SPARC T3, X86, Virtualisierung Und Management Elke Freymann & Stefan Hinker Oracle Hardware Sales Consulting

Total Page:16

File Type:pdf, Size:1020Kb

SPARC T3, X86, Virtualisierung Und Management Elke Freymann & Stefan Hinker Oracle Hardware Sales Consulting <Insert Picture Here> Sun Server News: SPARC T3, x86, Virtualisierung und Management Elke Freymann & Stefan Hinker Oracle Hardware Sales Consulting The following is intended to outline our general product direction. It is intended for information purposes only, and may not be incorporated into any contract. It is not a commitment to deliver any material, code, or functionality, and should not be relied upon in making purchasing decisions. The development, release, and timing of any features or functionality described for Oracle’s products remains at the sole discretion of Oracle. © 2010 Oracle Corporation and/or its affiliates. All rights reserved 3 Agenda • SPARC T3 – Acht sind nicht genug <Insert Picture Here> • x86 – Wenn acht nicht reichen, nehmen Sie zwei! • Virtualisierung – Viele Flöhe, kein Aufpreis • Management – Zum Flöhe hüten bestens geeignet © 2010 Oracle Corporation and/or its affiliates. All rights reserved 4 Agenda • SPARC T3 – Acht sind nicht genug • x86 – Wenn acht nicht reichen, nehmen Sie zwei! • Virtualisierung – Viele Flöhe, kein Aufpreis • Management – Zum Flöhe hüten bestens geeignet 5 UltraSPARC T1 – Die erste 8-Kern CPU UltraSPARC-T1 Details Process: 90nm Transistors: 300M Clock Speed: 1.2GHz Cores/Socket: 8 Threads/Core: 4 Thread Model: CMT Oracle multiplier: 0.25 Memory: DDR2-533 L1$: 16K I$ 8K D$ / core L2$: 3MB shared L3$: None Power: 79W maximum 6 2005 7 UltraSPARC T2: Server on a Chip 42 GB/s read, 21 GB/s write 2–8 DIMMs Dual- Dual- Dual- Dual- • 8 SPARC V9 cores @ 1.2–1.4GHz channel channel channel channel > 8 threads per core FB-DIMM FB-DIMM FB-DIMM FB-DIMM > 2 execution pipelines per core > 1 instruction/cycle per pipeline x10 write x14 read > 1 FPU per core @ 4.0 GT/s > 1 SPU (crypto) per core Memory Memory Memory Memory controller controller controller controller > 4 MB, 16-way, 8-bank L2$ L2$L2$ BankL2$ L2$L2$ BankL2$ L2$L2$ BankL2$ L2$L2$ BankL2$ Bank Bank Bank Bank Bank Bank Bank Bank 4 MB L2$ • 4 FB-DIMM DRAM controllers CrossbarCrossbar 16 KB I$ 16 KB I$ 16 KB I$ 16 KB I$ 16 KB I$ 16 KB I$ 16 KB I$ 16 KB I$ • 2.5 GHz x 8 PCI-Express interface 8 KB D$ 8 KB D$ 8 KB D$ 8 KB D$ 8 KB D$ 8 KB D$ 8 KB D$ 8 KB D$ • 2 x 10 Gb on-chip Ethernet FPU FPU FPU FPU FPU FPU FPU FPU SPU SPU SPU SPU SPU SPU SPU SPU C1 C2 • Technology: TI 65nm New C3 C4 C5 C6 C7 C8 8 threads per core 2 • Die size: 342mm 2 execution pipes 1 op/cycle per pipe Sys I/F • Power: < 95 W (nominal) buffer NI switch PCIe U core 10 Gb Ethernet SSI, JTAG X8 @ 2.5 GHz Debug port 2 GB/s each direction 8 2007 9 Weltweit erster 16-Kern Server Prozessor • 16 Kerne, 32 Pipelines, 128 Threads • Verschlüsselung integriert • PCIe 2.0 integriert © 2010 Oracle Corporation and/or its affiliates. All rights reserved 10 SPARC T3 Prozessor Übersicht DDR3 DDR3 DDR3 DDR3 DDR3 DDR3 DDR3 DDR3 • 16 Kerne @ 1.65 GHz 2x8 PCIe 2.0 @ 5 Gbps 6 Coherence Links 8 GB/s each direction – 128 Threads BoB BoB BoB BoB Memory Memory PEU PEU • TSMC N40G+ Technology Controller