Power 7 Overview
- 1. Enterprise Power Systems Transition to POWER7 technology … the next generation
- 2. Agenda Introductions Jim McGuire, IBM Client Executive – Lockheed Martin Account Tom Lambert, IBM Hardware Sales Leader – Lockheed Martin Account Larry Mills, IBM IT Architect – Lockheed Martin Account Bill Linn, IBM Power Technical Specialist – Lockheed Martin Account Jeff Fier, IBM Computational Scientist Marc West (on phone), IBM HPC Sales Power 7 Introduction – focus on 32 way and higher systems Power 7 750/755 Power 7 770 Power 7 780 IBM HPC solutions Open Discussion on Lockheed Martin’s MM3D Application Server
- 3. IBM Power Systems continues to grow… gaining market share and outselling HP and Sun Industry’s most popular UNIX enterprise servers Sustained performance leadership Leadership virtualization with PowerVM Innovative modular flexibility Application availability through Live Partition Mobility Non-disruptive growth with CoD Broad application support through AIX, IBM i, Linux for Power and Lx86
- 4. IBM POWER Processor Roadmap - 3 Year Revolution - 18 month “+” evolution 2004 2001 2007 2010 POWER4/4+ Dual Core Chip Multi Processing Distributed Switch Shared L2 Dynamic LPARs (32) 180nm, POWER5/5+ Dual Core & Quad Core Md Enhanced Scaling 2 Thread SMT Distributed Switch + Core Parallelism + FP Performance + Memory bandwidth + 130nm, 90nm POWER6/6+ Dual Core High Frequencies Virtualization + Memory Subsystem + Altivec Instruction Retry Dyn Energy Mgmt 2 Thread SMT + Protection Keys 65nm POWER7/7+ 4,6,8 Core 32MB On-Chip eDRAM Power Optimized Cores Mem Subsystem ++ 4 Thread SMT++ Reliability + VSM & VSX Protection Keys+ 45nm, 32nm POWER8 Future First Dual Core in Industry Hardware Virtualization for Unix & Linux Fastest Processor In Industry Most POWERful & Scalable Processor in Industry IBM is the leader in Processor and Server design
- 5. POWER7 Processor Chip Physical Design: 567mm 2 Technology: 45nm lithography, Cu, SOI, eDRAM 1.2B transistors Equivalent function of 2.7B eDRAM efficiency * Statements regarding SMP servers do not imply that IBM will introduce a system with this capability. Features: Eight processor cores 12 execution units per core 4 Way SMT per core 32 Threads per chip 256KB L2 per core 32MB on chip eDRAM shared L3 Dual DDR3 Memory Controllers 100GB/s Memory bandwidth per chip sustained Scalability up to 32 Sockets 360GB/s SMP bandwidth/chip 20,000 coherent operations in flight Two I/O Mezzanine (GX++) System Buses Binary Compatibility with POWER6
- 6. POWER7 Core 64-bit Power ISA Architecture v2.06 Out of Order Execution Wider Dispatch & Issue Capability 6 Wide Dispatch (2 branches per group) 8 Wide issue 12 Execution Units 2 Fixed Point Units 2 Load Store Units also do Simple FX ops 4 Double Precision Floating Point Units 1 Vector Unit 1 Branch 1 Condition Register 1 Decimal Floating Point Unit Units include distributed Recovery Function Cache Design: L1 32KB 4-way set associative I-Cache L1 32KB 8-way set associative D-Cache L1 cache latency reduced from 4 to 2 cycles L2 256KB 8-way associative (i & d combined) Tightly coupled to core. 8 cycles away Add Boxes 256KB L2 IFU CRU/BRU ISU DFU FXU VSX FPU LSU
- 7. The Competition Nehalem-EP (Xeon 5570) for 2 socket servers 45nm, 269mm2, 4 cores, 8MB cache Westmere-EP (Xeon 5680) for 2 socket servers 32nm, 240mm2, 6 cores, 12MB cache Nehalem-EX (Xeon 7560) for up to 8 socket servers 45nm, 684mm2, 8 cores, 24MB cache Tukwila (Itanium 9350) 65nm, 700mm2, 4 cores, 30MB cache
- 8. L3 Cache Memory Controller Fabric Controller Fundamental Computer Architecture POWER7 (Intel Nehalem) 8 cores X 4w SMT (8 core 2w SMT) 32MB L3 EDRAM (24MB SRAM) 360GB/s (103GB/s) 105GB/s Memory BW (32GB/s) 3.86 Ghz (2.26 Ghz) 7 8B SMP interconnect buses 8 DDR3 buffered memory channels 980GB/s Cache BW Core L1 I-cache L1 D-cache L2 Cache Core L1 I-cache L1 D-cache L2 Cache Core L1 I-cache L1 D-cache L2 Cache Core L1 I-cache L1 D-cache L2 Cache Core L1 I-cache L1 D-cache L2 Cache Core L1 I-cache L1 D-cache L2 Cache Core L1 I-cache L1 D-cache L2 Cache Core L1 I-cache L1 D-cache L2 Cache
- 9. L3 Cache Memory Controller Fabric Controller Fundamental Computer Architecture POWER7 Edge 2X cores in SMT 1.3X larger L3 Cache 3.5X IO B/W 3.3X Larger Memory B/W 1.7X higher frequency 7 8B SMP interconnect buses 8 DDR3 buffered memory channels 980GB/s Cache BW Core L1 I-cache L1 D-cache L2 Cache Core L1 I-cache L1 D-cache L2 Cache Core L1 I-cache L1 D-cache L2 Cache Core L1 I-cache L1 D-cache L2 Cache Core L1 I-cache L1 D-cache L2 Cache Core L1 I-cache L1 D-cache L2 Cache Core L1 I-cache L1 D-cache L2 Cache Core L1 I-cache L1 D-cache L2 Cache
- 10. “ Mainframe-class”, a term used by Intel to describe processor enhancements, is derived from the IBM innovations that built the legendary RAS mainframes provide #1,2,3 - See “POWER6 RAS” in backup; See the following URLs for addition details:http://www-03.ibm.com/systems/migratetoibm/systems/power/availability.html http://www-03.ibm.com/systems/migratetoibm/systems/power/virtualization.html The same people who develop mainframes develop Power Systems RAS Feature POWER7™ SPARC Integrity Xeon Application/Partition RAS Live Partition Mobility Yes No No Yes Live Application Mobility Yes No No No Partition Availability priority Yes No No No System RAS OS independent First Failure Data Capture Yes No No No Memory Keys Yes No No No Processor RAS Processor Instruction Retry Yes Yes No No Alternate Processor Recovery Yes No No No Dynamic Processor Deallocation Yes Yes Yes No Dynamic Processor Sparing Yes Yes 2 Yes 2 No Memory RAS Chipkill Yes Yes Yes Yes Survives Double Memory Failures Yes No No No Selective Memory Mirroring Yes No No No Redundant Memory Yes Yes Yes Yes I/O RAS Extended Error Handling Yes No No No
- 11. Alpha Particle Emission Testing IBM’s SOI SRAM cells 9x more robust than bulk SRAM cells* IBM’s SOI latches are 6x more robust than bulk latches* P7 is IBM’s 4 th POWER generation where emission testing was part of the design process to go beyond baseline technology advantages * Statements based on comparisons of IBM’s 65nm SIO and bulk technologies.
