DOCSLIB.ORG

Performance Prediction Based on Inherent Program Similarity

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Performance Prediction Based on Inherent Program Similarity Performance Prediction based on Inherent Program Similarity Kenneth Hostey, Aashish Phansalkarz, Lieven Eeckhouty, Andy Georgesy, Lizy K. Johnz and Koen De Bosscherey yELIS, Ghent University, Belgium zECE, The University of Texas at Austin fkehoste,leeckhou,ageorges,[email protected] faashish,[email protected] ABSTRACT 1. INTRODUCTION A key challenge in benchmarking is to predict the performance of From a benchmark consumer point-of-view, a key challenge is to an application of interest on a number of platforms in order to de- determine the platform that yields the best performance for a given termine which platform yields the best performance. This paper application of interest. Ideally, the user's application of interest proposes an approach for doing this. We measure a number of is his best benchmark. However, in many practical circumstances microarchitecture-independent characteristics from the application the user has to rely on the performance scores of a standardized of interest, and relate these characteristics to the characteristics of benchmark suite for estimating the performance of the application the programs from a previously profiled benchmark suite. Based of interest for two reasons. First, it is too difficult or costly to port on the similarity of the application of interest with programs in the the application program of interest to a wide range of platforms. benchmark suite, we make a performance prediction of the applica- Second, there are many platforms for which the performance needs tion of interest. We propose and evaluate three approaches (normal- to be measured before making a choice about which platform yields ization, principal components analysis and genetic algorithm) to to the best performance for the given application. transform the raw data set of microarchitecture-independent char- A popular tool for estimating performance of an application pro- acteristics into a benchmark space in which the relative distance gram on an unavailable platform is detailed cycle-accurate proces- is a measure for the relative performance differences. We evalu- sor simulation. However, next to not solving the porting problem, ate our approach using all of the SPEC CPU2000 benchmarks and simulation is very time consuming and thus is difficult to use in real hardware performance numbers from the SPEC website. Our practice. framework estimates per-benchmark machine ranks with a 0.89 av- This motivates us to come up with a different solution to this erage and a 0.80 worst case rank correlation coefficient. ubiquitous problem in benchmarking. The methodology proposed in this paper uses the already known performance scores of stan- dardized benchmark suites on the systems of our interest. As a Categories and Subject Descriptors part of our methodology we measure a set of microarchitecture- C.4 [Performance of Systems]: Modeling Techniques, Performance independent characteristics for the new application of interest and attributes relate them to the same characteristics of the benchmarks in the standardized benchmark suite. The microarchitecture-independent characteristics capture the inherent program behavior that is un- General Terms biased towards a particular microarchitecture. We then use the Experimentation, Measurement, Performance knowledge of similarity between the application of interest and the corresponding benchmarks to predict the performance of the appli- cation of interest. In other words, we use the standardized bench- Keywords marks as proxies for our application of interest based on similarity. Performance Modeling, Workload Characterization, Inherent Pro- The key issue in a methodology that uses program similarity gram Behavior based on microarchitecture-independent program characteristics is to determine how differences in microarchitecture-independent char- acteristics translate into differences in performance. We propose and evaluate three approaches for achieving that, namely normal- ization, principal components analysis and a genetic algorithm, of which the genetic algorithm shows to be the most accurate. A ge- netic algorithm learns how to rescale the benchmark space so that Permission to make digital or hard copies of all or part of this work for the Euclidean distance in the benchmark space becomes a more personal or classroom use is granted without fee provided that copies are accurate measure for performance differences when running the not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to benchmarks on a variety of platforms. republish, to post on servers or to redistribute to lists, requires prior specific We evaluate our framework for predicting machine ranks using permission and/or a fee. SPEC published speedup rates that cover various commercial ma- PACT'06, September 16–20, 2006 , Seattle, Washington, USA. chines with different ISAs, compiler settings and microprocessors. Copyright 2006 ACM 1­59593­264­X/06/0009 ...