Controller DMU Coherence Unit Coherence Unit • Neue 9-stufige FPU SIU NIU 2 XAUI 6MB L2$, 16 BANKS, 24 WAY – Fused Multiply/Add N2 SPARC 10GE CORE CROSSBAR 16KB I$ NCU 8KB D$ RAM C C C C C C C C C C C C C C C C • 1 1 1 1 1 1 1 FGU 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 CTU Crypto – 4x FBDIMM2 @ 6.4Gb/s 8 threads per core @1.65 GHz • I/O und Netzwerk – 2x PCIe2 x8 links @ 5GTS – 2x 10GbE (XAUI interface) • Power: 146 – 169 Watt © 2010 Oracle Corporation and/or its affiliates. All rights reserved 11 SPARC T3 – Neues Power Management • CPU Takt Anpassung • Taktrate entsprechend der Auslastung anpassen • Memory Power Management • Idle-Modus für nicht/wenig genutztes RAM • Power Limit • Power-Obergrenze für das Gesamtsystem • Wird die Grenze erreicht, werden einzelne Komponenten im Verbrauch reduziert © 2010 Oracle Corporation and/or its affiliates. All rights reserved 12 SPARC T3 Server: Jede Menge Skalierbarkeit Hoch integriert, höchster Durchsatz SPARC T3-4 H I G H S Y SPARC T3-1B Blade S for Blade 6000 SPARC T3-2 • 64 cores T E • 512 threads M SPARC T3-1 • 5 RU T • 16x PCIe EM H • 32 cores R • Best Scale O • 256 threads • Most Crypto U • 16 cores • 3 RU G • 128 threads H • 10x PCIe • 2 RU P • Medium scale U • 6x PCIe • T • Middleware 16 cores • Entry-level Consolidation • 128 threads • Price/performance • Best density • General purpose CONSOLIDATION VIRTUALIZATION HIGH © 2010 Oracle Corporation and/or its affiliates. All rights reserved 13 Herausragende Leistung Oracle Business Applications • SPECjEnterprise2010 (Weblogic Server) • Oracle JD Edwards EnterpriseOne 9.0.1 • Siebel CRM 8.1.1 • E-Business Suite R12.12 (HR SelfService) • Consumer E-Commerce • Communications Order & Service Management • PeopleSoft Enterprise Financials 9.0 • SPECjvm2008 (HotSpot Java VM) • X86 to SPARC Virtualization http://www.oracle.com/us/solutions/performance-scalability/sun-sparc-enterprise-t-servers-078532.html © 2010 Oracle Corporation and/or its affiliates. All rights reserved 14 Performance – Ein Beispiel SPARC T3-4 & Oracle 11g schlägt IBM Power7 • SPECjEnterprise Benchmark • SPARC T3-4 Application & SPARC T3-2 Database • T3-4 erreicht 9456.28 SPECjEnterprise2010 EjOPS@Standard • 32% schneller als IBM p750 (ebenfalls 4 CPU) • 5% bessere Leistungsdichte als Power7 SPARC T3-4 IBM Power7 0 5.000 10.000 Application Tier DB Tier JOPS@standard Database SPARC T3-4 (64 core) SPARC T3-2 9,456.28 Oracle11g IBM p750 (32 core) IBM PS702 7,172.93 DB2 9.7 © 2010 Oracle Corporation and/or its affiliates. All rights reserved 15 16 Kerne mit Zukunft 5 Jahres-Ausblick Cores 4x Threads 32x SPARC Memory Kapazität 16x 1-64 Sockets Datenbank TPM 40x +2x Throughput Java Ops Pro Sekunde 10x +1.5x Single Strand M-series 8-64 Sockets +2x Throughput Heute T-series 1-8 Sockets M-series +3x Throughput 8-64 Sockets +6x Throughput T-series +1.5x Single Strand 1-4 Sockets M-series +3x Single Strand 1-64 Socket T-series + 20% 1-4 Socket + 2x Throughput Solaris 10 Solaris 11 Solaris 11 Solaris 11 Solaris 11 Update Update Update Update 2010 2011 2012 2013 2014 2015 © 2010 Oracle Corporation and/or its affiliates. All rights reserved 16 Agenda • SPARC T3 – Acht sind nicht genug <Insert Picture Here> • x86 – Wenn acht nicht reichen, nehmen