Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

IBM Power leading Cognitive Systems

644 views

Published on

Why Power Systems?

Published in: Technology
  • Be the first to comment

IBM Power leading Cognitive Systems

  1. 1. © IBM Corporation, 2017 Power Systems is part of IBM Cognitive Systems Accelerated compute and storage delivered on premise, in the cloud or via Watson
  2. 2. © IBM Corporation, 2017© IBM Corporation, 2017 2 The value of computing is changing, from scaling process automation to scaling knowledge via actionable insights
  3. 3. © IBM Corporation, 2017© IBM Corporation, 2017Source: Gartner, “Architecting the On-Demand Digital Business”; Drue Reeves, Kyle Hilgendorf, Kirk Knoernschild, August 16, 2016 3
  4. 4. © IBM Corporation, 2017 IBM Power Systems 4 Designed for cognitive era Power technology designed for big data and analytics workloads Cloud delivery Hyperscale or hybrid cloud with improved economics Open, collaborative innovation platform Enabling cognitive business and cloud economics through Linux and open collaborations Best optimized System for business applications
  5. 5. © IBM Corporation, 2017 Financial sector Retail Telecom Public & security Manufacturing Transportation Defence Oracle SAP & SAP HANA Temenos Flexcube Infosys Infor Comptel Best Economics / TCO Hybrid Cloud Open Source DB / MW Performance guarantee RAS Security Solid roadmap Power Systems represents over 60% of Unix market 5
  6. 6. © IBM Corporation, 2017 2004 2007 2010 2014 POWER8/8NV 22nm POWER6/6+ 65/65 nm Dual Core High Frequencies Virtualization + Memory Subsystem + Altivec Instruction Retry Dynamic Energy Mgmt SMT + Protection Keys POWER5/5+ 130/90 nm Dual Core Enhanced Scaling SMT Distributed Switch + Core Parallelism + FP Performance + Memory Bandwidth + Virtualization Twelve Cores SMT++ On Chip IO Buses Transactional Memory FPGA Support CAPI Encryption PCIe Acceleration NVLINK1.0 (Power8’) POWER7/7+ 45/32 nm Eight Cores On-Chip eDRAM Power-Optimized Cores Memory Subsystem ++ SMT++ Reliability + VSM & VSX Protection Keys+ POWER9 14nm Extreme Analytics Optimization Extreme Big Data Optimization On-chip accelerators NVLINK2.0 Strategy built around a Solid Technology Roadmap 6
  7. 7. © IBM Corporation, 2017 UNSTRUCTURED IN-MEMORY STRUCTURED Flash for extreme performance Massive IO bandwidth Continuous data load Parallel processing Large-scale memory processing Optimized for a broad range of big data & analytics workloads: Processors flexible, fast execution of analytics algorithms Memory large, fast workspace to maximize business insight Cache ensure continuous data load for fast responses 4X threads per core vs. x86 (up to 1536 threads per system) 4X memory bandwidth vs. x861 (up to 16TB of memory) 4X more cache vs. x862 (up to 800MB cache per socket) IBM POWER8 brings performance and scale 7
  8. 8. © IBM Corporation, 2017 POWER8 Product Line Power E870C Power E880C Power E850C Power S822 Power S814, S824 Power S824LPower S812L S822L S812LC S822LC 65% for Scale Out Guaranteed Utilization vs. industry average of 30–40% 80% for Enterprise Power S814 8
  9. 9. © IBM Corporation, 2017 Top 5 Reasons to Upgrade to POWER8 1 2 3 4 5 Greater Performance Significant Cost Savings Increased Capacity, Flexibility, and Scale Seamless Integration and Simple Migration Future Proof, Cloud Ready Solution
  10. 10. © IBM Corporation, 2017 10
  11. 11. © IBM Corporation, 2017 High-end system capacity increases with POWER8 Power E880 offers nearly 40% more performance with 25% fewer cores, enabling workload growth and lowering software costs rPerf performance for fully-configured systems. Power 795 rPerf using 64c LPARs 64c 64c 128c 96c 64c 80c 192c 128c 256c 128c Power 595 ----- Power 795 ----- Power 780 Power 795 Power E870 Power E880 3,906+39%