- 12. Power Systems with AIX deliver 99.997% uptime - 54% of IT executives and managers say that they require 99.99% or better availability for their applications Power Systems with AIX ® Best availability of UNIX, Linux ® , Windows choices Only 15 minutes of downtime per year Best reliability No severe outages Best serviceability Systems with Call Home have 1/3 the unscheduled repair actions Source: ITIC 2009 Global Server Hardware & Server OS Reliability Survey Results , July 7, 2009
- 13. Security PowerVM has never had a single reported security vulnerability. Source: National Vulnerability Database, http://nvd.nist.gov/
- 14. POWER7 Has Clear Performance Leadership On Major Workloads PER SOCKET vs. Best Published (4/18/2010) Intel Offering * Source: http://www.spec.org/ IBM p570 POWER6 results to be submitted on 5/21/07: All other results as of 04/27/07; ** Source: www.tpc.org/ IBM p570 POWER6 result to be submitted on5/21/07; All other results as of 04/27/07 See next page for full detail
- 15. POWER 7 Workload Optimization Intelligent threads pick ST/SMT2/SMT4 mode of operation Automatic, dynamic movement of thread vs. throughput performance across 2x range Intelligent Cache technology optimizes cache utilization flowing it from core to core Turbo Core Mode enables max core performance for databases Active Cores have a 2X increase in L3 Cache size per core Chips run at increased frequency 3.86 vs 4.14Ghz Provides up to 1.5X per core to core performance gain over P6. Unused core SMT4 SMT2 SMT1 Core Performance Thread 1 1 Thread 1 Thread 2 1.56 2.27 Max core Mode Throughput Orientation incredible parallelization Thread 1 Thread 2 Thread 3 Thread 4 SMT4 SMT1 SMT2 Thread 1 1.19 Thread 1 1.92 Thread 2 Thread 1 Thread 2 Thread 3 Thread 4 2.84 Turbo Core Mode max per core performance for databases 20-25% per core gain
- 16. POWER 7 Dominates Intel’s Offerings POWER 7 vs. Nehalem-EX (Xeon 7560) POWER 7 vs. Westmere-EP (Xeon 5680) POWER 7 vs. Nehalem-EP (Xeon 5570) POWER 7 vs. Tukwila (Itanium 9350) Core Count = 1.33x 2x 2x Micro-Architecture ++ ++ ++ ++++ Frequency 1.7x 1.16x 1.3x 2.25x # of Threads / Core 2x 2x 2x 2x (+ SMT vs. HMT) Cache 1.33x (+ DRAM advantage) 2.67x (+ DRAM advantage) 4x (+ DRAM advantage) 1.1x (+ DRAM advantage) Memory Bandwidth 3x 5x 5x 3x SMP Bandwidth 3.5x (+coherency advantage) 7x (+coherency advantage) 7x (+coherency advantage) 3.5x (+coherency advantage) Max Glueless SMP 4x (32 vs. 8) 16x (32 vs. 2) 16x (32 vs. 2) 4x (32 vs. 8)
- 17. Power is the innovation that will fuel the growth The future of UNIX provides virtualization without limits and exponential ROI Virtualization without Limits increases flexibility and reduces costs Workload-optimizing systems improve service levels with assured performance Consolidation that delivers exponential ROI Dynamic Energy Optimization that balances performance and efficiency Resiliency without Downtime
- 18. IBM Power 755 Server High Performance Computing with POWER7 Smarter Systems for a Smarter Planet
- 19. Focus Application Areas for IBM Power HPC Systems Weather & Environmental models Medical and Life Sciences Basic Research Engineering / Scientific and emerging technologies Predicting the path of the next hurricane Modeling the Human Brain Discovering the secrets of the Universe Tomorrow’s technologies today BRINGING OUR STRENGTH TO BEAR
- 20. HPC Market Opportunity IDC Segment Definitions SuperComputing : Systems configured for large problems and high throughput, >$500K Divisional : Systems for throughput environments, $250K>ASP>$499K Departmental : Systems for throughput environments, $100K<$250K Workgroup: Systems for throughput environments, ASP<$100K Source: IDC Worldwide Technical Computing Server 2009 – 2013 Forecast Update – March, 2009
- 21. Power 755 Market Opportunity HPC has been traditional strength for Power Architecture™: Blue Gene®/P debuted at #2 in Nov, 2007 with 5 systems in the TOP500 Power® 575 debuted at #18 in June, 2008 with 7 POWER6™ systems in the TOP500 Cell/BE debuted at #1 in June, 2008 with 3 Cell/BE systems in the TOP500 Power 755 targets broader market opportunity in Divisional and Departmental Computing segments Source: IDC Worldwide Technical Computing Server 2009 – 2013 Forecast Update – March, 2009
- 22. Power Systems HPC Roadmap Power 755, Blue Gene, Power 575 Mid Range Departmental & Divisional Systems Extreme Scaling Mid to High-end Capability 2010 2012 2011 Blue Gene/Q Power 575 (POWER6™) Blue Gene/P POWER7 (P7 IH) Power 755 Power 755 For information only – subject to change without notice
- 23. IBM Power 755 Performance per node 2X improvement in Single Instruction Multiple Data (SIMD) acceleration Full AltiVec™ (VMX) instruction set support Extended VSX instruction set Up to 8.4 TFlops per Rack (10 nodes per Rack) Cluster Interconnect 2-Port InfiniBand 12X DDR IBM HPC software stack Boost frequency for better performance & performance/watt IBM Power 755 (8236-E8C) 4-socket, 4U server 8-core POWER7 processors 32-core 3.3GHz configuration Up to 256GB of memory Up to 64 clustered nodes Energy Star-qualified GA: 2/19/2010
- 24. 1.7 Times More floating point operations per watt performed by the Power 755 compared to Sun Blade X6440* 1.55 Times More performance than Sun X6270 Blades with Intel’s fastest Xeon X5570 processor on NAMD molecular dynamics simulation The fastest, most energy efficient 4-socket system on the planet * Based on Little Green500. http://www.green500.org IBM Power 755
- 25. Power 755 boosts NAMD performance over Sun X6275 - Xeon X5570 results The Power 755 increases performance by over 55% on NAMD’s molecular dynamics simulations. IBM Power 755 Four Power 755 nodes versus eight X6275 nodes All results are current as of 2/8/2010. IBM Power 7555 with POWER7; 4/32c/32t; 1 -4 nodes results: http://www-03.ibm.com/systems/power/hardware/reports/system_perf.html SUN X6275 with Intel Xeon X5570; 4p/16c/16t 1 – 8 nodes results: http://blogs.sun.com/BestPerf/entry/sun_blade_6048_and_sun1 Sun Sun Sun
- 26. Power 755 delivers superior SPECint_rate2006 performance compared to 4-socket HP x86 servers 2.4X more performance than HP DL585 G6 3.4X more performance than HP DL580 G5 IBM Power 755
- 27. Power 755 delivers superior SPECfp_rate2006 performance compared to 4-socket HP x86 servers 2.9X more performance than HP DL585 G6 5X more performance than HP DL580 G5 IBM Power 755
- 28. Power 755 Rack w/ 8 Nodes (6.8TF) and 76.8 TB (Usable)
- 29. Power 755 Cluster w/ 34 Nodes (28.7 TF) & 326.4TB (Usable)
- 30. Power 755 HPC Cluster Node Support IB-DDR Interconnect Up to 10 Nodes per Rack GA Levels XLF v13.1 VAC/C++ v11.1 Compilers Beta (GA 06/2010) ESSL v5.1 ESSL xCAT v2.3.x GPFS v3.3 PESSL v3.3 LL v4.1 PE v5.2.x HPC Stack Levels AIX 6.1 H Linux Operating Systems 64 nodes (32 Cores/node) 54 TFlops Scaling
- 31. Power 755 – Performance and Energy Efficiency Power 755 targets the growing Divisional and Departmental HPC Segments Weather Reservoir modeling Financial Services Selected Computational Chemistry/Molecular Dynamics Power 755 offers leadership performance 1.7X greater floating point operations per watt performed by the IBM Power 755 compared to Sun Blade X6440 1.55X better NAMD performance than Sun Blade X6275 Greater than 2.9X SPECfp_rate2006 versus HP DL580 and DL585 Power 755 compared to Power 575 delivers: 40% better performance per node 1/3rd less power consumption per node 37% less floor space for a 64 node configuration
- 32. 44 Power your planet. Smarter Systems for a Smarter Planet.
