Understanding and Using PC Bench (TM) 8.0 Copyright (c) 1992, 1993 by Ziff-Davis Publishing Company, L.P. All rights reserved. Manual release date: November 1993 with PC Bench version 8.0 LICENSE AGREEMENT FOR ZIFF-DAVIS' PC BENCH (TM) 8.0 READ THIS AGREEMENT CAREFULLY BEFORE USING THE SOFTWARE EMBODIED IN THE PC BENCH{SYMBOL 228 \f "Symbol"} 8.0 DISKETTE (OR, IF DOWNLOADED, IN THE DOWNLOADED FILE(S)). Embodied in the PC Bench 8.0 diskette ("diskette") (or, if downloaded, in the downloaded file(s)) is the PC Bench 8.0 computer program and related documentation (the "Software"). Ziff-Davis Publishing Company, L.P., having a place of business at One Park Avenue, New York, New York 10016 ("Ziff") is the licensor under this Agreement and you are the licensee. By using the Software, in whole or in part, you agree to be bound by the terms of this Agreement. If you do not agree to the terms of this Agreement, promptly return the Software (or, if downloaded, delete the Software) to the Ziff-Davis Benchmark Operation at One Copley Parkway, Suite 510, Morrisville, North Carolina 27560. Title to the Software and all copyrights, trade secrets and other proprietary rights therein are owned by Ziff. All rights therein, except those expressly granted to you in this Agreement, are reserved by Ziff. 1. Limited License This Agreement grants you only limited rights to use the Software. Ziff grants you a non-exclusive, non-transferable license to use the Software on a single dedicated computer or on a file server networked with multiple PC computers for the sole purpose of conducting benchmark tests to measure the performance of computer hardware and operating system configurations. You have the right to make a single copy of the Software for archival purposes and the right to transfer a copy of the Software across a network only to the PC computers attached to the network. Ziff hereby grants you the right to publish benchmark test results obtained by you from your use of the Software provided that with the publication of each such result you: A. Identify Ziff-Davis, the name and version number of the benchmark Software used (i.e., Ziff-Davis' PC Bench (TM) Ver. 8.0); B. Identify the exact name, processor speed and type, amount of RAM, amount of secondary RAM cache, if any, hard disk size, type of hard disk controller, and size of hardware hard disk cache, if any, of the PC used for the test (e.g., WXY Corp. Model 466 with 66-MHz Intel (R) 486DX2-66 CPU, 8 MB of RAM, 64 KB RAM cache, 200 MB hard disk, IDE controller, and no hardware disk cache); C. Identify the exact video adapter name and refresh rate that produced the result (e.g., XYZ Corp. XYZ Video adapter with a refresh rate of 72 Hz); D. Identify the operating system version (e.g., MS-DOS (R) 5.0), size and type of software disk cache, if any (e.g., SMARTDRV 2MB cache), video resolution and color depth (e.g., 800 by 600 pixels with 256 colors), and any other special conditions used to achieve the result (e.g., disk compression utility ABC version 1.0 enabled); E. State that all products used in the test were shipping versions available to the general public; F. State that the test was performed without independent verification by Ziff; and G. Follow proper trademark usage and acknowledge Ziff's trademark rights (e.g., "[ ] achieved a score of X DOSMark (TM) units on the PC Bench (TM) 8.0 test. PC Bench and DOSMark are trademarks of Ziff-Davis Publishing Company, L.P."). This Agreement and your rights hereunder shall automatically terminate if you fail to comply with any provision of this Agreement. Upon such termination, you agree to cease all use of the Software, to delete the Software, and to destroy all copies of the diskette and other materials contained in this package in your possession or under your control, or, if downloaded, to destroy any and all copies of the Software in your possession or under your control. 2. Additional Restrictions A. You shall not (and shall not permit other persons or entities to) rent, lease, sell, sublicense, assign, or otherwise transfer the Software or this Agreement. Any attempt to do so shall be void and of no effect. B. You shall not (and shall not permit other persons or entities to) reverse engineer, decompile, disassemble, merge, modify, include in other software or translate the Software, or use the Software for any commercial purposes, except for the publication of test results, as provided above. C. You shall not (and shall not permit other persons or entities to) remove or obscure Ziff's copyright, trademark or other proprietary notices or legends from any of the materials contained in this package or downloaded. 