  12. 12. © IBM Corporation, 2017 Performance per core moves up with POWER8 14 rPerf performance estimate for fully configured systems divided by number of cores rPerfPer-core 595 5GHz 770 3.5GHz 780 4.14GHz TurboCore 795 4.25GHz TurboCore 770 4.2GHz 780 4.4GHz E870 4.19GHz E880 4.35GHz …up to 1.5X over POWER7+ up to 2X over POWER7 up to 2.5X over fastest POWER6 rPerfPer-core
  13. 13. © IBM Corporation, 2017 The Path to High Availability • Full redundancy has always been the best option for high availability • Traditionally, this path has been out of reach for most clients • Power Enterprise Pools makes it possible • Move your applications with Live Partition Mobility and move activations as well! • Supports your cloud environment • Enables workload balancing • Simplifies systems maintenance • Works for disaster recovery • User controlled, always available, no hassle • Move everything 15 A Power Enterprise Pool puts a client in the drivers seat on the road to continuous availability with reduced TCO. Power Enterprise Pool
  14. 14. © IBM Corporation, 2017 What makes IBM Power Systems the best platform for clients’ mission critical SAP HANA deployments? Flexibility  Superior virtualization and management features to afford flexibility and maximum utilization Resiliency  Unsurpassed RAS (reliability, availability, serviceability) characteristics to support mission critical SAP applications Performance  Highest throughput per core to deliver faster business results – up to 2x Intel-based alternatives
  15. 15. © IBM Corporation, 2017 New World record set by IBM Power E870 on SAP BW Enhanced Mixed Load Standard Application Benchmark with 2 Billion records (1) IBM Power Enterprise SystemE870 on the SAP BW Extended mixed load standard applicationbenchmark running SAP Netweaver 7.31 application; 4 processors / 40 cores / 320 threads, POWER8; 4.19GHz, 1024 GB memory, 192.750 adhoc navigationsteps per hours on SuSE Linux Enterprise Server 11 and SAP Hana 1.0, Certification#: 2015024 Result valid as of June 1, 2015. Source: http://www.sap.com/benchmark. (2) Dell PowerEdge R930, on the SAP BW Extended mixed load standard applicationbenchmark running SAP Netweaver 7.31 application; 4 processors / 72 cores / 144 threads, , Intel Xeon Processor E-7 8890 v3 ,2.5 GHz; 1536 GB memory, 172.450 adhoc navigationsteps per hours on SuSE Linux Enterprise Server 11 and SAP Hana 1.0, Certification#: 2015014 (3) Dell PowerEdge R920, on the SAP BW Extended mixed load standard applicationbenchmark running SAP Netweaver 7.31 application; 4 processors / 60 cores / 120 threads, , Intel Xeon Processor E-7 4890 v2 ,2.8 GHz; 1024 GB memory, 137,010 adhoc navigationsteps per hours on SuSE Linux Enterprise Server 11 and SAP Hana 1.0, Certification#: 2014044 (4) HP DL580 Gen8, on the SAP BW Extended mixed load standard applicationbenchmark running SAP Netweaver 7.30 application;4 processors / 60 cores / 120 threads, , Intel Xeon Processor E-7 4880 v2 ,2.5 GHz; 1024 GB memory, 126,980 adhoc navigationsteps per hours on SuSE Linux Enterprise Server 11 and SAP Hana 1.0, Certification#: 2014009 SAP and all SAP logos are trademarks or registered trademarks of SAP AG in Germany and in several other countries. All other product and service names mentionedare the trademarks of their respective companies. 17 “IBM set a world record in the industry leading SAP BW-EML Standard Application Benchmark at 2 billion records ... twice the performance per core over previous benchmarks.” >> Kyle Garman SVP & Managing Director, Global Strategic Partners SAP