- 33. POWER7 Vector/Scalar Unit 64 Entry Vector/Scalar Register File 128-bit wide registers Used for 32b/64b scalar as well as 4x32B/2x64b SIMD instructions Four floating-point execution units Each FP unit capable of single or double precision Each FP unit can complete a multiply-add instruction per cycle (2FLOPS) Maximum throughput 2 FLOPS x 4 = 8FLOPS/cycle Each FPU can also execute FP divide and sqrt Floating Point Operations are ANSI/IEEE standard 754-1985 Compliant
- 34. IBM Power 770 with POWER7 processors Enhancing the industry’s most popular mid-range server with more performance, energy efficiency and scalability Modular scalability Upgrades from POWER6 Capacity on Demand Live Partition Mobility PowerVM Energy Efficiency IBM extends performance leadership with POWER7 Multi-core delivers more total system AND per core performance Over 4x total system aggregate throughput than 16 core 570’s More performance per core than the 5.0 GHz 570 system Similar portfolio of modular, scalable offerings as POWER6 New workload optimizing features expand flexibility Intelligent Threads™ utilize more threads when workloads benefit Active Memory Expansion™ provides more effective memory The Roadmap to Continuous Availability Redundant system clocks with dynamic fail-over Hot-node Repair available for all nodes * Upgrades available for POWER6 570 systems (9117-MMA) Increased Energy Control and Automation Over 3X improvement in performance per watt Dynamic Energy Optimization maximizes performance or energy efficiency as thermal conditions and policy allow * Hot Node Repair planned for 4Q 2010
- 35. IBM Power 780 with POWER7 processors A new option for growth supporting the highest performance per core and per system with enterprise class features Extreme scalability TurboCore flexibility Capacity on Demand PowerCare Service PowerVM IBM extends performance leadership EVEN MORE Multi-core delivers more total system AND per core performance Over 4.8X total system aggregate throughput than the 5.0 GHz 16 core 570 system Over 2X the performance per core than today’s 32 core 570 New workload optimizing features expand flexibility TurboCore™ for max per core performance for databases Intelligent Threads™ utilize more threads when workloads benefit Active Memory Expansion™ provides more effective memory Enterprise Features 24 x 7 standard warranty PowerCare included with every system Upgrades available for POWER6 570 systems (9117-MMA) The Roadmap for Continuous Availability Redundant system clocks with dynamic fail-over Hot-node Repair available for all nodes * * Hot Node Repair planned for 4Q 2010
- 36. POWER7 Innovations available with Power Modular Systems Technology / Performance for Faster ROI Options for 4-cores, 6 cores, & 8 cores per socket Support for up to 2 TB * DDR3 memory per system Dual memory controllers for increased memory bandwidth Frequency boost for increased performance eDRAM technology for on-chip L3 cache Workload Optimizing Features for Added Flexibility Exclusive new TurboCore mode available with Power 780 Intelligent Threads (SMT-4) for additional capacity Active Memory Expansion TM for increased effective memory Integrated split back plane for dedicated partition support Integrated, separate media controller for partition flexibility Availability Features for Today’s 24 X 7 Workloads Integrated RAID support Redundant clock failover Hot Node Repair for all nodes ** Hot Repair for GX adapters GX++ support for pureScale availability Dynamic Energy Optimization for Increased Efficiency Enhanced TPMD chip for thermal/energy monitoring and control Frequency reduction during low demand for energy reduction SFF SAS drives for increased energy efficiency * Large DIMM planned for 4Q 2010 ** Hot Node Repair planned for 4Q 2010
- 37. IBM Power 780 delivers performance with efficiency 780 delivers over 3X the performance per core of HP Superdome and Sun M9000 780 delivers over 5.8X the performance per watt of HP Superdome and Sun M9000 Performance Per Core Performance Per Watt Source: http://www.spec.org IBM results available at announcement. All other r esults as of 01/27/10. Not all results listed. Performance per KWatt is calculated by dividing the performance by the recommended maximum power usage for site planning. This defines the requirement for the power infrastructure. Actual power used by the systems will be less than this value for all of the systems. For HP systems, this information is contained in the QuickSpecs available through www.hp.com. For Sun systems, this information is available through the respective Site Planning Guides available through www.sun.com. POWER6 HP Superdome Sun M9000 POWER7 POWER6 HP Superdome Sun M9000 POWER7 24,392 44,800 5,600 6,400 Maximum energy requirement (WATTs) 148 26 832 October 2008 16/32/64 IBM Power 570 (4.2 GHz POWER6) 68 12.875 1648 September 2006 64/128/128 HP Integrity Superdome (1.6 GHz Itanium 2) 10.1 39.5 Per core 2586 2530 SPECint_rate2006 58 October 2009 64/256/512 Sun SPARC Enterprise M9000 395 February 2010 8/64/256 IBM Power 780 (3.8 GHz POWER7) Per KWatt Date Chip/Core/Thread System
- 38. Why TurboCore? TurboCore mode extends per core performance Clock speed increases from 3.8 to 4.1 GHz L3 cache doubles from 4 MB per core to 8 MB per core Memory bandwidth per core doubles I/O bandwidth per core doubles Physical memory per core doubles TurboCore mode provides the best option for minimizing software costs Provides over 2X the rPerf per core as the POWER6 32 core 570 TurboCore mode allows clients choice and minimizes risk Clients purchase the system with identical components Clients choose the mode they wish to run Clients can change the mode at any time TurboCore mode eases the transition to highly parallel multi-core systems Start with TurboCore mode when transitioning from POWER6 systems Turn off and grow system when application environment is ready to leverage greater levels of parallel computing
- 39. What is TurboCore? Technology Four of the eight cores and L2 cache are turned off Shared L3 cache is now doubled for remaining cores Remaining cores run at 4.1 GHz frequency Remaining cores share memory and I/O bandwidth Provides up to 22% per core performance gain compared to having all eight cores turned on Unused cores are powered off to improve energy efficiency Implementation 3.8 GHz processors are purchased with Power 780 system as needed Required number of activations are purchased with Power 780 system System is configured by user for TurboCore mode operations System is powered on and cores are made available in TurboCore mode (up to 4 cores per processor) System can be reconfigured by user with a reboot The entire system is either in TurboCore mode or MaxCore mode Unused Core TurboCore Cores Memory Interface P7 Core L2 G X S M P F A B R I C P O W E R B U S L3 Cache P7 Core L2 P7 Core L2 P7 Core L2 P7 Core L2 P7 Core L2 P7 Core L2 P7 Core L2