3. Limited Warranty and Limited Liability THE SOFTWARE IS PROVIDED "AS IS" WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESS OR IMPLIED, INCLUDING, WITHOUT LIMITATION, ANY WARRANTY OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS TO THE RESULTS AND PERFORMANCE OF THE SOFTWARE IS ASSUMED BY YOU, AND ZIFF ASSUMES NO RESPONSIBILITY FOR THE ACCURACY OR APPLICATION OF OR ERRORS OR OMISSIONS IN THE SOFTWARE. IN NO EVENT SHALL ZIFF BE LIABLE FOR ANY DIRECT, INDIRECT, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE USE OR INABILITY TO USE THE SOFTWARE, EVEN IF ZIFF HAS BEEN ADVISED OF THE LIKELIHOOD OF SUCH DAMAGES OCCURRING. ZIFF SHALL NOT BE LIABLE FOR ANY LOSS, DAMAGES OR COSTS, ARISING OUT OF, BUT NOT LIMITED TO, LOST PROFITS OR REVENUE, LOSS OF USE OF THE SOFTWARE, LOSS OF DATA OR EQUIPMENT, THE COSTS OF RECOVERING SOFTWARE, DATA OR EQUIPMENT, THE COST OF SUBSTITUTE SOFTWARE OR DATA, CLAIMS BY THIRD PARTIES, OR OTHER SIMILAR COSTS. THE ONLY WARRANTY MADE BY ZIFF IS THAT THE ORIGINAL PHYSICAL MEDIA IN WHICH THE SOFTWARE IS EMBODIED AND WHICH IS DISTRIBUTED BY ZIFF SHALL BE FREE OF DEFECTS IN MATERIALS AND WORKMANSHIP. ZIFF'S ENTIRE LIABILITY AND THE USER'S EXCLUSIVE REMEDY SHALL BE LIMITED TO THE REPLACEMENT OF THE ORIGINAL PHYSICAL MEDIA IF DEFECTIVE. THE WARRANTIES AND REMEDIES SET FORTH HEREIN ARE EXCLUSIVE AND IN LIEU OF ALL OTHERS, ORAL OR WRITTEN, EXPRESS OR IMPLIED. NO ZIFF AGENT OR EMPLOYEE, OR THIRD PARTY, IS AUTHORIZED TO MAKE ANY MODIFICATION OR ADDITION TO THIS WARRANTY. SOME STATES DO NOT ALLOW EXCLUSION OR LIMITATION OF IMPLIED WARRANTIES OR LIMITATION OF LIABILITY FOR INCIDENTAL OR CONSEQUENTIAL DAMAGES; SO THE ABOVE LIMITATIONS OR EXCLUSIONS MAY NOT APPLY TO YOU. 4. U.S. Government Restricted Rights The Software is licensed subject to RESTRICTED RIGHTS. Use, duplication or disclosure by the Government or any person or entity acting on its behalf is subject to restrictions as set forth in subdivision (c)(1)(ii) of the Rights in Technical Data and Computer Software Clause at DFARS (48 CFR 252.227-7013) for DoD contracts, in paragraphs (c)(1) and (2) of the Commercial Computer Software-Restricted Rights clause in the FAR (48 CFR 52.227-19) for civilian agencies, on in other comparable agency clauses. The contractor/manufacturer is the Ziff-Davis Benchmark Operation, One Copley Parkway, Suite 510, Morrisville, North Carolina 27560. 5. General Provisions Nothing in this Agreement constitutes a waiver of Ziff's rights under U.S copyright laws or any other Federal, state, local or foreign law. You are responsible for installation, management, and operation of the Software. This Agreement shall be construed, interpreted and governed under New York law. If any provision of this Agreement shall be held by a court of competent jurisdiction to be illegal, invalid or unenforceable, the remaining provisions shall remain in full force and effect. Trademarks Above DISK (R) is a registered trademark of Above Software. Computer Select (R), Computer Shopper (R), PC/Computing (R), PC Direct (R), PC Expert (R), PC Magazine (R), PC Week (R), and WinBench (R) are registered trademarks and DOSMark (TM), First Looks (TM), MacBench (TM), NetBench (TM), PC Bench (TM), PC Direkt (TM), PC Professionell (TM), ServerBench (TM), Windows (TM) Activ (TM), and Winstone (TM) are trademarks of Ziff-Davis Publishing Company, L.P. CompuServe (R) is a registered trademark of CompuServe, Inc. dBASE (R) and Paradox (R) are registered trademarks of Borland International, Inc. Harvard Graphics (R) is a registered trademark of Software Publishing Corp. Intel (R) is a registered trademark of Intel Corporation. Lotus (R) and 1-2-3 (R) are registered trademarks of Lotus Development Corporation. Microsoft (R) and MS-DOS (R) are registered trademarks and Windows (TM) and Windows (TM) Sources (TM) are trademarks of Microsoft Corporation. Micro Channel (TM) is a trademark of International Business Machines Corporation. SCO (R) is a registered trademark of The Santa Cruz Operation, Inc. UNIX (R) is a registered trademark of UNIX System Laboratories, Inc. WordPerfect (R) is a registered trademark of WordPerfect Corporation. Preface This manual tells you how to install, set up, and run PC Bench (TM) on your PC and how to interpret your results. In this preface, you can find: + A list of notes for this release of PC Bench. + A reading path telling you which chapters to read if you wish to understand the concepts behind PC Bench and if you need to run PC Bench and understand your results. + A summary of the chapters in this manual. + A list of conventions PC Bench and this manual use. Introducing PC Bench The Ziff-Davis Benchmark Operation (ZDBOp) developed PC Bench 8.0 as part of a family of benchmarks to help you evaluate your PC's performance. Each Ziff-Davis benchmark lets you test different aspects of a PC's performance. PC Bench is a subsystem-level benchmark you can use to measure the performance of your PC's processor, memory, disk, and video subsystems. The scores you get with PC Bench give you a measure of the performance of these subsystems when you run DOS applications. PC Bench's DOSMark (TM) provides a general, overall measure of your PC's performance under DOS. You can get copies of PC Bench and other Ziff-Davis benchmarks free of charge by contacting the Ziff-Davis Benchmark Operation (ZDBOp) or by downloading the benchmark you want from ZiffNet. To request benchmarks