  16. 16. © IBM Corporation, 2017 Top Reasons IBM Power Systems are better than Oracle SPARC Servers 18IBM / CustomerXXX Confidential © Copyright IBM Corporation 2017 IBM Power Systems SPARC M7 HW/OS RAS Features Includes all key RAS features: • Redundant system clocks and controllers • Concurrent firmware updates • Dynamic processor sparing…and more Lacks key RAS features: • No concurrent firmware updates • No dynamic processor sparing…and less Performance • Industry-leading performance and benchmarks from 2-16 sockets per server • 2x - 3x greater per core performance over SPARC when leveraging PowerVM • Unsubstantiated performance claims • Exaggerated claims using unrealistic or unavailable configurations • Industry benchmarks show poor per core performance Planned Downtime • LPM can be used to ensure the workloads and users are not disrupted during planned maintenance • OracleVM for SPARC has limited and restricted capabilities for moving workloads. Not available at all on Solaris containers. Requires VCS or Sun Cluster for planned downtime Virtualization • Allows optimization of Oracle licenses to reduce number required by driving sustained utilization levels higher through dynamic resource reallocation • All features are certified and supported for use with Oracle No CPU or memory virtualization: • Logical domains are allocated physical CPUs; No shared processor pools; No micro-partitioning support • No shared memory or memory compression Risk • Proven platform for Oracle workloads, both DB and Applications with many industry references to support reliability and stability of platform • M7 is new Oracle platform; The only enterprise class industry benchmark results show lackluster per core performance. Limited RAS features mean the systems are designed to be clustered using expensive Oracle software Roadmap • History of a clear roadmap with timely execution • Three of the last seven server generations were late; Missed on server capacity and performance targets
  17. 17. © IBM Corporation, 2017 POWER8 vs. SPARC performance 19IBM / CustomerXXX Confidential © Copyright IBM Corporation 2017 Oracle M6-32 result touted as “World Record for 32-Processor Systems” (it had 384 cores) IBM Power Systems E870: • 2.7x more users per core than M6-32 (SPARC M6) • 2.1x more users per core than T7-2 (SPARC M7) • 2.0x more users per core than M7-8 (SPARC M7) 443 365 481 508 997 0 200 400 600 800 1,000 1,200 Oracle M5-32 Oracle M6-32 Oracle T7-2 Oracle M7-8 IBM E870 Users/Core SAP SD Standard Application Benchmark 2-Tier: SD Benchmark Users per core POWER8 2x faster than SPARC M7 (1) Oracle SPARC Server M5-32 on the two-tier SAP SD standard application benchmark running SAP enhancement package 5 for the SAP ERP 6.0 application; 32 processors/192 cores/1536 threads, SPARC M5; 3.60 GHz, 4,096 GB memory; 85,050 SD benchmark users, running Solaris® 11 and Oracle 11g; Certification # 2013009. Source: http://www.sap.com/benchmark. (2) Oracle SPARC Server M6-32 on the two-tier SAP SD standard application benchmark running SAP enhancement package 5 for the SAP ERP 6.0 application; 32 processors/384 cores/3072 threads, SPARC M6; 3.60 GHz, 16 TB memory; 140,000 SD benchmark users, running Solaris® 11 and Oracle 11g; Certification # 2014008. Source: http://www.sap.com/benchmark. (3) Oracle SPARC Server T7-2 on the two-tier SAP SD standard application benchmark running SAP enhancement package 5 for the SAP ERP 6.0 application; 2 processors/64 cores/512 threads, SPARC M7; 4.133 GHz, 1 TB memory; 30,800 SD benchmark users, running Solaris® 11 and Oracle 12c; Certification # 2015050. Source: http://www.sap.com/benchmark. (4) Oracle SPARC Server M7-8 on the two-tier SAP SD standard application benchmark running SAP enhancement package 5 for the SAP ERP 6.0 application; 8 processors/256 cores/2048 threads, SPARC M7; 4.133 GHz, 4 TB memory; 130,000 SD benchmark users, running Solaris® 11 and Oracle 12c; Certification # 2016020. Source: http://www.sap.com/benchmark. (5) IBM Power Enterprise System E870 on the two-tier SAP SD standard application benchmark running SAP enhancement package 5 for the SAP ERP 6.0 application; 8 processors / 80 cores / 640 threads, POWER8; 4.19GHz, 2048 GB memory, 79,750 SD benchmark users running AIX® 7.1 and DB2® 10.5, Certification #: 2014034. Source: http://www.sap.com/benchmark. SAP and all SAP logos are trademarks or registered trademarks of SAP AG in Germany and in several other countries. All other product and service names mentioned are the trademarks of their respective companies.