$5.00. benchmark space uarch-indep data chars transform data uarch-indep transform chars build data application performance transform of interest numbers benchmark suite predict performance based on programs in neighborhood Figure 1: The framework proposed in this paper for predicting performance based on microarchitecture-independent program characteristics. Current practice, which uses the average rank across all bench- the microarchitecture-independent characteristics using the same marks for predicting ranks for specific applications of interest, a- data transformation matrix as above. This locates the application chieves an average 0.83 and a worst case 0.64 rank correlation co- of interest in the benchmark space. Performance is then predicted efficient for the estimated speedups versus the measured speedups. by appropriately weighting the performance numbers of the bench- Our framework based on inherent program similarity achieves an marks in the neighborhood of the application of interest. average correlation coefficient of 0.89; the worst case correlation We now discuss a number of aspects of this framework: (i) the coefficient that we observe is 0.79. These results demonstrate that microarchitecture-independent characteristics, (ii) how to build the our framework is indeed capable of tracking performance differ- data transformation matrix, and (iii) how to compute a performance ences across platforms with different ISAs, compilers and micro- number for the application of interest. processors. To the best of our knowledge, this paper is the first to propose a methodology for predicting machine ranks for indi- 2.1 Microarchitecture­independent character­ vidual programs based on microarchitecture-independent program istics similarity. Ideally, the program characteristics that serve as input to our This paper is organized as follows. We first detail on our per- framework should be platform-independent characteristics. In other formance prediction framework. We then present our experimental words, they should be compiler-independent, ISA-independent and setup followed by the evaluation of our framework. Finally, we microarchitecture-independent in order to capture the true inherent discuss related work and conclude. program behavior. Since this is difficult to do, we take a prag- matic approach and use microarchitecture-independent character- istics. The characteristics that we collect are specific to a given 2. PERFORMANCE PREDICTION FRAME­ ISA and a given compiler, however they are independent of a given WORK microarchitecture, i.e., the characteristics are independent of cache Figure 1 illustrates the framework that we propose in this paper size, branch predictor size, processor core configuration, etc. As for predicting performance based on microarchitecture-independent will be shown in the evaluation section of this paper, these charac- program similarity. The framework assumes a collection of pro- teristics, inspite of being ISA-dependent and compiler-dependent, grams which we call the benchmark suite. For each of these bench- are accurate enough for tracking performance across different plat- marks, we have a collection of microarchitecture-independent char- forms with different ISAs and compilers. acteristics as well as performance numbers on a (number of) plat- Table 1 summarizes the 47 microarchitecture-independent char- form(s). The performance numbers could be obtained from simu- acteristics that we use in this paper. The range of microarchitecture- lation or from real hardware execution. These microarchitecture- independent characteristics is fairly broad in order to cover all ma- independent characteristics along with the performance numbers jor program behaviors such as instruction mix, inherent ILP, work- are then used to build a data transformation matrix — building the ing set sizes, memory strides, branch predictability, etc. Measur- data transformation matrix can also be done without using perfor- ing these program characteristics can be done efficiently through mance numbers, thus using microarchitecture-independent charac- instrumentation which is substantially faster than simulation. We teristics solely (hence the dashed line between the `performance include the following characteristics: numbers' box and the `build data transform' box in Figure 1). Once Instruction mix. We include the percentage of loads, stores, the data transformation matrix is computed, the original microarchi- control transfers, arithmetic operations, integer multiplies and floa-
Load more

Recommended publications
  • Performance of a Computer (Chapter 4) Vishwani D
    ELEC 5200-001/6200-001 Computer Architecture and Design Fall 2013 Performance of a Computer (Chapter 4) Vishwani D. Agrawal & Victor P. Nelson epartment of Electrical and Computer Engineering Auburn University, Auburn, AL 36849 ELEC 5200-001/6200-001 Performance Fall 2013 . Lecture 1 What is Performance? Response time: the time between the start and completion of a task. Throughput: the total amount of work done in a given time. Some performance measures: MIPS (million instructions per second). MFLOPS (million floating point operations per second), also GFLOPS, TFLOPS (1012), etc. SPEC (System Performance Evaluation Corporation) benchmarks. LINPACK benchmarks, floating point computing, used for supercomputers. Synthetic benchmarks. ELEC 5200-001/6200-001 Performance Fall 2013 . Lecture 2 Small and Large Numbers Small Large 10-3 milli m 103 kilo k 10-6 micro μ 106 mega M 10-9 nano n 109 giga G 10-12 pico p 1012 tera T 10-15 femto f 1015 peta P 10-18 atto 1018 exa 10-21 zepto 1021 zetta 10-24 yocto 1024 yotta ELEC 5200-001/6200-001 Performance Fall 2013 . Lecture 3 Computer Memory Size Number bits bytes 210 1,024 K Kb KB 220 1,048,576 M Mb MB 230 1,073,741,824 G Gb GB 240 1,099,511,627,776 T Tb TB ELEC 5200-001/6200-001 Performance Fall 2013 . Lecture 4 Units for Measuring Performance Time in seconds (s), microseconds (μs), nanoseconds (ns), or picoseconds (ps). Clock cycle Period of the hardware clock Example: one clock cycle means 1 nanosecond for a 1GHz clock frequency (or 1GHz clock rate) CPU time = (CPU clock cycles)/(clock rate) Cycles per instruction (CPI): average number of clock cycles used to execute a computer instruction.