from ZDBOp, use the form included at the back of this manual. If you have technical questions about PC Bench, see Appendix C for information on how to contact ZDBOp. Notes about PC Bench Please note the following before you run PC Bench: 1. You must read and agree to the license information in the front of this manual before running PC Bench. The same licensing information appears on your PC's screen the first time you run PC Bench after you install it and in the README.TXT file included with PC Bench. If you do not agree to the licensing information, delete all copies of PC Bench. If ZDBOp sent you the PC Bench diskette, return it and all accompanying materials (including any documentation) to ZDBOp at the following address: Ziff-Davis Benchmark Operation One Copley Parkway, Suite 510 Morrisville, NC 27560 For more information on licensing PC Bench, see Chapter 2. 2. PC Bench version 8.0 is the most recent release. This release differs from previous releases in many important ways. As a result, you can only meaningfully compare PC Bench 8.0 results with other PC Bench version 8.0 results. Do not attempt to compare version 8.0 results with results from earlier versions of PC Bench. Also, the database format for PC Bench 8.0 is not compatible with the database formats of earlier versions of PC Bench. 3. PC Bench runs DOS-based tests. Its results are not meant to reflect your PC's performance when running Windows (TM)-based applications. 4. ZDBOp tested PC Bench 8.0 with the EMM386 and HIMEM.SYS memory managers that ship with DOS. While PC Bench may run if you have another memory manager loaded on your PC, ZDBOp does not recommend using PC Bench as a tool for testing or comparing the performance of memory managers. Who should read this manual This manual is for people who want to run PC Bench and understand what its scores mean about the overall performance of their PCs. If you use PC Bench, you should already be familiar with PCs in general. (If you have questions about your PC or the applications you run on it, consult the documentation that came with your PC or the applications.) You can easily find the information about PC Bench you need in this manual. Early chapters explain PC Bench basics, and later chapters provide a more detailed explanation of how PC Bench works. This organization lets you choose the degree of technical information you need without having to read everything in the manual. In addition, you can find the tasks you want to perform without reading the entire manual. For example, if you just want to view PC Bench results, read Chapter 4. The table that follows lists some general concepts and tasks you may need to understand and which chapter contains information about them. If you need to: Read these chapters: (Concepts) Get a general overview of PC Bench Chapter 1 Learn about the PC Bench tests Chapter 5 Understand what the scores mean Chapter 5 Understand ways to use your PC's scores Chapter 5 Learn about what affects your PC's scores Chapter 6 Understand the concepts behind PC Bench Chapter 7 Understand where to find other PC Bench results Chapter 8 (Tasks) Determine the requirements to run PC Bench Chapter 2 Install and set up PC Bench Chapter 2 License PC Bench Chapter 2 Start PC Bench Chapter 3 Run the PC Bench tests Chapter 3 Change your PC's test settings Chapter 3 Create a test suite Chapter 3 Save results Chapter 4 Compare results Chapter 4 Select a baseline system Chapter 4 Print results Chapter 4 Use functions in the PC Bench program Chapters 3, 4 View results from a test run Chapter 4 Use PC Bench with other Ziff-Davis benchmarks Chapters 1, 5 Get tips on running PC Bench Chapter 6 Get a list of the PC Bench tests Appendix A Recover from a PC Bench error message Appendix B Get technical support from ZDBOp Appendix C Request Ziff-Davis benchmarks Appendix C, Benchmark Request Form Summary of this manual The following list contains a summary of the information in each chapter of this manual. Chapter 1 An Overview of PC Bench Presents a brief description of PC Bench and gives an overview of the scores PC Bench produces. Chapter 2 Installing PC Bench Gives you the software and hardware requirements for PC Bench. Explains how to install PC Bench and set up your PC to run the benchmark. Also tells you how to license and register your copy of PC Bench. NOTE: In most cases, you must install PC Bench before you can read or print this manual. If, however, you received a copy of the manual without first installing PC Bench on your PC, you can use the instructions in this chapter to install it. Chapter 3 Running PC Bench Tells you the different ways to run the PC Bench tests on your PC. Explains how to run PC Bench in demo and batch modes. Gives instructions on how to create your own custom test suites. Chapter 4 Working with PC Bench's Results Tells you how to use the PC Bench program screen to view results and compare results. Explains how to save and print results. Tells you how to import results from other PCs. Chapter 5 Understanding PC Bench's Results Explains