Sie zwei! • Virtualisierung – Viele Flöhe, kein Aufpreis • Management – Zum Flöhe hüten bestens geeignet © 2010 Oracle Corporation and/or its affiliates. All rights reserved 17 Sun x86 Rack-Mount Server Family • Vollständige Serverfamilie, aktualisiert mit Intel Xeon Prozessoren 5600 & 7500 • Ausgestattet mit Flash für optimale Leistung • Sun Fire X4800 Server Breiteste Auswahl an Betriebsystemen und Virtualisierungs-Optionen Sun Fire X4470 Server Sun Fire X4270 M2 Server Sun Fire X4170 M2 Server Sun Fire X2270 M2 Server © 2010 Oracle Corporation and/or its affiliates. All rights reserved 18 Sun Fire X4800 server Compute Up to 8x Intel Xeon Processor 7500 Series Up to 128x DDR3-1066 RDIMMs Up to 1 TB of memory I/O 8x PCIe 2.0 EM slots (8x lane per EM) 8x 10GbE SFP+ 8x GbE ports 8x 2.5” SAS-2/SATA/SSD drives Availability Hot-plug disks HW RAID 0,1,5,6,10,50,60 (w/REM) Hot-swap redundant 2000W PSUs 8-Socket Glueless 5 Rack Unit (RU) Hot-swap redundant fans Enterprise-Class SMP Hot-swap I/O and SP Rack System Management Oracle ILOM Service Processor Solaris, Linux, Windows or VMware © 2010 Oracle Corporation and/or its affiliates. All rights reserved 19 Exadata Database Machine X2-8 Full Rack Sun Fire X4800 als Hochleistungs-Datenbank Server High Performance Database Nodes Xeon 7560 @ 2.26 GHz 2x 8 CPUs 1066 MHz DDR3 RAM 2x 1TB 10 Gbit Ethernet to DC 2x 8 Ports • Für höchste Anforderungen – Insb. OLTP mit hoher Schreibaktivität • Skalierbarkeit – im Datenbank-Knoten: 64 Cores, 1TB – mit RAC: 2 Knoten pro Rack – Bei Bedarf mehrere Racks... © 2010 Oracle Corporation and/or its affiliates. All rights reserved 20 Exalogic Elastic Cloud Compute Nodes SunFire X4170 M2 – HPC-Performance für Java • Die schnellste Server Plattform der Welt für Anwendungen, liefert unübertroffene Java Performance • Optimale Integration mit der Exadata Database Maschine © 2010 Oracle Corporation and/or its affiliates. All rights reserved 21 ZFS Storage Appliances Second Generation Systems Standard Features (All Models) All data protocols included Advanced data services included 7210 7320 Entry-Level Clients and Applications STANDALONE Entry 7420 (All Models) CLUSTER- CAPABLE Oracle Solaris • Oracle Enterprise Linux Expandable Oracle Database, Middleware, and CLUSTER- Applications Oracle VM • VMware • Windows CAPABLE More than 50 business applications supported 7720 NEW BENEFITS Density and Capacity: Up to 1PB For Consolidation • Flash High-Density Everywhere and More Of It: Up to 4T Read Cache and 432G Write CLUSTER- Cache • More Processing Power: Up to 32 Cores Per Controller and CAPABLE 50% Better Performance © 2010 Oracle Corporation and/or its affiliates. All rights reserved 22 Agenda • SPARC T3 – Acht sind nicht genug <Insert Picture Here> • x86 – Wenn acht nicht reichen, nehmen Sie zwei! • Virtualisierung – Viele Flöhe, kein Aufpreis • Management – Zum Flöhe hüten bestens geeignet © 2010 Oracle Corporation and/or its affiliates. All rights reserved 23 Oracle VM Server for SPARC 2.0 • Neu mit SPARC T3 Servern – Auch für T2/T2Plus verfügbar • Verbessert: Alles – Performance – Resource Management – Power Management – Flexibilität – Verfügbarkeit SPARC Hypervisor T-Series Server © 2010 Oracle Corporation and/or its affiliates.