    [Show full text]
  • 6IF Ver. 1.X MANUAL
    5SMT / 5SMTS SiS 530 ALL-IN-ONE AT MAINBOARD OPERATION MANUAL DOC NO. UM-SMT-E4 …………………………………………………………………………………………………………… PRINTED IN TAIWAN SiS530 AT MAINBOARD TABLE OF CONTENTS CHAPTER & SECTION PAGE 1. INTRODUCTION ................................................................................... 1-1 1.1 SYSTEM OVERVIEW ...................................................................................1-1 1.2 SYSTEM BOARD LAYOUT .........................................................................1-2 2. FEATURES .............................................................................................. 2-1 2.1 2.1 MAINBOARD SPECIFICATIONS ........................................................2-1 3. HARDWARE SETUP ............................................................................. 3-1 3.1 UNPACKING...................................................................................................3-1 3.2 HARDWARE CONFIGURATION...............................................................3-2 3.2.1 CONNECTORS...................................................................................3-2 3.2.2 JUMPERS ............................................................................................3-17 3.2.3 QUICK INSTALLATION OF POPULAR CPU..............................3-23 3.2.4 INSTALLATION OF DEVICE DRIVER ........................................3-26 4. AWARD BIOS SETUP ........................................................................... 4-1 4.1 GETTING STARTED.....................................................................................4-1
    [Show full text]
  • AMD's Early Processor Lines, up to the Hammer Family (Families K8
    AMD’s early processor lines, up to the Hammer Family (Families K8 - K10.5h) Dezső Sima October 2018 (Ver. 1.1) Sima Dezső, 2018 AMD’s early processor lines, up to the Hammer Family (Families K8 - K10.5h) • 1. Introduction to AMD’s processor families • 2. AMD’s 32-bit x86 families • 3. Migration of 32-bit ISAs and microarchitectures to 64-bit • 4. Overview of AMD’s K8 – K10.5 (Hammer-based) families • 5. The K8 (Hammer) family • 6. The K10 Barcelona family • 7. The K10.5 Shanghai family • 8. The K10.5 Istambul family • 9. The K10.5-based Magny-Course/Lisbon family • 10. References 1. Introduction to AMD’s processor families 1. Introduction to AMD’s processor families (1) 1. Introduction to AMD’s processor families AMD’s early x86 processor history [1] AMD’s own processors Second sourced processors 1. Introduction to AMD’s processor families (2) Evolution of AMD’s early processors [2] 1. Introduction to AMD’s processor families (3) Historical remarks 1) Beyond x86 processors AMD also designed and marketed two embedded processor families; • the 2900 family of bipolar, 4-bit slice microprocessors (1975-?) used in a number of processors, such as particular DEC 11 family models, and • the 29000 family (29K family) of CMOS, 32-bit embedded microcontrollers (1987-95). In late 1995 AMD cancelled their 29K family development and transferred the related design team to the firm’s K5 effort, in order to focus on x86 processors [3]. 2) Initially, AMD designed the Am386/486 processors that were clones of Intel’s processors.
    [Show full text]
  • AMD Geode™ NX Processors BIOS Considerations Application Note
    AMD Geode™ NX Processors BIOS Considerations Application Note PID: 32483 Rev: A Issue Date: October 2004 © 2004 Advanced Micro Devices, Inc. All rights reserved. The contents of this document are provided in connection with Advanced Micro Devices, Inc. (“AMD”) products. AMD makes no representations or warranties with respect to the accuracy or completeness of the contents of this publication and reserves the right to make changes to specifications and product descriptions at any time without notice. No license, whether express, implied, arising by estoppel or otherwise, to any intellectual property rights is granted by this publication. Except as set forth in AMD’s Standard Terms and Conditions of Sale, AMD assumes no liability whatsoever, and disclaims any express or implied warranty, relating to its products including, but not limited to, the implied warranty of mer- chantability, fitness for a particular purpose, or infringement of any intellectual property right. AMD’s products are not designed, intended, authorized or warranted for use as components in systems intended for surgical implant into the body, or in other applications intended to support or sustain life, or in any other application in which the failure of AMD’s product could create a situation where personal injury, death, or severe property or environmental damage may occur. AMD reserves the right to discontinue or make changes to its products at any time without notice. Contacts www.amd.com Trademarks AMD, the AMD Arrow logo, AMD Athlon, AMD PowerNow!, and combinations thereof, and Geode are trademarks of Advanced Micro Devices, Inc. Microsoft and Windows are registered trademarks of Microsoft Corporation.