in detail the different scores PC Bench produces. Chapter 6 What Affects PC Bench's Results Provides tips on configuring your PC for optimal performance and tells you some factors that can influence your PC Bench scores. Chapter 7 The Concepts Behind PC Bench Explains how PC Bench measures the performance of your PC. Discusses the types of tests PC Bench runs. Chapter 8 How Your PC Compares Tells you where to find published results and standards to use in comparing the PC Bench scores for your PC. Appendix A The PC Bench Tests Gives you a list of all the tests that come with PC Bench. Appendix B Troubleshooting Tells you what to do if you must abort PC Bench while it is running a test. Lists some PC Bench error messages and tells you how to deal with those errors. Appendix C Technical Support Explains how to get technical support for PC Bench. Benchmark Request Form Ziff-Davis Benchmarks' Manual Order Form On-line documentation PC Bench comes with the following on-line documentation: README.TXT This file tells you how to license PC Bench and contains notes on PC Bench. You should read this information before you run PC Bench. PC Bench also includes two versions of this manual. They are: PCBENCH.DOC Microsoft (R) Word for Windows 2.0c version. PCBENCH.TXT ASCII text version; it does not contain any special formatting commands. (This file contains special touch-up work to help you print it.) With these two versions, you should be able to print a copy of this manual regardless of the type of software you have on your PC. Note, however, that the PCBENCH.TXT file is not formatted and does not contain illustrations of sample screens. NOTE: If you would like to order a printed copy of this manual, see the manual order form at the back of this book. In addition to these files, PC Bench includes an on-line Help system that contains most of the information in this manual. To access the PC Bench help topics, select Help from the main menu in the PC Bench program screen. Conventions PC Bench uses When you work with the PC Bench program screen, you can use some standard conventions with which you may already be familiar. For example, the PC Bench main menu bar uses drop-down menus to present options. When you select a menu title, such as "File," a drop-down menu appears and displays the menu items available. In many cases, the drop-down menus duplicate actions you can perform using PC Bench's function buttons in the PC Bench program screen. In some cases, the drop-down menus give you additional options, such as getting information on how to contact ZDBOp. PC Bench also uses several other standard conventions for selecting items from its main screen, windows, and dialog boxes. You can run PC Bench with or without a mouse attached to your PC. The following sections describe how to select items with a mouse and with keyboard commands. Using a mouse If you have a mouse attached to your PC, you can use it to select items by placing your cursor on the item you wish to select and then clicking your left mouse button. Generally, if you click on an item such as a function button or a test name in a list, you are selecting that item and PC Bench highlights it. If there is an action associated with the item you select (such as the Run function button), PC Bench takes the action associated with that item. For example, if you click on the PC Bench About function button, PC Bench displays a screen that tells you the names of the PC Bench development team. Using keyboard commands If you do not have a mouse attached to your PC or if you prefer not to use a mouse, you can: + Press the Alt key and the underlined access character of the item you want to select. (For example, to select the Print option in the File menu, press the Alt key, the "f" key, and then the "p" key.) This item can be one of the PC Bench function buttons, a menu bar title, or a menu item within a drop-down menu. + Press the Tab key or the Tab-Shift key combination to move the current focus from function to function on the PC Bench program screen. (You can identify the current focus by looking for the highlighted or outlined item on the screen.) PC Bench highlights a menu item by shading that item and highlights a function button by outlining the button with a dotted line. The status line at the bottom of the screen describes the current focus. When PC Bench highlights the item or function you want, press Enter to select that item. For example, if the current focus is on Help in the menu bar, when you press the Tab key, the focus goes to the About function button. You would then press Enter to bring up the About PC Bench screen. To de-select an item, press the Esc key. + Use the arrow keys to highlight a menu, submenu, or list item and then press the Enter key to select that item. For example, if you select the list of tests beside the Run function button on the PC Bench program screen, you can then use your PC's arrow keys to move the highlight in that list up or down. Once you highlight the test you wish to run, press the Enter key to