  18. 18. © IBM Corporation, 2017 IBM Power Systems 20 Designed for cognitive era Power technology designed for big data and analytics workloads Cloud delivery Hyperscale or hybrid cloud with improved economics Open, collaborative innovation platform Enabling cognitive business and cloud economics through Linux and open collaborations Best optimized System for business applications
  19. 19. © IBM Corporation, 2017 IBM Power Systems Enterprise Cloud Offering (C-models) On-Premises Cloud Hybrid Infrastructure Built-in Cloud Deployment Service Options Transform traditional infrastructure with automation, self-service and elastic consumption models Securely extend to Public Cloud with rapid access to compute services and API integration • OpenStack-based Cloud Management: enabling DevOps to Full production • Open source automation: installation and config. recipes • Flexible elastic private cloud capacity and consumption models • Cross Data Center Inventory and Performance Monitoring via the IBM Cloud • Manage VMs across on and off-premises clouds with a single pane of glass (e.g., VMware vRealize) • Securely connect traditional workloads with cloud-native apps (Power & API Connect, BlueMix) • Optional DR as a Service (GDR for Power) • Free access and capacity flexibility with SoftLayer - Free SoftLayer starter pack (12 server months) - Flexibility to run capacity On Premises or in SoftLayer • Design for Cloud Provisioning and Automation • Build for Infrastructure as a Service • Build for Cloud Capacity Pools across Data Centers 5 • Design for Hybrid Cloud with BlueMix • Deliver with automation for DevOps • Deliver with Database as a Service
  20. 20. © IBM Corporation, 2017 IBM Power Systems 22 Designed for cognitive era Power technology designed for big data and analytics workloads Cloud delivery Hyperscale or hybrid cloud with improved economics Open, collaborative innovation platform Enabling cognitive business and cloud economics through Linux and open collaborations Best optimized System for business applications
  21. 21. © IBM Corporation, 2017 Centralized Mainframes Cognitive Era E-Business Distributed Computing Smarter Planet Office Productivity Client/ Server Personal Computer Data Warehousing Big Data & Predictive Analytics Cognitive Data InsightContext Transactional Database Business Intelligence Big Data & Analytics Actionable Insight in context Reporting Our industry continues to transform through waves of change 23
  22. 22. © IBM Corporation, 2017 Level of ROI realized based on implementation of “critical” technologies Q7: What level of ROI have you realized based on your implementation of these technologies? – Institute of Business Value: Who’s leading the cognitive pack in digital operations? Critical technologies are paying off for adopters 24
  23. 23. © IBM Corporation, 2017 Overall Cognitive / Artificial Intelligence (AI) Space 25 Machine Learning Deep Learning Break tasks into Artificial Neural Networks New Data Sources: NoSQL, Hadoop & Analytics New class of applications Opportunity in ML Training  Pattern matching  Image matching (Consumer Photos)  Real-time decision support  Complex workflows  Data Lakes Extend Enterprise applications  Finance: Fraud detection / prevention  Retail: shopping advisors  Healthcare: Diagnostics and treatment  Supply chain and logistics Extend Predictive Analytics to Advance Analytics with AI Human Intelligence Exhibited by Machines Cognitive / ML/DL “Trained” using large amounts of data & ability to learn how to perform the task Growing across Compute, Middleware, and Storage
  24. 24. © IBM Corporation, 2017 With AI, our words will be a window into our mental health https://youtu.be/DnYUNQVcVnI In five years, what we say and write will be indicators of our mental health and physical wellbeing. Patterns in our speech and writing analyzed by cognitive systems will provide doctors and patients tell-tale signs to to predict and track early-stage developmental disorders, mental illness and degenerative neurological diseases more effectively and prompt us to seek treatment. Analyze speech for early detection: Find patterns in speech to accurately predict and monitor psychosis, schizophrenia, mania and depression. Written words provide indicators: Analyze written words to help evaluate our mental health, alerting us to a decline before it occurs. Cognitive assistants for mental health: Cognitive assistants and sensors in our smart phones could “listen” out for our wellbeing Automated mental health tools: Create automated tools to help doctors track and treat the progression of a patient’s neurological disease. 26