- 40. POWER7 TurboCore Example Single Node Power 780 system (TurboCore mode) One processor feature #4982 (0 of 16) Two POWER7 processors 64 MB internal L3 cache System is configured for TurboCore mode 8 POWER7 cores @ 4.1 GHz available Up to 8 CoD processor core activation features #5469 Best performance per core configuration Single Node Power 780 system (MaxCore mode) One processor feature #4982 (0 of 16) Two POWER7 processors 64 MB internal L3 cache System is configured for MaxCore mode 16 POWER7 cores @ 3.8 GHz available Up to 16 CoD processor core activation features #5469 Best total system capacity configuration Unused Core TurboCore Cores X X X X X X X X
- 41. POWER7 TurboCore Pricing Example Single Node Power 780 system System configured for TurboCore mode Eight cores are made available @ 4.1 GHz One processor feature #4982 (0 of 16) @ $6.00 Eight CoD activation features #5469 @ $1.00 each 1 x $6.00 + 8 x $1.00 = $14.00 $14.00 / 8 = $1.75 per core Unused Core TurboCore Cores Single Node Power 780 system System configured for MaxCore mode Sixteen cores are made available @ 3.8 GHz One processor feature #4982 (0 of 16) @ $6.00 16 CoD activation features #5469 @ $1.00 each 1 x $6.00 = 16 x $1.00 = $22.00 $22.00 / 16 = $1.375 per core Pricing examples are for illustrative purposes only and do not reflect actual pricing X X X X X X X X
- 42. Designed with the capacity for consolidation 4X memory per core than HP SD 15X memory bandwidth per core than HP SD 5.4X I/O bandwidth per core than HP SD You can use the tremendous capacity of the IBM Power™ 780 to run challenging applications in every virtual server. System data for HP from the HP Superdome Datasheet available at www.hp.com. System data for Sun from the Sun SPARC Enterprise M9000 Datasheet available at www.sun.com . Both are current as of 1/27/2010 Memory per core Memory bandwidth per core I/O bandwidth per core Capacity per core relative to the Power 780
- 43. Save up to 93% in annual energy costs! By consolidating nine 64-core HP Superdomes into ONE Power 780 system --Reduce floor space required by 91% --Reduce processing cores by 88% One Power 780 (@ 75% utilization) 576 total cores @ 1.6 GHz 9 HP Superdomes (@ 25% utilization ) 64 total cores @ 3.8 GHz Only 1 Rack – 7.6 sq. ft of floor space Up to $139k in energy savings per year! See Power 780 comparisons in backup for full substantiation details.
- 44. POWER7 continues to deliver more Performance per Watt rPerf per KWatt >3X increase in performance per watt over POWER6+ >30X increase in performance per watt since POWER4 >10 years of changing the UNIX landscape POWER6™ Power 570 4.2 GHz rPerf: 193.25 KWatts: 5.6 POWER6™ Power 570 4.7 GHz rPerf: 134.35 KWatts: 5.6 POWER7™ Power 780 3.8 GHz rPerf: 685.09 KWatts: 6.4 POWER5+™ p570 1.9 GHz rPerf: 85.20 KWatts: 5.2 POWER5™ p5-570 1.65 GHz rPerf: 68.4 KWatts: 5.2 POWER4+™ p670 1.5 GHz rPerf: 46.79 KWatts: 6.71 POWER4™ p670 1.1 GHz rPerf: 24.46 KWatts: 6.71
- 45. Go Green and Save with IBM EnergyScale Technology Manage your energy costs as never before IBM Systems Director Active Energy Manager™ for POWER exploits POWER7 processor-based EnergyScale™ technology to help you reduce energy consumption >3X improvement in performance per watt over POWER6 570 TPMD chip in every 770/780 system for monitoring and controlling thermal output and energy efficiency Dynamic Energy Optimization reduces energy usage up to 50% when workload or policy allows
- 46. Growth you don’t have to wait on @ $16/day! Enterprise Power Systems offer Capacity on Demand for growth, flexibility and availability Growth Start your system with as few as four cores Grow to 64 cores without disruption Most importantly, grow when you want to: quickly and without disruption to your operation Flexibility Activate resources in increments of one core and one GB memory Chose between permanent activations (purchase) or temporary activations (by the day or by the minute) Prices start as low as $8 per day (per core) and $1 per day (per GB) Try out that new application today for as little as $16 * ($8 for one core and $8 for 8 GB memory for one day) Availability Inactive resources are used for processor and memory sparing Inactive resources can be used for free trials of new applications Inactive resources can be used with temporary activations for emergency backup http://www.ohdeedoh.com/ohdeedoh/flickr-finds/flickr-finds-diy-growth-chart-040628 * US list price at announce for Power 770 with 3.1 GHz processors and AIX
- 47. Move up to Enterprise Class Features Power 750 Power 770 IBM Installed Two dedicated high speed GX++ adapter slots Six dedicated PCI Express adapter slots Up to 32 GB POWER7 memory per core Up to 16 GB standard memory per core Five PCI adapter slots (two PCI X and three PCI Express - two shared) One high speed GX++ adapter and one standard GX adapter slots shared with two PCI Express slots Integrated split backplane support and dedicated media controller Client Installed Capacity on Demand processors start at four cores
- 48. Power Systems Commercial Portfolio – Enterprise and Express BladeCenter Power 750/755 Express Power 770 Power 520 Express Power 595 Enterprise systems Extreme Performance and Scalability Highest memory and I/O bandwidths Capacity on Demand Hot-node Add & Repair Greatest Redundancy POWER7 Memory System Upgrades CE setup & service Granularity of capacity Express systems Choice – AIX, i, and/or Linux Flexibility – Rack, Tower, Blades Industrial Strength Virtualization Power 780
- 49. Move up to enterprise class RAS Optional Standard Not available * Requires two or more nodes ** Planned for 4Q 2010 Alternate Processor Recovery Hot GX Adapter Repair * * Hot-node Repair / Hot-memory Add for all nodes ** * * Dynamic Service Processor and System Clock Failover Memory Sparing * * * * Power 780 Storage Keys Processor Instruction Retry Dual disk controllers (split backplane) PowerVM™/Live Partition Mobility/Live Application Mobility * Redundant System Clocks * Redundant Service Processors Redundant / Hot Swap Power Supplies Concurrent Firmware Update Hot Swap DASD / Media / PCI Adapters * Hot-node Repair / Hot-memory Add Power 750 * Hot-node Add / Cold-node Repair POWER7 Enhanced Memory Redundant / Hot Swap Power Regulators Dynamic Processor Sparing Redundant / Hot Swap Fans & Blowers Hot GX Adapter Add and Cold Repair Power 770 RAS Item