select that test. General manual conventions This manual uses a few general conventions. The manual: + Portrays information you enter from your keyboard and system responses in a monospace or typewriter font. For example: C:\PCBENCH\BENCH.EXE + Uses italics to indicate variables where you must supply a value. For example: C:\PCBENCH\BENCH.EXE \Ofile indicates that you should enter a DOS file name after the \O option. + Uses the phrase "your PC" when talking about the PC you are testing. + Uses the term "highlight" an item to mean "move the focus to" an item. When you use keyboard commands to work with PC Bench, the item you wish to choose must be active before you take an action on that item. When an item is active, the focus for the screen is on that item. PC Bench highlights items in its screen by either outlining the item with a broken line (such as the Run function button) or shading the item (as with the File menu title in the menu bar). + Uses all capital letters to denote DOS file names and directories. For example, this manual contains sentences similar to the following: Before you install PC Bench, you should create a directory on your hard drive called \PCBENCH to hold the PC Bench files. + Separates the keys in the notation of a key combination with a hyphen (-). This means you do not have to press all the keys at the same time. For example, to select Save from the File menu, you would use the Alt-f-s key combination. You would first press the Alt key, then the "f" key, and finally, the "s" key. + Uses the term "select" to tell you when to click on an item using your PC's mouse or use a key combination to select an item. You can use your PC's mouse or keyboard to select items as the previous section describes. + Uses the following symbols to make it easier for you to know how to perform functions with and without a mouse: {SYMBOL 56 \f "Wingdings"} If you have a mouse attached to your PC, look for this symbol before instructional paragraphs. This symbol denotes paragraphs that tell you how to execute PC Bench functions using your mouse. (ZDBOp recommends you use PC Bench with a mouse if you have one available.) {SYMBOL 58 \f "Wingdings"} If you do not have a mouse attached to your PC, look for this symbol before instructional paragraphs. This symbol denotes paragraphs that tell you how to execute PC Bench functions from your keyboard. If an instructional paragraph does not have one of these symbols, the paragraph will use the term "select" as in the bulleted item above. In this case, you should use the method appropriate for your PC to select the item. Contacting ZDBOp You can mail or fax your questions about PC Bench to ZDBOp, or you can use ZiffNet, the Ziff-Davis on-line service to contact ZDBOp. Appendix C contains details on the different ways to contact ZDBOp. End of Preface Table of Contents Chapter 1: An Overview of PC Bench 1 The PC Bench basics 1 How PC Bench measures a PC's performance 2 Application profiling 3 Subsystem scores and the DOSMark 3 The subsystems PC Bench measures 4 Understanding the tests 5 The DOSMark 5 The Processor Score 5 The Video Score 6 The Disk Score 6 Using PC Bench results 7 Understanding the baseline system 8 Using PC Bench with other Ziff-Davis benchmarks 8 What's next 9 Chapter 2: Installing PC Bench 11 Hardware and software requirements 11 Installing PC Bench 12 Setting up your PC to run PC Bench 13 Licensing PC Bench 15 What's next 16 Chapter 3: Running PC Bench 17 What you should know before you run PC Bench 17 Changing your PC's test settings 18 Starting PC Bench 22 Using PC Bench command-line switches 23 Exiting PC Bench 25 Using the PC Bench program screen 26 Selecting tests 28 Running the PC Bench tests 30 Running the DOSMark Test Suite 32 Running other PC Bench built-in test suites 32 Running all the PC Bench tests 33 Running selected PC Bench tests 33 Running PC Bench in batch mode 34 Running PC Bench in demonstration mode 36 Running the tests included as utilities 37 Saving your test results 38 Creating and running custom test suites 38 Creating a custom test suite 38 Editing a custom test suite 40 Using previously created test suites 41 What's next 42 Chapter 4: Working with PC Bench's Results 43 Understanding the graph buttons 44 Comparing results 47 Selecting a baseline system 48 Adding a set of results to the display 50 Importing PC Bench results 51 Removing a set of results from the displayed results 54 Deleting results from the PC Bench database 56 Viewing your PC's configuration information 57 Printing results 59 What's next 61 Chapter 5 Understanding PC Bench's Results 63 The PC Bench scores 63 The DOSMark 64 The Processor Score 65 The Video Score 65 The Disk Score 66 The Memory Tests results 66 Other PC Bench tests 67 Why results differ 68 Publishing PC Bench results 69 Using PC Bench results 69 Using PC Bench with other benchmarks 70 What's next 71 Chapter 6: What Affects PC Bench's Results 73 Software and hardware factors that influence PC Bench's results 73 Tips for improving your PC Bench results 75 Following