Recommended publications
  • Datasheet Fujitsu Sparc Enterprise T5440 Server
    DATASHEET FUJITSU SPARC ENTERPRISE T5440 SERVER DATASHEET FUJITSU SPARC ENTERPRISE T5440 SERVER THE SYSTEM THAT MOVES WEB APPLICATION CONSOLIDATION INTO MID-RANGE COMPUTING. UP TO 4 HIGH PERFORMANCE PROCESSORS, HIGH MEMORY AND EXTENSIVE CONNECTIVITY PROVIDE THE INFRASTRUCTURE FOR BACK OFFICE AND DATA CENTER CONSOLIDATION TASKS. FUJITSU SPARC ENTERPRISE FOR WEB SECURITY, SPARC ENVIRONMENTS MEAN MANAGEABILITY AND EFFICIENCY AND PERFORMANCE RELIABILITY Fujitsu SPARC Enterprise throughput computing Based on a four socket design, Fujitsu SPARC servers are the ultimate in Web and front-end Enterprise T5440 provides up to 256 threads and business processes. Designed for space efficiency, 512GB of memory for outstanding workload low power consumption, and maximum compute consolidation. These servers can deliver outstanding performance they provide high throughput, data throughput performance in web and network energy-saving, and space-saving solutions, in Web environments while also delivering excellent server server deployment. Built on UltraSPARC T2 or consolidation capability for back office and UltraSPARC T2 Plus processors, everything is departmental database solutions. Fully supported by integrated together on each processor chip to reduce solid management and the top scalability and the overall component count. This speeds openness of the Solaris Operating system, you have performance lowers power use and reduces the ability to maximise thread utilization, deliver component failure. Add in the no-cost virtualization application capability, and scale as large as you technology from Logical Domains and Solaris need. Containers and you have a fully scalable environment for server consolidation. Finish it off with on-chip The intrinsic service management in Fujitsu SPARC encryption and 10 Giga-bit Ethernet freeways and Enterprise T5440 combined with the SPARC they provide the compete environment for secure hardware architecture and Solaris operating system data processing and lightening fast throughput.
    [Show full text]
  • Oracle® Developer Studio 12.6
    ® Oracle Developer Studio 12.6: C++ User's Guide Part No: E77789 July 2017 Oracle Developer Studio 12.6: C++ User's Guide Part No: E77789 Copyright © 2017, Oracle and/or its affiliates. All rights reserved. This software and related documentation are provided under a license agreement containing restrictions on use and disclosure and are protected by intellectual property laws. Except as expressly permitted in your license agreement or allowed by law, you may not use, copy, reproduce, translate, broadcast, modify, license, transmit, distribute, exhibit, perform, publish, or display any part, in any form, or by any means. Reverse engineering, disassembly, or decompilation of this software, unless required by law for interoperability, is prohibited. The information contained herein is subject to change without notice and is not warranted to be error-free. If you find any errors, please report them to us in writing. If this is software or related documentation that is delivered to the U.S. Government or anyone licensing it on behalf of the U.S. Government, then the following notice is applicable: U.S. GOVERNMENT END USERS: Oracle programs, including any operating system, integrated software, any programs installed on the hardware, and/or documentation, delivered to U.S. Government end users are "commercial computer software" pursuant to the applicable Federal Acquisition Regulation and agency-specific supplemental regulations. As such, use, duplication, disclosure, modification, and adaptation of the programs, including any operating system, integrated software, any programs installed on the hardware, and/or documentation, shall be subject to license terms and license restrictions applicable to the programs.