  25. 25. © IBM Corporation, 2017 Hyperimaging & AI will give us superhero vision Hyper-imaging and AI will give us superhero vision. In five years, our ability to “see” beyond visible light will reveal new insights to help us understand the world around us. This technology will be widely available throughout our daily lives, giving us the ability to perceive or see through objects and opaque environmental conditions anytime, anywhere. https://youtu.be/TaOOb89ilYk Health and nutrition: Take images of food to show its nutritional value or whether it's safe to eat. Pharmaceuticals: A hyperimage of a pharmaceutical drug could be used to determine if it’s fraudulent. Driving: See through fog or rain and detect black ice or if something is in the road. Gaming: Combine hyperimaging and augmented reality to create video games that allow users to see through objects. 27
  26. 26. © IBM Corporation, 2017 Macroscopes will help us understand Earth's complexity in infinite detail Macroscopes will help us understand Earth's complexity in infinite detail. The physical world before our eyes only gives us a small view into what’s an infinitely interconnected and complex world. Instrumenting and collecting masses of data from every physical object, big and small, and bringing it together will reveal comprehensive solutions for our food, water and energy needs. 28 https://youtu.be/qKMugpYD6tA Organize the IoT: New tools like macroscopes will organize all the world's data -- whether gathered by microscopes, telescopes or everything in between. Search data by time and space: Macroscope technology will be built on platforms that collect and curate geospatial data so it can be easily searched. Transform industries: Macroscopes will reveal new insights about some of the most fundamental problems we face, such as the availability of food, water and energy.
  27. 27. © IBM Corporation, 2017 Medical labs “on a chip” will serve as health detectives for tracing disease at the nanoscale 29 "Medical “labs on a chip” will serve as health detectives for tracing disease at the nanoscale. Nanotechnology “detectives” will let us know we’re unwell before we experience any symptoms. New techniques that detect tiny bio particles found in bodily fluids will reveal clues that, when combined with data from the Internet of Things, will give a full picture of our health. Early detection: Nanobiotechnology techniques will enable the detection of nanoscale-sized clues into our health, letting us know immediately if we have reason to see a doctor. Noninvasive testing: Convenient handheld devices could allow people to quickly and regularly measure the presence of biomarkers found in small amounts of bodily fluids to potentially reveal signs of disease. Know when we aren't contagious: Easily analyze fluids to know when you can no longer pass illnesses like the flu on to others. Complete view of health: Stream information from nanotechnology "detectives" into the cloud where it can be combined with data from other IoT enabled devices, like sleep monitors and smart watches, and analyzed by AI systems for insights. https://youtu.be/c0o0myb7Te4
  28. 28. © IBM Corporation, 2017 Smart sensors will detect environmental pollution at the speed of light Smart sensors will detect environmental pollution at the speed of light. Environmental pollutants won’t be able to hide thanks to new sensing technologies that accurately pinpoint and monitor the quality of our environment. Together with physical analytics combined with artificial intelligence, these technologies will unlock insights to help us prevent pollution and fully harness the promise of cleaner fuels like natural gas. 30 https://youtu.be/8NiVUzKFlzI Detect leaks: Pinpoint invisible methane and other chemical leaks in real-time. New insights: Use AI to analyze and extract sensor data to gain new insights about the spread of pollutants. Complex models: Combine sensor data with IoT data sets to build complex environmental models to detect pollutants. Respiratory illness detection: Detect contaminants in exhaled human breath to improve diagnosis of respiratory disease.