- 50. Two really are better than one! Availability with two nodes Redundant service processors with dynamic failover Redundant system clocks with dynamic failover Additional inactive processors for sparing Additional inactive memory for sparing Allows for continued use if one node fails Hot-node Repair available Performance with two nodes More memory bandwidth More I/O bandwidth Easier growth with CoD Use for balancing workload from other systems Dynamic Energy Optimization assists in maintaining energy efficiency High-availability with two systems Use two systems with PowerHA SystemMirror for high-availability Use Live Partition Mobility for workload balancing and availability during service or upgrades x 2 2
- 51. “… we would have a disk go bad on a Friday night. No one would have known it had happened except for the fact that Electronic Service Agent ™ called out and we got a call back from IBM alerting us to the problem. That protection alone, the risk prevention factor, makes the product hugely valuable to us." Hang up your phone and let your systems do the talking Benefits: System contacts IBM Support for you Immediately uploads error logs Faster diagnosis and time to repair Improved availability for your systems Customized maintenance information End to end, automated, closed loop support process Complimentary installation for P6 570 and P6 595 Features: Automatic reporting for your IBM Systems Secure, encrypted transmission to IBM Support Inventory at your fingertips Enables IBM Electronic Services Built into the HMC, AIX and IBM i for easy installation Nick Gimben, Sr. Systems Admin, Fossil Inc. Power = Support + Automation TM IBM Electronic Service Agent IBM Elec
- 52. What’s in it for me? High availability Secure, 24x7 proactive monitoring: downtime avoidance Less personnel time gathering information and reporting problems Accurate fixes Faster on-site response with parts, location, and problem information Automatic sending of system logs for problem determination and resolution Enables proactive tools My Notifications: customized, proactive recommendations Performance Management: manage system capacity My Systems: compare firmware levels across your datacenter HMC TM IBM Electronic Service Agent IBM Elec
- 53. “We recently completed an analysis of ESA on POWER6 processor-based Power 595 systems. One finding was dramatic: clients who didn’t activate ESA account for 70 percent of unexpected machine outages.” Ross Mauri General Manager IBM Power Systems
- 54. “ ESA and Call Home follow the industry norms for protecting data during network transport by using the Transport Layer Security (TLS) protocol. It also protects Call Home / IBM Support accounts by generating unique passwords for these accounts. Call Home uses protected channels (e.g. TLS, VPN) to transfer data from the HMC to IBM Support. The channels provide confidentiality and integrity protection of the data sent between the two entities.” Proven Security No customer business data transmitted to IBM Connectivity Methods Internet, VPN, Dial-up Proxy & authenticating firewall support Security Protocols https (SSL and TLS) 128 bit encryption; uses keys, certificates and tokens Secure storage System information stored in secure database behind 2 firewalls accessible by you with a protected password Accessible only by authorized IBM Support Representatives Atsec Information Security Corp. Oct 2008 assessment Did you know? 3,344 banks worldwide use Electronic Service Agent… because it’s secure. 11 million+ secure transactions per month 300 million+ total transactions TM IBM Electronic Service Agent IBM Elec
- 55. IBM Power Systems Comparisons * Planned availability in 4Q 2010 No 9 x 5 Standard 1 Up to 11 0.9GB/s 30 GB/s 1 per processor 8.5 GB/s 273 GB/s 8GB Up to 256GB 4 byte 3.3 GHz 32 One Power 755 8 byte 8 byte 4 byte SMP buses 2 per processor 2 per processor 1 per processor Memory controllers Up to four Up to four One Nodes Up to 2 TB* Up to 2 TB* Up to 512 GB System memory 3.8, 4.1 GHz 3.1, 3.5 GHz 3.0, 3.3, 3.55 GHz Frequency Yes 24 x 7 P7 Enhanced Memory Dynamic FSP & clocks Up to 640* Up to 13 3.6 or 7.3 GB/s 236 GB/s 17 or 34 GB/s 1088 GB/s 32 or 64 GB 4 – 64 Power 780 3.6 or 4.9 GB/s 0.9GB/s I/O Bandwidth per core (peak) 17 or 22 GB/s 8.5 GB/s Memory Bandwidth per core (peak) No 9 x 5 P7 Enhanced Memory Dynamic FSP & clocks Up to 640* Up to 11 236 GB/s 1088 GB/s 32 or 42 GB 4 – 64 Power 770 Up to 320* Maximum LPARs No 9 x 5 Standard Up to 11 30 GB/s 273 GB/s 16 or 21 GB 6, 12, 18, 24 or 8, 16, 24, 32 Power 750 Memory Bandwidth (peak) I/O Bandwidth (peak) Cores (single system image) Memory per core rPerf per core RAS Warranty PowerCare
- 56. POWER7 High-end Server Massive throughput, performance and scalability in a new POWER7 high-end system with up to 256 POWER7 processors and support for up to 1,000 partitions Large-scale consolidation of energy-wasting, under-utilized servers onto an energy-efficient high-end POWER7 system running AIX, i and/or Linux applications Improve infrastructure resilience – Enterprise Power Systems & Software are engineered to deliver the highest levels of Power Architecture™ reliability, availability & serviceability Enable rapid service delivery – Industry-leading virtualization and Capacity on Demand for processors and memory help provide seamless, non-disruptive growth Upgrades from Power 595 will enable clients to leverage their investment in POWER6 systems to deploy POWER7 performance, scalability and efficiency within their enterprise Delivering extraordinary scalability, performance and availability for Data Centers with the most demanding Unix, Linux and i applications
- 57. Power is the innovation that will fuel the growth in 2010 Power is effortlessly balancing hundreds of workloads Power is operating at over 90% utilization Power is Management with Automation Power is Intelligent Energy Optimization Power is Virtualization without Limits Power is Resiliency without Downtime
- 58. Backup
- 59. IBM Power 780 comparisons Performance per watt is calculated by dividing the performance in the table above by the recommended maximum power for site planning. Actual power used by the systems will be less than this value for all of the systems. The maximum power requirement for the Power 780 is 6,400 Watts and is available at http://www-01.ibm.com/common/ssi/index.wss - search for Power 780. Power consumption figures of 6400 W for the IBM Power 780, 12,196 W / 24,392 W for the HP Superdome and 44,800 W for the Sun SPARC Enterprise M9000 were based on the maximum rates published by IBM, HP and Sun Microsystems, respectively. The information for the HP Integrity Superdome is in “QuickSpecs HP Integrity Superdome Servers 16- processor, 32-processor, and 64- processor Systems” available at www.hp.com. The information for the Sun SPARC Enterprise M9000 is in the "Sun SPARC Enterprise M9000 Servers Site Planning Guide" available at www.sun.com.