    [Show full text]
  • RISC-V Bitmanip Extension Document Version 0.90
    RISC-V Bitmanip Extension Document Version 0.90 Editor: Clifford Wolf Symbiotic GmbH [email protected] June 10, 2019 Contributors to all versions of the spec in alphabetical order (please contact editors to suggest corrections): Jacob Bachmeyer, Allen Baum, Alex Bradbury, Steven Braeger, Rogier Brussee, Michael Clark, Ken Dockser, Paul Donahue, Dennis Ferguson, Fabian Giesen, John Hauser, Robert Henry, Bruce Hoult, Po-wei Huang, Rex McCrary, Lee Moore, Jiˇr´ıMoravec, Samuel Neves, Markus Oberhumer, Nils Pipenbrinck, Xue Saw, Tommy Thorn, Andrew Waterman, Thomas Wicki, and Clifford Wolf. This document is released under a Creative Commons Attribution 4.0 International License. Contents 1 Introduction 1 1.1 ISA Extension Proposal Design Criteria . .1 1.2 B Extension Adoption Strategy . .2 1.3 Next steps . .2 2 RISC-V Bitmanip Extension 3 2.1 Basic bit manipulation instructions . .4 2.1.1 Count Leading/Trailing Zeros (clz, ctz)....................4 2.1.2 Count Bits Set (pcnt)...............................5 2.1.3 Logic-with-negate (andn, orn, xnor).......................5 2.1.4 Pack two XLEN/2 words in one register (pack).................6 2.1.5 Min/max instructions (min, max, minu, maxu)................7 2.1.6 Single-bit instructions (sbset, sbclr, sbinv, sbext)............8 2.1.7 Shift Ones (Left/Right) (slo, sloi, sro, sroi)...............9 2.2 Bit permutation instructions . 10 2.2.1 Rotate (Left/Right) (rol, ror, rori)..................... 10 2.2.2 Generalized Reverse (grev, grevi)....................... 11 2.2.3 Generalized Shuffleshfl ( , unshfl, shfli, unshfli).............. 14 2.3 Bit Extract/Deposit (bext, bdep)............................ 22 2.4 Carry-less multiply (clmul, clmulh, clmulr)....................
    [Show full text]
  • Debugging Multicore & Shared- Memory Embedded Systems
    Debugging Multicore & Shared- Memory Embedded Systems Classes 249 & 269 2007 edition Jakob Engblom, PhD Virtutech [email protected] 1 Scope & Context of This Talk z Multiprocessor revolution z Programming multicore z (In)determinism z Error sources z Debugging techniques 2 Scope and Context of This Talk z Some material specific to shared-memory symmetric multiprocessors and multicore designs – There are lots of problems particular to this z But most concepts are general to almost any parallel application – The problem is really with parallelism and concurrency rather than a particular design choice 3 Introduction & Background Multiprocessing: what, why, and when? 4 The Multicore Revolution is Here! z The imminent event of parallel computers with many processors taking over from single processors has been declared before... z This time it is for real. Why? z More instruction-level parallelism hard to find – Very complex designs needed for small gain – Thread-level parallelism appears live and well z Clock frequency scaling is slowing drastically – Too much power and heat when pushing envelope z Cannot communicate across chip fast enough – Better to design small local units with short paths z Effective use of billions of transistors – Easier to reuse a basic unit many times z Potential for very easy scaling – Just keep adding processors/cores for higher (peak) performance 5 Parallel Processing z John Hennessy, interviewed in the ACM Queue sees the following eras of computer architecture evolution: 1. Initial efforts and early designs. 1940. ENIAC, Zuse, Manchester, etc. 2. Instruction-Set Architecture. Mid-1960s. Starting with the IBM System/360 with multiple machines with the same compatible instruction set 3.
    [Show full text]
  • Sun SPARC Enterprise T5440 Servers
    Sun SPARC Enterprise® T5440 Server Just the Facts SunWIN token 526118 December 16, 2009 Version 2.3 Distribution restricted to Sun Internal and Authorized Partners Only. Not for distribution otherwise, in whole or in part T5440 Server Just the Facts Dec. 16, 2009 Sun Internal and Authorized Partner Use Only Page 1 of 133 Copyrights ©2008, 2009 Sun Microsystems, Inc. All Rights Reserved. Sun, Sun Microsystems, the Sun logo, Sun Fire, Sun SPARC Enterprise, Solaris, Java, J2EE, Sun Java, SunSpectrum, iForce, VIS, SunVTS, Sun N1, CoolThreads, Sun StorEdge, Sun Enterprise, Netra, SunSpectrum Platinum, SunSpectrum Gold, SunSpectrum Silver, and SunSpectrum Bronze are trademarks or registered trademarks of Sun Microsystems, Inc. in the United States and other countries. All SPARC trademarks are used under license and are trademarks or registered trademarks of SPARC International, Inc. in the United States and other countries. Products bearing SPARC trademarks are based upon an architecture developed by Sun Microsystems, Inc. UNIX is a registered trademark in the United States and other countries, exclusively licensed through X/Open Company, Ltd. T5440 Server Just the Facts Dec. 16, 2009 Sun Internal and Authorized Partner Use Only Page 2 of 133 Revision History Version Date Comments 1.0 Oct. 13, 2008 - Initial version 1.1 Oct. 16, 2008 - Enhanced I/O Expansion Module section - Notes on release tabs of XSR-1242/XSR-1242E rack - Updated IBM 560 and HP DL580 G5 competitive information - Updates to external storage products 1.2 Nov. 18, 2008 - Number
    [Show full text]
  • Day 2, 1640: Leveraging Opensparc