  29. 29. © IBM Corporation, 2017 If the business can dream it… …IT can address it. IT leaders are machine learning heroes who bring cognitive visions to life. Personalize customer journeys Prevent fraud Predict terrorist attacks Prevent disease Cognitive disruption in the enterprise: from big data to intelligent data + fast insights 31
  30. 30. © IBM Corporation, 2017 Cognitive - Data & Analytics Use Cases AUTOMOTIVE Auto sensors reporting location, problems COMMUNICATIONS Location-based advertising CONSUMER PACKAGED GOODS Sentiment analysis of what’s hot, problems $ FINANCIAL SERVICES Risk & portfolio analysis New products EDUCATION & RESEARCH Experiment sensor analysis HIGH TECHNOLOGY / INDUSTRIAL MFG. Mfg. quality Warranty analysis LIFE SCIENCES Clinical trials MEDIA/ENTERTAINMENT Viewers / advertising effectiveness ON-LINE SERVICES / SOCIAL MEDIA People & career matching HEALTH CARE Patient sensors, monitoring, EHRs OIL & GAS Drilling exploration sensor analysis RETAIL Consumer sentiment TRAVEL & TRANSPORTATION Sensor analysis for optimal traffic flows UTILITIES Smart Meter analysis for network capacity, LAW ENFORCEMENT & DEFENSE Threat analysis - social media monitoring, photo analysis 32
  31. 31. © IBM Corporation, 2017 Internet Services Medicine Media & Entertainment Security & Defense Autonomous Machines  Cancer cell detection  Diabetic grading  Drug discovery  Pedestrian detection  Lane tracking  Recognize traffic signs  Face recognition  Video surveillance  Cyber security  Video captioning  Content based search  Real time translation  Image/Video classification  Speech recognition  Natural language processing Deep Learning Is Sweeping Across Industries 33
  32. 32. © IBM Corporation, 2017 IBM Power Systems 34 Designed for cognitive era Power technology designed for big data and analytics workloads Cloud delivery Hyperscale or hybrid cloud with improved economics Open, collaborative innovation platform Enabling cognitive business and cloud economics through Linux and open collaborations Best optimized System for business applications
  33. 33. © IBM Corporation, 2017 Hybrid and private cloud deployments • SAP HANA on Power • HE Cloud models • IBM’s OpenPOWER LC Advanced analytics and open source databases AI, deep learning and machine learning • PowerAI + Minsky • Kinetica • ODP: IBM BigInsights, Hortonworks • EDB, MongoDB, Redis Offerings for the Cognitive Era 35
  34. 34. © IBM Corporation, 2017 Power System Built with Collaborative Innovation Open Source Workloads Now hyper-focused on expanding Cognitive/AI industry applications Enterprise Support/Subscription model OpenPOWER 299 OpenPOWER members contribute to 87 OpenPOWER ready products and 17 servers delivering choice to industry Close partnership with major AI/accelerator industry leader Nvidia 36 Open Frameworks Highly optimized & accelerated Cognitive/AI frameworks Cognitive/AI SDK for deployment and deployment tools OpenCAPI High bandwidth open interconnect to attach to accelerators and SCM Making machine learning and AI more affordable
  35. 35. © IBM Corporation, 2017 Price/Performance Full system stack innovation required Moore’s Law IT innovation can no longer come from just the processor Technology and Processors 2000 2020 Firmware / OS Accelerators Software Storage Network Full Stack Acceleration (Lower is better) The next era of innovation comes from: • Accelerators • I/O Attach Devices • Software Optimization The Cognitive Era demands datacenter innovation 37
  36. 36. © IBM Corporation, 2017 IBM | OpenPOWER Advantage IBM OpenPOWER Foundation Open Source Software
  37. 37. © IBM Corporation, 2017 IBM’s OpenPOWER Servers DIFFERENT BY DESIGN Disrupt your industry without disrupting your datacenter
  38. 38. © IBM Corporation, 2017 IBM OpenPOWER Servers Solve data intensive business challenges Faster insights for data intensive deployments like supply chain optimization, agile asset management, fraud prevention and enterprise data management. Drive datacenter efficiency Compatible with your existing datacenter and cloud environments, delivering superior price-performance for data analytics workloads. Accelerate innovation to become a disruptor Only system ready for the cognitive era with solution-specific hardware accelerators and the POWERAccel family of technologies.