- 60. IBM Power 780 comparisons The virtualized system count and energy savings were derived from several factors: A performance ratio factor was applied to the virtualization scenario based on SPECint_rate2006. The performance factor is simply the SPECint_rate2006 result per core of the Power 780 divided by the per core result of the HP or Sun system. Power 780 (64-core, 8 chips, 8 cores per chip, 3.8 GHz) SPECint_rate2006 2,530 peak as of 2/8/2010. HP Superdome (64-core, 32 chips, 2 cores per chip) 1.6 GHz, SPECint_rate2006 824 peak published October 2006. Sun SPARC Enterprise M9000 (256-core, 64 chips, 4 cores per chip) 2.88 GHz, SPECint_rate2006 2,586 peak published October 2009. SPEC® results available at: www.spec.org A virtualization factor of 3.157X was applied to the virtualization scenario using utilization assumptions derived from an Alinean white paper on server consolidation. The tool assumes 19% utilization of existing servers and 60% utilization of new servers. Source - www.ibm.com/services/us/cio/optimize/opt_wp_ibm_systemp.pdf. Air conditioning power requirement estimated at 50% of system power requirement. Energy cost of $.1031 per kWh is based on 2009 YTD US Average Retail price to commercial customers per US DOE at http://www.eia.doe.gov/cneaf/electricity/epm/table5_6_b.html as of 1/27/2010. The reduction in floor space, power, cooling and software costs depends on the specific customer, environment, application requirements, and the consolidation potential. Actual numbers of virtualized systems supported will depend on workload levels for each replaced system. System data for HP from the HP Superdome Datasheet and HP Integrity Superdome Server — specifications both available at www.hp.com. System data for Sun from the Sun SPARC Enterprise M9000 Tech Specs available at www.sun.com. Data is current as of January 27, 2010.
- 61. Special notices This document was developed for IBM offerings in the United States as of the date of publication. IBM may not make these offerings available in other countries, and the information is subject to change without notice. Consult your local IBM business contact for information on the IBM offerings available in your area. Information in this document concerning non-IBM products was obtained from the suppliers of these products or other public sources. Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products. IBM may have patents or pending patent applications covering subject matter in this document. The furnishing of this document does not give you any license to these patents. Send license inquires, in writing, to IBM Director of Licensing, IBM Corporation, New Castle Drive, Armonk, NY 10504-1785 USA. All statements regarding IBM future direction and intent are subject to change or withdrawal without notice, and represent goals and objectives only. The information contained in this document has not been submitted to any formal IBM test and is provided "AS IS" with no warranties or guarantees either expressed or implied. All examples cited or described in this document are presented as illustrations of the manner in which some IBM products can be used and the results that may be achieved. Actual environmental costs and performance characteristics will vary depending on individual client configurations and conditions. IBM Global Financing offerings are provided through IBM Credit Corporation in the United States and other IBM subsidiaries and divisions worldwide to qualified commercial and government clients. Rates are based on a client's credit rating, financing terms, offering type, equipment type and options, and may vary by country. Other restrictions may apply. Rates and offerings are subject to change, extension or withdrawal without notice. IBM is not responsible for printing errors in this document that result in pricing or information inaccuracies. All prices shown are IBM's United States suggested list prices and are subject to change without notice; reseller prices may vary. IBM hardware products are manufactured from new parts, or new and serviceable used parts. Regardless, our warranty terms apply. Any performance data contained in this document was determined in a controlled environment. Actual results may vary significantly and are dependent on many factors including system hardware configuration and software design and configuration. Some measurements quoted in this document may have been made on development-level systems. There is no guarantee these measurements will be the same on generally-available systems. Some measurements quoted in this document may have been estimated through extrapolation. Users of this document should verify the applicable data for their specific environment. Revised September 26, 2006
- 62. Special notices (cont.) IBM, the IBM logo, ibm.com AIX, AIX (logo), AIX 6 (logo), AS/400, Active Memory, BladeCenter, Blue Gene, CacheFlow, ClusterProven, DB2, ESCON, i5/OS, i5/OS (logo), IBM Business Partner (logo), IntelliStation, LoadLeveler, Lotus, Lotus Notes, Notes, Operating System/400, OS/400, PartnerLink, PartnerWorld, PowerPC, pSeries, Rational, RISC System/6000, RS/6000, THINK, Tivoli, Tivoli (logo), Tivoli Management Environment, WebSphere, xSeries, z/OS, zSeries, AIX 5L, Chiphopper, Chipkill, Cloudscape, DB2 Universal Database, DS4000, DS6000, DS8000, EnergyScale, Enterprise Workload Manager, General Purpose File System, , GPFS, HACMP, HACMP/6000, HASM, IBM Systems Director Active Energy Manager, iSeries, Micro-Partitioning, POWER, PowerExecutive, PowerVM, PowerVM (logo), PowerHA, Power Architecture, Power Everywhere, Power Family, POWER Hypervisor, Power Systems, Power Systems (logo), Power Systems Software, Power Systems Software (logo), POWER2, POWER3, POWER4, POWER4+, POWER5, POWER5+, POWER6, POWER7, pureScale, System i, System p, System p5, System Storage, System z, Tivoli Enterprise, TME 10, TurboCore, Workload Partitions Manager and X-Architecture are trademarks or registered trademarks of International Business Machines Corporation in the United States, other countries, or both. If these and other IBM trademarked terms are marked on their first occurrence in this information with a trademark symbol (® or ™), these symbols indicate U.S. registered or common law trademarks owned by IBM at the time this information was published. Such trademarks may also be registered or common law trademarks in other countries. A current list of IBM trademarks is available on the Web at "Copyright and trademark information" at www.ibm.com/legal/copytrade.shtml The Power Architecture and Power.org wordmarks and the Power and Power.org logos and related marks are trademarks and service marks licensed by Power.org. UNIX is a registered trademark of The Open Group in the United States, other countries or both. Linux is a registered trademark of Linus Torvalds in the United States, other countries or both. Microsoft, Windows and the Windows logo are registered trademarks of Microsoft Corporation in the United States, other countries or both. Intel, Itanium, Pentium are registered trademarks and Xeon is a trademark of Intel Corporation or its subsidiaries in the United States, other countries or both. AMD Opteron is a trademark of Advanced Micro Devices, Inc. Java and all Java-based trademarks and logos are trademarks of Sun Microsystems, Inc. in the United States, other countries or both. TPC-C and TPC-H are trademarks of the Transaction Performance Processing Council (TPPC). SPECint, SPECfp, SPECjbb, SPECweb, SPECjAppServer, SPEC OMP, SPECviewperf, SPECapc, SPEChpc, SPECjvm, SPECmail, SPECimap and SPECsfs are trademarks of the Standard Performance Evaluation Corp (SPEC). NetBench is a registered trademark of Ziff Davis Media in the United States, other countries or both. AltiVec is a trademark of Freescale Semiconductor, Inc. Cell Broadband Engine is a trademark of Sony Computer Entertainment Inc. InfiniBand, InfiniBand Trade Association and the InfiniBand design marks are trademarks and/or service marks of the InfiniBand Trade Association. Other company, product and service names may be trademarks or service marks of others. Revised February 9, 2010