    Leveraging OpenSPARC ESA Round Table 2006 on Next Generation Microprocessors for Space Applications G.Furano, L.Messina – TEC-EDD OpenSPARC T1 • The T1 is a new-from-the-ground-up SPARC microprocessor implementation that conforms to the UltraSPARC architecture 2005 specification and executes the full SPARC V9 instruction set. Sun has produced two previous multicore processors: UltraSPARC IV and UltraSPARC IV+, but UltraSPARC T1 is its first microprocessor that is both multicore and multithreaded. • The processor is available with 4, 6 or 8 CPU cores, each core able to handle four threads. Thus the processor is capable of processing up to 32 threads concurrently. • Designed to lower the energy consumption of server computers, the 8-cores CPU uses typically 72 W of power at 1.2 GHz. G.Furano, L.Messina – TEC-EDD 72W … 1.2 GHz … 90nm … • Is a cutting edge design, targeted for high-end servers. • NOT FOR SPACE USE • But, let’s see which are the potential spin-in … G.Furano, L.Messina – TEC-EDD Why OPEN ? On March 21, 2006, Sun made the UltraSPARC T1 processor design available under the GNU General Public License. The published information includes: • Verilog source code of the UltraSPARC T1 design, including verification suite and simulation models • ISA specification (UltraSPARC Architecture 2005) • The Solaris 10 OS simulation images • Diagnostics tests for OpenSPARC T1 • Scripts, open source and Sun internal tools needed to simulate the design and to do synthesis of the design • Scripts and documentation to help with FPGA implementation
    [Show full text]
  • SPARC Enterprise Oracle VM Server for SPARC Important Information
    SPARC Enterprise Oracle VM Server for SPARC Important Information C120-E618-06EN October 2012 Copyright © 2007, 2012, Oracle and/or its affiliates and FUJITSU LIMITED. All rights reserved. Oracle and/or its affiliates and Fujitsu Limited each own or control intellectual property rights relating to products and technology described in this document, and such products, technology and this document are protected by copyright laws, patents, and other intellectual property laws and international treaties. This document and the product and technology to which it pertains are distributed under licenses restricting their use, copying, distribution, and decompilation. No part of such product or technology, or of this document, may be reproduced in any form by any means without prior written authorization of Oracle and/or its affiliates and Fujitsu Limited, and their applicable licensors, if any. The furnishings of this document to you does not give you any rights or licenses, express or implied, with respect to the product or technology to which it pertains, and this document does not contain or represent any commitment of any kind on the part of Oracle or Fujitsu Limited, or any affiliate of either of them. This document and the product and technology described in this document may incorporate third-party intellectual property copyrighted by and/or licensed from the suppliers to Oracle and/or its affiliates and Fujitsu Limited, including software and font technology. Per the terms of the GPL or LGPL, a copy of the source code governed by the GPL or LGPL, as applicable, is available upon request by the End User.
    [Show full text]
  • Performance Analysis of Multiple Threads/Cores Using the Ultrasparc T1
    Performance Analysis of Multiple Threads/Cores Using the UltraSPARC T1 Dimitris Kaseridis and Lizy K. John Department of Electrical and Computer Engineering The University of Texas at Austin {kaseridi, ljohn}@ece.utexas.edu Abstract- By including multiple cores on a single chip, Chip to the Server-on-Chip execution model. Under such an envi- Multiprocessors (CMP) are emerging as promising ways of utiliz- ronment, the diverged execution threads will place dissimilar ing the additional die area that is available due to process scaling demands on the shared resources of the system and therefore, at smaller semiconductor feature-size technologies. However, due to resource contention, compete against each other. Con- such an execution environment with multiple hardware context sequently, such competition could result in severe destructive threads on each individual core, that is able to execute multiple threads of the same or different workloads, significantly diverges interference between the concurrently executing threads. Such from the typical, well studied, uniprocessor model and introduces behavior is non-deterministic since the execution of each a high level of non-determinism. There are not enough studies to thread significantly depends on the behavior of the rest of the analyze the performance impact of the contention of shared re- simultaneously executing applications, especially for the case sources of a processor due to multiple executing threads. We of CMP where multiple processes run on each individual core. demonstrate the existence destructive interference on Chip Mul- So far, many researchers have recognized the need of tiprocessing (CMP) architectures using both a multiprogrammed Quality of Service (QoS) that both the software [6] and hard- and a multithreaded workload, on a real, Chip Multi-Threaded ware stack [7-10] has to provide to each individual thread in (CMT) system, the UltraSPARC T1 (Niagara).