proper testing procedures 75 Other ways to improve your PC Bench results 76 Using PC Bench to best advantage 76 What's next 77 Chapter 7: The Concepts Behind PC Bench 79 The design goal 79 Profiling applications 80 Understanding the "market-centered" concept 81 Using a weighted harmonic mean 81 Understanding the Disk Mix 82 PC Bench's Protected Mode Large Instruction Mix 82 Correlating the score to applications 83 What's next 83 Chapter 8: How Your PC Compares 85 Checking published results 85 Using PCs within your organization 86 Publishing PC Bench scores 87 What's next 87 Appendix A: The PC Bench Tests 89 Processor tests 89 EGA/VGA text mode tests 91 EGA/VGA graphics mode tests 92 Disk tests 93 Disk throughput tests 93 DOS disk access 99 Memory timing tests 99 Tests included as utilities 102 Battery Rundown Test 102 Time the Timer Test 102 VGA Compatibility Test 102 EGA/VGA Monitor Quality Test 103 Appendix B: Troubleshooting 105 What to do if you have problems 105 Running PC Bench with memory managers 107 Running PC Bench in batch mode 107 PC Bench error messages 108 General database errors 112 Appendix C: Technical Support 113 Requesting a benchmark 113 Ordering a printed manual 114 Contacting ZDBOp 114 Benchmark Request Form Ziff-Davis Benchmarks' Manual Order Form Acknowledgments Chapter 1 An Overview of PC Bench This chapter presents a brief description of PC Bench and the results it produces. After reading this chapter, you should have a basic understanding of how PC Bench measures your PC's performance and what its results mean. NOTE: PC Bench is available free of charge on a 3.5" diskette. To request a copy of PC Bench or any other Ziff-Davis benchmark, use the form at the back of this manual. You can also download the benchmark you want from ZiffNet. See Appendix C for more information. The PC Bench basics PC Bench is a subsystem-level, synthetic benchmark you can use to measure the performance of your PC's processor, memory, disk, and video subsystems. PC Bench's results give you a measure of the performance you can expect from your PC when you run DOS applications. PC Bench's DOSMark provides an overall measure of your PC's performance. The PC Bench tests reflect the most common operations that best-selling DOS applications perform. ZDBOp developed the PC Bench tests by profiling such applications to determine the types of operations the applications typically perform. Thus, PC Bench's results correlate closely with the performance of actual applications. PC Bench's user interface lets you enter information about your PC, run tests, create test suites, and compare results. You perform most PC Bench functions from the PC Bench program screen. Once you run a test, PC Bench displays its results in the PC Bench program screen in graph format. (For more information on how to use the PC Bench program screen, see Chapters 3 and 4.) NOTE: PC Bench version 8.0 is the most recent release of this product. This release differs from previous releases in many important ways. As a result, you can only meaningfully compare PC Bench version 8.0 results with other PC Bench version 8.0 results. Do not attempt to compare your PC's version 8.0 results with results from earlier versions of PC Bench. Also, the database format for PC Bench 8.0 is not compatible with the database formats of earlier versions of PC Bench. How PC Bench measures a PC's performance PC Bench tests your PC's processor and disk subsystems using test mixes of operations that closely reflect the typical operations DOS applications perform. A "mix" is a single test consisting of a combination of operations. PC Bench's mixes use operations that subsystems execute when running applications. NOTE: The PC Bench video subsystem tests do not use mixes; they report raw performance results for different VGA video operations. PC Bench then weights the most important video results and includes them in its DOSMark. ZDBOp modeled the PC Bench tests on operations like those generated by DOS applications. PC Bench does not run applications during its tests. The sections below explain briefly how the PC Bench tests measure your PC's performance. For more information on the types of tests in PC Bench, see the section "Understanding the tests" later in this chapter. Application profiling The individual tests that PC Bench contains reflect extensive profiling of leading DOS applications. Profiling is a method of monitoring an application to see the types of operations it performs. By using application profiles as targets, PC Bench's tests can accurately reflect what the applications do. Extensive profiling research provided information on processor, memory, disk, and video access patterns in complex usage scenarios in actual applications. ZDBOp used the information from profiling to improve PC Bench's testing techniques with such features as: + A new 16-bit Protected Mode Large Instruction mix. As applications have changed, PC Bench's instruction-mix technology has changed with them. The large working set size (approximately 450 KB) of