- 63. Notes on benchmarks and values The IBM benchmarks results shown herein were derived using particular, well configured, development-level and generally-available computer systems. Buyers should consult other sources of information to evaluate the performance of systems they are considering buying and should consider conducting application oriented testing. For additional information about the benchmarks, values and systems tested, contact your local IBM office or IBM authorized reseller or access the Web site of the benchmark consortium or benchmark vendor. IBM benchmark results can be found in the IBM Power Systems Performance Report at http://www.ibm.com/systems/p/hardware/system_perf.html . All performance measurements were made with AIX or AIX 5L operating systems unless otherwise indicated to have used Linux. For new and upgraded systems, AIX Version 4.3, AIX 5L or AIX 6 were used. All other systems used previous versions of AIX. The SPEC CPU2006, SPEC2000, LINPACK, and Technical Computing benchmarks were compiled using IBM's high performance C, C++, and FORTRAN compilers for AIX 5L and Linux. For new and upgraded systems, the latest versions of these compilers were used: XL C Enterprise Edition V7.0 for AIX, XL C/C++ Enterprise Edition V7.0 for AIX, XL FORTRAN Enterprise Edition V9.1 for AIX, XL C/C++ Advanced Edition V7.0 for Linux, and XL FORTRAN Advanced Edition V9.1 for Linux. The SPEC CPU95 (retired in 2000) tests used preprocessors, KAP 3.2 for FORTRAN and KAP/C 1.4.2 from Kuck & Associates and VAST-2 v4.01X8 from Pacific-Sierra Research. The preprocessors were purchased separately from these vendors. Other software packages like IBM ESSL for AIX, MASS for AIX and Kazushige Goto’s BLAS Library for Linux were also used in some benchmarks. For a definition/explanation of each benchmark and the full list of detailed results, visit the Web site of the benchmark consortium or benchmark vendor. TPC http://www.tpc.org SPEC http://www.spec.org LINPACK http://www.netlib.org/benchmark/performance.pdf Pro/E http://www.proe.com GPC http://www.spec.org/gpc VolanoMark http://www.volano.com STREAM http://www.cs.virginia.edu/stream/ SAP http://www.sap.com/benchmark/ Oracle Applications http://www.oracle.com/apps_benchmark/ PeopleSoft - To get information on PeopleSoft benchmarks, contact PeopleSoft directly Siebel http://www.siebel.com/crm/performance_benchmark/index.shtm Baan http://www.ssaglobal.com Fluent http://www.fluent.com/software/fluent/index.htm TOP500 Supercomputers http://www.top500.org/ Ideas International http://www.ideasinternational.com/benchmark/bench.html Storage Performance Council http://www.storageperformance.org/results Revised March 12, 2009
- 64. Notes on HPC benchmarks and values Revised March 12, 2009 The IBM benchmarks results shown herein were derived using particular, well configured, development-level and generally-available computer systems. Buyers should consult other sources of information to evaluate the performance of systems they are considering buying and should consider conducting application oriented testing. For additional information about the benchmarks, values and systems tested, contact your local IBM office or IBM authorized reseller or access the Web site of the benchmark consortium or benchmark vendor. IBM benchmark results can be found in the IBM Power Systems Performance Report at http://www.ibm.com/systems/p/hardware/system_perf.html . All performance measurements were made with AIX or AIX 5L operating systems unless otherwise indicated to have used Linux. For new and upgraded systems, AIX Version 4.3 or AIX 5L were used. All other systems used previous versions of AIX. The SPEC CPU2000, LINPACK, and Technical Computing benchmarks were compiled using IBM's high performance C, C++, and FORTRAN compilers for AIX 5L and Linux. For new and upgraded systems, the latest versions of these compilers were used: XL C Enterprise Edition V7.0 for AIX, XL C/C++ Enterprise Edition V7.0 for AIX, XL FORTRAN Enterprise Edition V9.1 for AIX, XL C/C++ Advanced Edition V7.0 for Linux, and XL FORTRAN Advanced Edition V9.1 for Linux. The SPEC CPU95 (retired in 2000) tests used preprocessors, KAP 3.2 for FORTRAN and KAP/C 1.4.2 from Kuck & Associates and VAST-2 v4.01X8 from Pacific-Sierra Research. The preprocessors were purchased separately from these vendors. Other software packages like IBM ESSL for AIX, MASS for AIX and Kazushige Goto’s BLAS Library for Linux were also used in some benchmarks. For a definition/explanation of each benchmark and the full list of detailed results, visit the Web site of the benchmark consortium or benchmark vendor. SPEC http://www.spec.org LINPACK http://www.netlib.org/benchmark/performance.pdf Pro/E http://www.proe.com GPC http://www.spec.org/gpc STREAM http://www.cs.virginia.edu/stream/ Fluent http://www.fluent.com/software/fluent/index.htm TOP500 Supercomputers http://www.top500.org/ AMBER http://amber.scripps.edu/ FLUENT http://www.fluent.com/software/fluent/fl5bench/index.htm GAMESS http://www.msg.chem.iastate.edu/gamess GAUSSIAN http://www.gaussian.com ANSYS http://www.ansys.com/services/hardware-support-db.htm Click on the "Benchmarks" icon on the left hand side frame to expand. Click on "Benchmark Results in a Table" icon for benchmark results. ABAQUS http://www.simulia.com/support/v68/v68_performance.php ECLIPSE http://www.sis.slb.com/content/software/simulation/index.asp?seg=geoquest& MM5 http://www.mmm.ucar.edu/mm5/ MSC.NASTRAN http://www.mscsoftware.com/support/prod%5Fsupport/nastran/performance/v04_sngl.cfm STAR-CD www.cd-adapco.com/products/STAR-CD/performance/320/index/html NAMD http://www.ks.uiuc.edu/Research/namd HMMER http://hmmer.janelia.org/ http://powerdev.osuosl.org/project/hmmerAltivecGen2mod
- 65. Notes on performance estimates rPerf for AIX rPerf (Relative Performance) is an estimate of commercial processing performance relative to other IBM UNIX systems. It is derived from an IBM analytical model which uses characteristics from IBM internal workloads, TPC and SPEC benchmarks. The rPerf model is not intended to represent any specific public benchmark results and should not be reasonably used in that way. The model simulates some of the system operations such as CPU, cache and memory. However, the model does not simulate disk or network I/O operations. rPerf estimates are calculated based on systems with the latest levels of AIX and other pertinent software at the time of system announcement. Actual performance will vary based on application and configuration specifics. The IBM eServer pSeries 640 is the baseline reference system and has a value of 1.0. Although rPerf may be used to approximate relative IBM UNIX commercial processing performance, actual system performance may vary and is dependent upon many factors including system hardware configuration and software design and configuration. Note that the rPerf methodology used for the POWER6 systems is identical to that used for the POWER5 systems. Variations in incremental system performance may be observed in commercial workloads due to changes in the underlying system architecture. All performance estimates are provided "AS IS" and no warranties or guarantees are expressed or implied by IBM. Buyers should consult other sources of information, including system benchmarks, and application sizing guides to evaluate the performance of a system they are considering buying. For additional information about rPerf, contact your local IBM office or IBM authorized reseller. ======================================================================== CPW for IBM i Commercial Processing Workload (CPW) is a relative measure of performance of processors running the IBM i operating system. Performance in customer environments may vary. The value is based on maximum configurations. More performance information is available in the Performance Capabilities Reference at: www.ibm.com/systems/i/solutions/perfmgmt/resource.html Revised April 2, 2007
Editor's Notes
- #5: POWER7 will deliver new features and functions to the Power family of Processors The enhancements include: Additional core density On chip cache using the energy saving technology developed in IBM Research Energy efficient core New on chip memory controller technology providing support for DDR3 memory. The memory will deliver more than 3X memory bandwidth of the POWER6 chip Support for both single and dual precision SIMD processing Support for additional Storage Protection Keys
- #6: Let’s take a closer look now at the POWER7 Chip. POWER7 is fabricated in IBMs 45nm Silicon on insulator technology using copper interconnect and embedded dram for the L3. The chip is 567mm square and contains 1.2B transistors. However if you consider each EDRAM cell has the function of a 6T SRAM cell the chip actually has the equivalent function of a 2.7B transistors chip. The chip as you can see has 8 processor cores each with 12 execution unit capable of running 4 way SMT. I’ll share some core details in a few slide. To feed the processor cores: We have two memory controllers one on each side of the chip. Each memory controller supports 4 channels of DDR3 memory. Combined these 8 channels provided 100GBs sustained memory bandwidth. On the top and bottom of the chip are our seven 8 byte multi-processor links providing 360GB/s bandwidth to make balanced SMP systems scalable to 32 sockets.