    [Show full text]
  • Sparc Enterprise T5440 Server Architecture
    SPARC ENTERPRISE T5440 SERVER ARCHITECTURE Unleashing UltraSPARC T2 Plus Processors with Innovative Multi-core Multi-thread Technology White Paper July 2009 TABLE OF CONTENTS THE ULTRASPARC T2 PLUS PROCESSOR 0 THE WORLD'S FIRST MASSIVELY THREADED SYSTEM ON A CHIP (SOC) 0 TAKING CHIP MULTITHREADED DESIGN TO THE NEXT LEVEL 1 ULTRASPARC T2 PLUS PROCESSOR ARCHITECTURE 3 SERVER ARCHITECTURE 8 SYSTEM-LEVEL ARCHITECTURE 8 CHASSIS DESIGN INNOVATIONS 13 ENTERPRISE-CLASS MANAGEMENT AND SOFTWARE 19 SYSTEM MANAGEMENT TECHNOLOGY 19 SCALABILITY AND SUPPORT FOR INNOVATIVE MULTITHREADING TECHNOLOGY21 CONCLUSION 28 0 The UltraSPARC T2 Plus Processors Chapter 1 The UltraSPARC T2 Plus Processors The UltraSPARC T2 and UltraSPARC T2 Plus processors are the industry’s first system on a chip (SoC), supplying the most cores and threads of any general-purpose processor available, and integrating all key system functions. The World's First Massively Threaded System on a Chip (SoC) The UltraSPARC T2 Plus processor eliminates the need for expensive custom hardware and software development by integrating computing, security, and I/O on to a single chip. Binary compatible with earlier UltraSPARC processors, no other processor delivers so much performance in so little space and with such small power requirements letting organizations rapidly scale the delivery of new network services with maximum efficiency and predictability. The UltraSPARC T2 Plus processor is shown in Figure 1. Figure 1. The UltraSPARC T2 Plus processor with CoolThreads technology 1 The UltraSPARC
    [Show full text]
  • Oracle's SPARC T5-2, SPARC T5-4, SPARC T5-8, and SPARC T5-1B Server Architecture Oracle's SPARC T5-2, SPARC T5-4, SPARC T5-8, and SPARC T5-1B Server Architecture
    An Oracle White Paper February 2014 Oracle's SPARC T5-2, SPARC T5-4, SPARC T5-8, and SPARC T5-1B Server Architecture Oracle's SPARC T5-2, SPARC T5-4, SPARC T5-8, and SPARC T5-1B Server Architecture Introduction ....................................................................................... 1 Comparison of SPARC T5–Based Server Features........................... 2 SPARC T5 Processor ........................................................................ 3 Taking Oracle’s Multicore/Multithreaded Design to the Next Level 5 SPARC T5 Processor Architecture ................................................ 6 SPARC T5 Processor Cache Architecture ..................................... 8 SPARC T5 Core Architecture ........................................................ 9 Oracle Solaris for Multicore Scalability............................................. 16 Oracle Solaris 11 Operating System ................................................ 18 Oracle Solaris Predictive Self Healing, Fault Management Architecture, and Service Management Facility ....................................................... 19 Oracle Solaris Cryptographic Frameworks................................... 19 End-to-End Virtualization Technology .............................................. 19 A Multithreaded Hypervisor ......................................................... 20 Oracle VM Server for SPARC ...................................................... 20 Oracle Solaris Zones ................................................................... 21 Enterprise-Class