    [Show full text]
  • MARIE: an Introduction to a Simple Computer
    00068_CH04_Null.qxd 10/18/10 12:03 PM Page 195 “When you wish to produce a result by means of an instrument, do not allow yourself to complicate it.” —Leonardo da Vinci CHAPTER MARIE: An Introduction 4 to a Simple Computer 4.1 INTRODUCTION esigning a computer nowadays is a job for a computer engineer with plenty of Dtraining. It is impossible in an introductory textbook such as this (and in an introductory course in computer organization and architecture) to present every- thing necessary to design and build a working computer such as those we can buy today. However, in this chapter, we first look at a very simple computer called MARIE: a Machine Architecture that is Really Intuitive and Easy. We then pro- vide brief overviews of Intel and MIPs machines, two popular architectures reflecting the CISC and RISC design philosophies. The objective of this chapter is to give you an understanding of how a computer functions. We have, therefore, kept the architecture as uncomplicated as possible, following the advice in the opening quote by Leonardo da Vinci. 4.2 CPU BASICS AND ORGANIZATION From our studies in Chapter 2 (data representation) we know that a computer must manipulate binary-coded data. We also know from Chapter 3 that memory is used to store both data and program instructions (also in binary). Somehow, the program must be executed and the data must be processed correctly. The central processing unit (CPU) is responsible for fetching program instructions, decod- ing each instruction that is fetched, and performing the indicated sequence of operations on the correct data.
    [Show full text]
  • A Comparative Study of 64 Bit Microprocessors Apple, Intel and AMD
    Special Issue - 2016 International Journal of Engineering Research & Technology (IJERT) ISSN: 2278-0181 ICACT - 2016 Conference Proceedings A Comparative Study of 64 Bit Microprocessors Apple, Intel and AMD Surakshitha S Shifa A Ameen Rekha P M Information Science and Engineering Information Science and Engineering Information Science and Engineering JSSATE JSSATE JSSATE Bangalore, India Bangalore, India Bangalore, India Abstract- In this paper, we draw a comparative study of registers, even larger .The term may also refer to the size of microprocessor, using the three major processors; Apple, low-level data types, such as 64-bit floating-point numbers. Intel and AMD. Although the philosophy of their micro Most CPUs are designed so that the contents of a architecture is the same, they differ in their approaches to single integer register can store the address of any data in implementation. We discuss architecture issues, limitations, the computer's virtual considered an appropriate size to architectures and their features .We live in a world of constant competition dating to the historical words of the work with for another important reason: 4.29 billion Charles Darwin-‘Survival of the fittest ‘. As time passed by, integers are enough to assign unique references to most applications needed more processing power and this lead to entities in applications like databases. an explosive era of research on the architecture of Some supercomputer architectures of the 1970s and microprocessors. We present a technical & comparative study 1980s used registers up to 64 bits wide. In the mid-1980s, of these smart microprocessors. We study and compare two Intel i860 development began culminating in a 1989 microprocessor families that have been at the core of the release.