  39. 39. © IBM Corporation, 2017 New “POWER8 with NVLink” Chip First CPU Designed for Accelerated Computing 41 High Performance Cores Fast & Large Memory System Fast PowerAccel Interconnects for Accelerators Faster Cores than x86 Larger Caches Per Core than x86 5x Faster Data Communication between CPU & GPUs CAPI NVLink PCIe P8 POWER8
  40. 40. © IBM Corporation, 2017 PowerAI takes advantage of NVLink between POWER8 & P100 to increase system bandwidth • NVLink between CPUs and GPUs enables fast memory access to large data sets in system memory • Two NVLink connections between each GPU and CPU-GPU leads to faster data exchange P100 GPU Power8 CPU GPU Memory System Memory P100 GPU 80 GB/s GPU Memory NVLink 115 GB/s P100 GPU Power8 CPU GPU Memory System Memory P100 GPU 80 GB/s GPU Memory NVLink 115 GB/s 80 GB/s IBM ”Minsky” Server (S822LC for HPC) 42
  41. 41. © IBM Corporation, 2017 BMW & IBM Partnership to develop in-car AI and IOT services with IBM’ Watson 43 Six distinct areas • Self-healing • Self-socializing • Self-learning • Self-figuring • Self-integrating • Fully autonomous
  42. 42. © IBM Corporation, 2017 Supply Chain Optimization Fraud Prevention Agile Asset Management Enterprise Data Management Solve data intensive business challenges 44
  43. 43. © IBM Corporation, 2017 Performance claims & Proof points 45 IBM Power System S822LC with 4 Tesla P100s –v- Xeon E5-2640 v4 system with 4 Tesla K80s Date of test: Aug 12, 2016 All results are based on running Kinetica “Filter by geographic area” queries on data set of 280 million simulated Tweets with 1 up to 80 simultaneous query streams each with 0 think time Source: https://www.ibm.com/developerworks/linux/perfcol/bigdata.html 2.5xbetter performance IBM Power® System S822LC for Big Data virtualized with Maria DB –v- Intel Xeon E5-2690 v4 running on sysbench workload Results are based on IBM internal testing of single system running multiple virtual machines with Sysbench read-only workload and are current as of August 22, 2016. Performance figures are based on running 24 M record scale factor per VM. Power8 total system throughput produced 15,903 tps and Xeon E5-2690 produced 11,727 tps. Individual results will vary depending on individual workloads, configurations and conditions. Source: https://www.ibm.com/developerworks/linux/perfcol/virtualization.html 1.8xprice-performance advantage 35% more tps/server
  44. 44. © IBM Corporation, 2017 Performance claims & Proof points 46 IBM Power® System S822LC for Big Data delivers 2.12x price performance leadership over Intel Xeon E5-2699 v4 with virtualized EDB Postgres Server 9.5 Results are based on IBM internal testing of single system running multiple virtual machines with pgbench select only work and are current as of August 25, 2016. Performance figures are based on running a 300 scale factor running processor in favor performance modes. Power8 total system throughput produced 577,671 tps and Xeon E5-2699 produced 588,901 tps Source: https://www.ibm.com/developerworks/linux/perfcol/virtualization.html 2.1xprice-performance leadership Power S822LC for HPC with four P100 GPUs using IBM Caffe compared to Intel Xeon E5-2640 v4 with eight M40 GPUs Competitor testing was done on 27-Sep-2016 Power S822LC testing was done on 04-Oct-2016: https://www.ibm.com/developerworks/linux/perfcol/mldl.html 24%faster
  45. 45. © IBM Corporation, 2017 Simplifying the Use of Accelerators 47 Libraries Directives and Abstraction Custom Code module madd_device_module use cudafor contains subroutine madd_dev(a,b,c,sum,n1,n2) real,dimension(:,:),device :: a,b,c real :: sum integer :: n1,n2 type(dim3) :: grid, block !$cuf kernel do (2) <<<(*,*),(32,4)>>> do j = 1,n2 do i = 1,n1 a(i,j) = b(i,j) + c(i,j) sum = sum + a(i,j) enddo enddo end subroutine end module Hardware • Innovation such as Power8 with NVLink and CAPI/OpenCAPI which integrates accelerators in such as way that that are user for software to use. Software • Advancements in directives based software approach and libraries to lessen the effort for software to take advantage of the power of accelerators
  46. 46. © IBM Corporation, 2017© IBM Corporation, 2017 Thank you
  47. 47. © IBM Corporation, 2017 Useful URLs & #tags • Deep Learning on OpenPOWER 49
  48. 48. © IBM Corporation, 2017 Work on these stories (not all AI) • http://www-03.ibm.com/press/us/en/pressrelease/51562.wss • http://www-03.ibm.com/press/us/en/pressrelease/51623.wss • http://www-03.ibm.com/press/us/en/pressrelease/51488.wss 50

×