROWH2O Driverless AI on DGX and NGC Business Joint Sales and Marketing Global Initiative $40m Series C Round of Funding
21 FOR MORE INFORMATION Webinar: Interpretability with Driverless AI Webinar: Accelerated Machine Learning on GPUs Blog: GOAI: Open GPU Accelerated Data Analytics Blog: Crossing the Chasm of AI Interpretability Blog: H2O Offers GPU-Accelerated ML for Enterprise Transformation Solution Brief: NVIDIA and H2O: GPU-Accelerated ML
22 A NEW COMPUTING ERA Performance Beyond Moore’s law Driverless AI on DGX and NGC Cross-Industry Solutions for ML
23

More Related Content

H2O World 2017 Keynote - Jim McHugh, VP & GM of Data Center, NVIDIA

  • 1. Jim McHugh | NVIDIA VP and GM of Data Center | December 2017 NVIDIA GPU ACCELERATED COMPUTING
  • 2. 2 NVIDIA GPU Computing Gaming VR AI & HPC Self-Driving Cars
  • 3. 3 GPU COMPUTING AT THE HEART OF AI Big Bang of Modern AI 103 105 107 1.5X per year 40 Years of CPU Trend Data Single-threaded perf GPU-Computing perf 1.5X per year 1.1X per year 1000X by 2025 Performance Beyond Moore’s Law Original data up to the year 2010 collected and plotted by M. Horowitz, 
 F. Labonte, O. Shacham, K. Olukotun, L. Hammond, and C. Batten New plot and data collected for 2010-2015 by K. Rupp 1980 1990 2000 2010 2020
  • 4. 4 RISE OF NVIDIA GPU COMPUTING Global GTC Attendees 10X in 5 Years GPU Developers 16X in 5 Years CUDA Downloads 6X in 5 Years 2012 2017 2M670,00022,000 2012 2017 2012 2017
  • 5. 5 ML & DL CHALLENGES Slow Processes Model Complexity Resource Constraints Hours to build GBT 30+ Days Data Transformation Weeks Feature Engineering Months Scoring Pipelines <100Grandmasters in the world
  • 6. 6 REVOLUTIONARY 
 AI PERFORMANCE Performance up to 100 CPUs 21 billion transistors – 5120 CUDA cores New Tensor Core architecture inspired 
 by the demands of deep learning Volta is the Most Advanced Data Center GPU Ever Built
  • 7. 7 MAXIMIZING PERFORMANCE ON VOLTA GPU Generational Training Scaling 0 3 6 9 12 K80 V100 Tensor Core ResNet-152 Training, 8x K80 (16 GPUs total) compared with 8x V100 NVLink GPUs using NVIDIA 17.10 containers Greater Than 10x Performance K80 vs. V100
  • 8. 8 NVIDIA TENSORRT 
 10X BETTER DATA CENTER TCO 160 CPU Servers 45,000 Images / Second 65 KWatts
  • 9. 9 NVIDIA TENSORRT 
 10X BETTER DATA CENTER TCO 1 NVIDIA HGX with 8 Tesla V100 GPUs 45,000 Images / Second 3 KWatts 1/6 the Cost | 1/20 the Power 4 Racks in a Box
  • 10. 10 NVIDIA VOLTA IN EVERY CLOUD, 
 EVERY DATACENTER
  • 11. 11 NVIDIA DGX FAMILY AI Workstation DGX Station
 The Personal AI Supercomputer DGX-1
 The Essential Instrument for AI Research Cloud-Scale AIAI Datacenter NVIDIA GPU Cloud Cloud Service with the Highest Deep Learning Efficiency
  • 12. 12
  • 13. 13 THREE STEPS TO NVIDIA GPU CLOUD www.nvidia.com/ngc SIGN UP To get an NGC account, 
 go to: www.nvidia.com/ngcsignup DEPLOY IMAGE On Amazon EC2, choose a P3 instance and deploy the NVIDIA Volta Deep Learning AMI for NGC PULL CONTAINER Pick your desired framework and pull the container into your instance 1 2 3
  • 14. 14 ANNOUNCING H2O DRIVERLESS AI 
 ON NVIDIA GPU CLOUD Up to 40X Faster
  • 15. 15
  • 17. 17 NVIDIA GPU ACCELERATED 
 DRIVERLESS AI ADOPTION Downloads Activation Installs 1250+ 300+600+
  • 18. 18 COLLABORATION THROUGH OPEN SOURCE gpuopenanalytics.com GPU DATA FRAME | APACHE ARROW
  • 20. 20 NVIDIA + H2O CommunityEngineering GPU DATA FRAME | APACHE ARROWH2O Driverless AI on DGX and NGC Business Joint Sales and Marketing Global Initiative $40m Series C Round of Funding
  • 21. 21 FOR MORE INFORMATION Webinar: Interpretability with Driverless AI Webinar: Accelerated Machine Learning on GPUs Blog: GOAI: Open GPU Accelerated Data Analytics Blog: Crossing the Chasm of AI Interpretability Blog: H2O Offers GPU-Accelerated ML for Enterprise Transformation Solution Brief: NVIDIA and H2O: GPU-Accelerated ML
  • 22. 22 A NEW COMPUTING ERA Performance Beyond Moore’s law Driverless AI on DGX and NGC Cross-Industry Solutions for ML
  • 23. 23