this mix lets it accurately reflect application CPU and RAM usage. + A new Disk Mix technology. PC Bench's Disk Mix accurately reflects the way today's applications use a PC's disk subsystem. The Disk Mix performs a broad range of disk operations over a large set of files. The Disk Mix works well and accurately with the software and hardware disk caches used in today's PCs, including such disk caching tools as SMARTDRV. + More accurate weights that provide realistic measures of a PC's performance. For example, the individual video tests weight 8- and 16-bit writes to both the text and graphics region of VGA display memory. These tests provide an accurate portrayal of the way DOS applications actually use a PC's video subsystem. Subsystem scores and the DOSMark PC Bench provides four main scores you can use to measure your PC's performance: the DOSMark score and the Processor, Video, and Disk subsystem scores. PC Bench produces each subsystem score from a different set of tests. (For more information on these tests, see the next section, "Understanding the tests.") The DOSMark is a weighted, harmonic mean of the Processor, Video, and Disk subsystem scores. PC Bench also weights the results of the scores of its individual tests to produce the subsystem scores. Using a harmonic mean provides a way to aggregate the scores for the individual tests into a weighted, overall score. PC Bench measures the total number of test operations and then divides that number by the total time it takes to perform these operations. PC Bench then weights this number to account for the different levels of importance each operation within a test group has. Basically, the more frequently an application uses an operation, the more weight PC Bench gives its test for that operation. NOTE: PC Bench also reports an absolute value for each individual test you run. The subsystems PC Bench measures PC Bench measures your PC's processor, memory, disk, and video subsystems. Each of these subsystems includes different components. In evaluating these subsystems, PC Bench considers both the software and hardware that make up the subsystem. The processor and memory subsystems include the main processor chip (for example, Intel{SYMBOL 226 \f "Symbol"} 80486 DX), any internal or external cache and cache controller, and main memory. The processor subsystem also includes the math co-processor if your PC has one. The disk subsystem includes the hard disk, the disk controller, and any hardware and software disk caches your PC uses. Finally, the video subsystem is the video adapter, whether on the motherboard or a plug-in card. Understanding the tests The following sections explain briefly the different PC Bench built-in test suites and the scores these tests produce. (A test suite is a group of tests you run at once with a single command.) For complete information on the different PC Bench tests you can run and what the different scores mean about your PC's performance, read Chapter 5. (If you need to run the PC Bench tests, skip to Chapter 3.) The DOSMark When you execute the DOSMark Test Suite, PC Bench runs the tests necessary to produce the Processor, Video, and Disk scores and then calculates the weighted harmonic mean of these three scores to produce a single number called the DOSMark. The DOSMark provides an overall measure of your PC's performance when you run DOS applications on your PC. You can use the DOSMark to compare your PC with other PCs. The Processor Score The core of PC Bench's Processor Score is the 16-bit Protected Mode Large Instruction Mix. This new Large Instruction Mix reflects the larger working-set sizes of today's applications. The approximate working-set size of this mix is 450 KB. ZDBOp designed this mix to accurately simulate the way popular DOS applications use a PC's processor. The 16-bit Protected Mode Large Instruction Mix reflects the way applications stress a PC's CPU subsystem in such areas as bus utilization, CPU cache usage, CPU cache interface, external cache usage, main memory, and the CPU instruction set. PC Bench's built-in Processor Test Suite also includes tests for the floating point unit (FPU). When you run this suite, PC Bench takes the results of the CPU and FPU tests and calculates a weighted harmonic mean. The weighted result reflects the relative effect of a PC's CPU and FPU on typical applications. NOTE: PC Bench includes a math co-processor test that requires an FPU. If your PC does not have an FPU, PC Bench uses FPU-emulation for this particular test. Results with FPU-emulation will be significantly slower than those with an FPU present. The Video Score The Video Score is a combination of different individual video tests. The Video Test Suite uses low-level video tests to accurately measure how your PC's video subsystem performs when running applications under DOS. When you run the Video Test Suite, PC Bench takes the results for the individual video tests in the test suite and calculates the weighted harmonic mean of those results. The resulting score is your PC's Video