    [Show full text]
  • Athlon XP 2600+ Reviewed Athlon XP 2600+ Reviewed by Johan De Gelas – August 2002
    Ace’s Hardware Athlon XP 2600+ Reviewed Athlon XP 2600+ Reviewed By Johan De Gelas – August 2002 AMD Takes Up the Gauntlet with Flair The last few quarters have been pure agony for AMD. Sales and turnover figures plunged to terrible lows and AMD's results where written in bloody red ink. Thoroughbred, the 130 nanometer Athlon XP was limited to a 66 MHz increase over its older brother Palomino. The race with Intel was lost: Intel's 2.53 GHz Pentium 4 was arguably the fastest x86 processor in the world. After launching the Thoroughbred Athlon XP 2200+, many people on the Internet were forecasting doom scenarios for AMD. The disappointing initial clockspeeds were, however, no surprise for the hardware veterans among you. When Thoroughbred was launched, we stated: "Our sources indicate that AMD's yields at 0.13µ are pretty good, but the current bin splits have a lot of room for improvement. While 1.5V is good enough to get the Thoroughbred up to 1600 MHz, higher clockspeeds are only attained with some kind of overclocking. Now this is hardly of a reason for panic, as it is even sort of a "tradition" for AMD. With far fewer resources for debugging and tuning, new process technology almost always experiences some teething problems." Contrary to Intel, AMD has neither funds nor fabs for backup plans, so moves to new process technology or new processor layouts are always somewhat risky. New process technology or layout changes often initially result in yield, clock scaling, and/or binsplit problems, and the time that it takes to resolve these issues tends to be very variable.
    [Show full text]
  • EC Declaration of Conformity
    EC Declaration of Conformity We IWILL Corp. No. 10, Wu Chuan 3rd Rd., Hsin Chuang City, Taipei, Taiwan, R.O.C. Declare under sole responsibility that the XA100 Plus motherboard Meets the intent of Directive 89/336/ECC for Electromagnetic Compatibility. Compliance was demonstrated to the following specifications as listed in the official Journal of the European Communities: EN 50081-1 Emissions: EN 55022 Radiated, Class B EN 55022 Conducted, Class B EN 60555-2 Power Harmonics EN 50082-1 Immunity: IEC 801-2 Electrostatic Discharge IEC 801-3 RF Radiate IEC 801-4 Fast Transient V1.3C 1 About This Manual This manual introduces to the users how this XA100 PLUS Ultra ATA-2 ( Ultra DMA 33 ) and 100 MHz bus motherboard is manufactured. All useful information is described in later chapters. Keep this manual for future upgrades or system configuration changes. Chapter 0 Overview An overview of some special and basic features of the XA100 Plus motherboard Chapter 1 Features The detailed features and specifications of the XA100 Plus motherboard Chapter 2 Quick Installation The quick reference for experienced users Chapter 3 Hardware Installation The detailed information on jumper settings, connections, and hardware installation Chapter 4 AWARD BIOS Setup The detailed information of the system AWARD BIOS setup options Chapter 5 Software and Tools The detailed information for installing software and tools Warning Most of the features of this motherboard have passed strict testing, and are subject to change at any time without notice. IWILL assumes no responsibility or liability for any malfunction that occurs due to the future technical changes made by the respective component manufacturers.
    [Show full text]
  • AMD-K6-2® Processor Data Sheet
    Preliminary Information AMD-K6-2® Processor Data Sheet © 2000 Advanced Micro Devices, Inc. All rights reserved. The contents of this document are provided in connection with Advanced Micro Devices, Inc. ("AMD") products. AMD makes no representations or warranties with respect to the accuracy or completeness of the contents of this publication and reserves the right to make changes to specifications and product descriptions at any time without notice. No license, whether express, implied, arising by estoppel or otherwise, to any intellectual property rights is granted by this publication. Except as set forth in AMD's Standard Terms and Conditions of Sale, AMD assumes no liability whatsoever, and disclaims any express or implied warranty, relating to its products including, but not limited to, the implied warranty of merchantability, fitness for a particular purpose, or infringement of any intellectual property right. AMD's products are not designed, intended, authorized or warranted for use as components in systems intended for surgical implant into the body, or in other applications intended to support or sustain life, or in any other application in which the failure of AMD's product could create a situation where personal injury, death, or severe property or environmental damage may occur. AMD reserves the right to discontinue or make changes to its products at any time without notice. Trademarks AMD, the AMD logo, K6, 3DNow!, and combinations thereof, K86, and Super7 are trademarks, and AMD-K6 and RISC86 are registered trademarks of Advanced Micro Devices, Inc. Microsoft, Windows, and Windows NT are registered trademarks of Microsoft Corporation. NetWare is a registered trademark of Novell, Inc.