    [Show full text]
  • Opensparc – an Open Platform for Hardware Reliability Experimentation
    OpenSPARC – An Open Platform for Hardware Reliability Experimentation Ishwar Parulkar and Alan Wood Sun Microsystems, Inc. James C. Hoe and Babak Falsafi Carnegie Mellon University Sarita V. Adve and Josep Torrellas University of Illinois at Urbana- Champaign Subhasish Mitra Stanford University IEEE SELSE 4 - March 26, 2008 www.OpenSPARC.net Outline 1.Chip Multi-threading (CMT) 2.OpenSPARC T2 and T1 processors 3.Reliability in OpenSPARC processors 4.What is available in OpenSPARC 5.Current university research using OpenSPARC 6.Future research directions IEEE SELSE 4 – March 26, 2008 2 www.OpenSPARC.net World's First 64-bit Open Source Microprocessor OpenSPARC.net Governed by GPLv2 Complete processor architecture & implementation Register Transfer Level (RTL) Hypervisor API Verification suite and architectural models Simulation model for operating system bringup on s/w IEEE SELSE 4 – March 26, 2008 3 www.OpenSPARC.net Chip Multithreading (CMT) Instruction- Low Low Low Medium Low High level Parallelism Thread-level Parallelism High High High High High Instruction/Data Large Large Medium Large Large Working Set Data Sharing Low Medium High Medium High Medium IEEE SELSE 4 – March 26, 2008 4 www.OpenSPARC.net Memory Bottleneck Relative Performance 10000 CPU Frequency DRAM Speeds 1000 2 Years 100 Every Gap 2x -- CPU 6 10 -- 2x Every DRAM Years 1 1980 1985 1990 1995 2000 2005 Source: Sun World Wide Analyst Conference Feb. 25, 2003 IEEE SELSE 4 – March 26, 2008 5 www.OpenSPARC.net Single Threading HURRY Up to 85% Cycles Waiting for Memory
    [Show full text]
  • Ultrasparc T1 Sparc History Sun + Sparc = Ultrasparc
    ULTRASPARC T1 SUN + SPARC = ULTRASPARC THE PROCESSOR FORMERLY KNOWN AS “NIAGARA” Processor Cores Threads/Core Clock L1D L1I L2 Cache UltraSPARC IIi 1 1 550Mhz, 650Mhz 16KiB 16KiB 512KiB UltraSPARC IIIi 1 1 1.593Ghz I D 1MBa UltraSPARC III 1 1 1.05-1.2GHz 64KiB 32KiB 8MiBb UltraSPARC IV 2c 1 1.05-1.35Ghz 64KiB 32KiB 16MiBd UltraSPARC IV+ 1 2 1.5Ghz I D 2MiBe UltraSPARC T1 8 4 1.2Ghz 32KiB 16KiBf 3MiBg UltraSPARC T2h 16 (?) 8 2Ghz+ (?) ? ? ? Slide 1 Slide 3 aOn-chip bExternal, on chip tags cUltraSPARC III cores d8MiB per core e32MiB off chip L3 fI/D Cache per core g4 way banked hSecond-half 2007 This work supported by UNSW and HP through the Gelato Federation SPARC HISTORY INSTRUCTION SET ➜ Scalable Processor ARCHitecture ➜ RISC! ➜ 1985 – Sun Microsystems ➜ Berkeley RISC – 1980-1984 ➜ Load–store only through registers ➜ MIPS – 1981-1984 ➜ Fixed size instructions (32 bits) ➜ register + register Slide 2 Architecture v Implementation: Slide 4 ➜ register + 13 bit immediate ➜ SPARC Architecture ➜ Branch delay slot ➜ SPARC V7 – 1986 X Condition Codes ➜ SPARC Interntaional, Ltd – 1989 V (V9) CC and non-CC instructions ➜ SPARC V8 – 1990 V (V9) Compare on integer registers ➜ SPARC V9 – 1994 ➜ Synthesised instructions ➜ Privileged v Non-Privileged SUN + SPARC = ULTRASPARC 1 CODE EXAMPLE 2 CODE EXAMPLE V9 REGISTER WINDOWS void addr(void) { int i = 0xdeadbeef; } 00000054 <addr>: Slide 5 54: 9d e3 bf 90 save %sp, -112, %sp Slide 7 58: 03 37 ab 6f sethi %hi(0xdeadbc00), %g1 5c: 82 10 62 ef or %g1, 0x2ef, %g1 60: c2 27 bf f4 st %g1, [ %fp + -12 ] 64:
    [Show full text]