- #7: Next let’s take a closer look at the POWER7 core. The smaller P7 core provides additional performance over our previous generation Power6 core by: Having a shorter wider pipeline with better utilization leveraging SMT4 and Out of order execution The net is higher performance even with a smaller core in equivalent technology thus saving power. Taking a look at the chip floor plan you can see. Two fixed point pipelines. The two LSU pipes , The load store pipes are also capable executing simple fixed point instructions. FPU pipelines capable of 4 double precision multiply add operations per cycle or 8 flops/cycle. This unit also handles vector instructions. The instruction fetch unit which also executes branch and condition register instructions. The decimal floating point unit < A widened Instruction sequencing unit capable of dispatching 6 instructions per cycle including 2 branches and issuing up to 8 instructions per cycle. <click> In POWER7 we took advantage of the out of order execution to switch from a dedicated recovery unit to a distributed one using the flush and refetch capability in the OOO machine. The core caches on Power7 have been improved by making the L1 instruction and data cache 32KB and reducing the access time from 4 to 2 cycles and backing them with a 256KB L2 cache integrated with the core to be only 8 cycles away.
- #8: EP – cut backs EX – enterprise ITF will die Itanic
- #11: Our RAS results are better because we start with a full systems view. We have very challenging for each element of RAS and we measure our systems performance. as we approach a new generation of Power, we attack those elements which have had the greatest impact on reliability, availability, or serviceability. we can do this because we design the HW, firmware and OS together. Just as an example, look at the way we address processor execution errors. before an instruction executes, we save status information about the processor. if the instruction fails for any reason, we reload that status and retry the instruction using Processor Intstruction Retry. Most of the time, the instruction will work because most problems are intermittent caused by events like bombardment of the chip by alpha particles flipping a bit. This kind of event becomes more common as we make technology denser and the size of the alpha particles becomes larger relative to the distance between bits in execution reguisters. Some of the time, retry doesn’t work on the processor because it has a hard failure. In that case, if another processor is available, we use Alternate Processor Recovery andload the status into the other processor and try there to avoid any application outage. Hardware Instruction Retry requires cooperation between HW and firmware. Alternate Processor Recovery requires the additional cooperation of the OS. We develop all of them so we include that cooperation. Itanium and x86 systems have neither. We have similar features throughout the system. As you can see in the chart above, if Xeon does get all the RAS features itanium has, it will be an improvement, but it will still leave Xeon based systems well behind Power systems.
- #13: Now let’s look at reliability, availability, and serviceability. A recent survey (independent - not vender funded) of 400 IT users worldwide by ITIC showed that the combination of AIX and Power Systems provides the best result in each of these categories. Our availability is 99.997% - 2 ½ times the next best UNIX alternative and 10 times better than Windows on x86. 54% of IT execs surveyed say they need 99.99% availability or better. With these kind of results it is no wonder that more and more of them are choosing Power systems. Note that Solaris on SPARC has better availability than Linux on x86. If your client moves to x86, they will be taking a step backward.
- #15: To really put balanced performance into perspective, however, we have placed four of the leading performance results on this one chart. The telling statistic, however, is that regardless of workload, POWER7 technology performs. This means that regardless of your workload, POWER7 systems can deliver industry leading performance for your business. This means that you no longer have to buy specialized systems for different workloads. This means that you can feel safe in consolidating multiple workloads onto the p770, knowing that you will get the best possible performance for each of your applications
- #18: Virtualization without Limits increases flexibility and reduces costs: Expanded system capability teamed with PowerVM’s performance, scalability and flexibility Workload-optimizing systems improve service levels with assured performance: PowerVM, Intelligent Threads and TurboCore mode enable you to optimize the performance of your workloads in a virtualized environment Consolidation that delivers exponential ROI: Industry’s leading performance, scalability and virtualization now unbounded with DB2 pureScale Dynamic Energy Optimization that balances performance and efficiency: >3X increased performance per watt, new EnergyScale features integrated with Active Energy Manager Resiliency without Downtime: Non-disruptive application upgrades from POWER6 and the improvements to the road for Continuous Availability
- #19: Good morning. This morning I am going to take you through a presentation that cover the POWER7 Express rack and tower roll out. With a focus on the products being announced in Feb.. I will talk about how there positioned and how they compared to legacy Power products as well as competitive servers. There will be a session with Patrick O’rourke tomorrow morning that will go into technical detail on all POWER7 offering and features.
- #23: Ian…FAST flash
- #24: The Power 755 is a 4 socket, 4U rack-optimized server supporting 8-core POWER7 processors and up to 256GB of memory. The Power 755 is a high performance compute node targeted at small to mid-size clusters. It delivers better than 3X improvement in power than current power offerings. POWER7 processors support AltiVec™ instruction set and extended VSX SIMD (single instruction multiple data) acceleration which can execute up to eight single-precision or double-precision floating point operations per clock cycle per core to improve fine-grained parallelism and accelerate data processing. IBM HPC software stack has the development tools, libraries, file systems and system management software necessary to administer a Power 755 server cluster. There will be a HPC technical session on the Power 755 on Wed..
- #46: On Slide 7, Clients will now be able to add the p520 and p550 to the list of POWER6 based servers that they can monitor and manage via IBM system director for active energy manager. The ability to collect and report power consumption and system themal data and leverage features such as power capping to control system power under a specified limit and power saving to reduce power when workload or policy allows. The POWER6 chip is also designed to conserve power and reduce heat generated by the server. A feature called “nap” mode enables processor clocks to be dynamically turned off when there is no useful work to be done and turned back on when there are instructions to be executed. This features is supported on all Power6 systems In summary, the POWER6 architecture with Acitve energy manager and POWER6 EnergyScale technology can help clients improve energy efficiency and reduce energy related costs .