    [Show full text]
  • Athlon XP 3200+ Versus Pentium 4 C by Johan De Gelas – May 2003
    Ace’s Hardware Gaming CPU Performance: Athlon XP 3200+ Versus 3 GHz Pentium 4 C Gaming CPU Performance: Athlon XP 3200+ Versus Pentium 4 C By Johan De Gelas – May 2003 Athlon XP 3200+ Reviewed As AMD feels that the Athlon 64 still isn't ready for prime time, the Athlon "Barton" core remains their flagship desktop processor that must face the fierce competition of Santa Clara. Today, AMD claims that a 2.2 GHz Athlon with 512 KB cache and a 400 MHz FSB should carry the 3200+ performance rating. Clockspeeds higher than 2.2 GHz seem to be a hurdle too high for the current Athlon. In our humble opinion, the primary limiting factor is power dissipation. The Athlon faces two problems: it is not capable of reaching high clockspeeds without a core voltage of 1.65V and it is hard to get rid of the heat of the resulting power dissipation when you have to dissipate it through a very small die without a heat-spreader. The carefully mounted heat-spreader and the SOI circuits on the Opteron show, however, that AMD has solved this problem with its next generation processors. But for now, eight months after the introduction of the 2.25 GHz AMD Athlon (2800+), the Athlon is still running at 2.2 GHz. Intel seems to face the same problem with the Pentium 4, as the 3 GHz P4 has been the fastest processor for 7 (!) months now. Nevertheless, when we knew that the new Athlon XP 3200+ was a 2.2 GHz part, we could not help but to be a bit disapointed.
    [Show full text]
  • Bapco Sysmark 25 White Paper
    SYSmark® 25 July 13th, 2020 Revision 1.1 Revision History: 1.0 – Initial version of the SYSmark 25 White Paper 1.1 – Corrected typo about RAM channels in Section 2.8, Corrected reference link names in Section 2.1, Updated APPENDIX A with information on licensing. BAPCo® SYSmark® 25 White Paper Page 1 of 34 About BAPCo Business Applications Performance Corporation (BAPCo) is a non-profit consortium with a charter to develop and distribute a set of objective performance benchmarks for personal computers based on popular computer applications and industry standard operating systems. For more information about BAPCo or a complete list of the current membership, see our website at http://www.bapco.com/. BAPCo® SYSmark® 25 White Paper Page 2 of 34 Table of Contents 1 Introduction .......................................................................................................................................... 4 2 BAPCo Development Process ................................................................................................................ 5 2.1 Milestone Overview ...................................................................................................................... 6 2.2 Usage Model, Sub Scenario, and Scenario Selection .................................................................... 7 2.3 Scenario Grouping ......................................................................................................................... 8 2.4 Application Selection ...................................................................................................................
    [Show full text]
  • Sysmark® 2002 Performance Benchmark: Analysis Presentation Benchmarks & Athlon XP Performance
    SYSmark® 2002 Performance Benchmark: Analysis Presentation Benchmarks & Athlon XP Performance • Benchmarks are currently the only way for business and home consumers to clearly measure the real-world performance of PCs. Benchmarks must accurately and objectively represent real-world application performance relevant to the end-user. •BAPCo® is a non-profit consortium, originally founded in 1995 by Intel, IBM, Dell and NCR. SYSmark® is BAPCo’s flagship application benchmark. • The AMD Athlon™ XP processor routinely outperformed comparable Intel Pentium® 4 processors using the SYSmark 2001 benchmark. • On January 31, 2002, BAPCo released SYSmark 2002. 2 SYSmark 2001 & 2002 Similarities Internet Content SYSmark Both SYSmark 2001 & SYSmark Creation 2002 Version 2002 are application-based Adobe Photoshop v6.0.1 benchmarks that attempt to Adobe Premier Same reflect usage patterns for end Macromedia Dreamweaver Same users in the areas of Internet Macromedia Flash Same Content Creation and Office Windows Media Encoder v7.01 Productivity. Office Productivity SYSmark 2002 Version The chart on the right explains Microsoft Word v2002 that SYSmark 2002 uses the same Microsoft Excel v2002 applications as SYSmark 2001 Microsoft Power Point v2002 with updated versions where Microsoft Outlook v2002 applicable. Microsoft Access v2002 Dragon Naturally Speaking Same McAfee VirusScan Same Netscape Communicator Same WinZip Computing Same 3 SYSmark 2002 Inconsistencies AMD tested SYSmark 2001 and 2002 on similarly configured systems based on the AMD Athlon XP 2000+ and the Intel Pentium 4 2.0A processors. When compared to the SYSmark 2001 benchmark results, the SYSmark 2002 benchmark results suggested a significant performance decline in the AMD Athlon XP 2000+ processor’s performance and a significant increase in Intel Pentium 4 2.0A processor’s performance.
    [Show full text]