Score. The individual video tests measure 8- and 16-bit VGA write operations involving both text and graphics. The Video Score is a single number that provides an overall measure of your PC's DOS video performance. The Disk Score The Disk Score lets you gauge the performance of your PC's disk subsystem. That subsystem includes the hard disk, disk controller, and any hardware or software disk caches. PC Bench 8.0 uses a new disk-testing methodology: its Disk Mix. The PC Bench Disk Mix performs a range of disk operations over a large set of files. This mix uses over 100 files spread across six directories. The Disk Mix works with multiple files concurrently, creates and deletes files, moves data in groups of different sizes, intersperses write and read operations, and generally uses the disk subsystem the way applications do. The Disk Mix works with a PC's disk subsystem via the standard Int 21H interface, so you can use the Disk score as an accurate measure regardless of whether you are using a software or hardware disk cache. ZDBOp based PC Bench's Disk Mix on disk usage in profiled applications. Thus, the Disk Mix works with your PC's disk subsystem the way applications do. Using PC Bench results PC Bench produces a result for each individual test it runs even if that test is part of a test suite. For example, if you run the DOSMark Test Suite, you will get an absolute result for each individual test in the suite in addition to the DOSMark for your PC. You can use the different results PC Bench produces to compare your PC with others and to see how well your PC handles certain operations. PC Bench stores its results in a database, and you can use results in this database for comparison. This ability lets you: + Run PC Bench multiple times on the same PC with different configurations and compare the results for each test run. For example, you could run PC Bench on a PC with and without software disk caching and compare the disk scores. + Compare PC Bench results for your PC to the results for other PCs. One way to compare PC Bench results for different PCs is to use published results from Ziff-Davis publications. (For a list of these publications, see Chapter 8.) Also, your organization may have more than one type of PC or may have PCs with different configurations. If so, you can run PC Bench on the different machines within your organization to compare their performance. For more information on using results from different PCs, see the section "Importing PC Bench results" in Chapter 4. NOTE: If you want to compare different PCs in the same way, make sure all the PCs you want to compare were set up the same way. For example, make sure you do not compare the results of a PC that had disk caching enabled and a PC that did not unless you are trying to gauge the effect of the disk cache. Otherwise, your comparison may not reflect the true potential of both systems fairly. For more suggestions on ways to use your PC Bench results, see the section "Using PC Bench results" in Chapter 5. Understanding the baseline system When you display results in the PC Bench program screen, you can choose which set of results you want to use as a baseline system. Once you set a baseline system, this is your comparison system for all other sets of results. PC Bench sets the baseline system's results to 100% and displays all other results as a percentage of the baseline system's results. When you use a baseline system, you can quickly determine from the displayed graphs how results for different PCs compare. For complete information on how to set a baseline system, see the section "Selecting a baseline system" in Chapter 4. Using PC Bench with other Ziff-Davis benchmarks Although PC Bench tests the speed of a PC's processor, memory, disk, and video subsystems to determine how they perform when they use DOS applications, it does not give a complete measure of your PC's performance if you also use Windows-based applications. To get that information, you can use PC Bench in combination with two other Ziff-Davis benchmarks: WinBench (R) and Winstone (TM) 94. + You can use WinBench to measure the performance of your PC's graphics and disk subsystems under Windows. ZDBOp based the WinBench tests on profiles of leading Windows applications. Thus, the WinBench tests provide an accurate measurement of your PC's graphics and disk performance under Windows. + Use Winstone 94 to measure the overall performance of your PC as it runs actual Windows-based applications. With Winstone 94 you can measure the effect that changing individual system components has on your PC's overall performance when running Windows and Windows-based applications. Winstone 94 performs tasks within the applications it runs like those typical users perform with those applications. Winstone 94 is the sole Microsoft-certified benchmark for the testing of industry-standard PCs running Windows 3.1-based applications. You can use the form at the back of this manual to request copies of WinBench, Winstone 94, and other Ziff-Davis benchmarks from ZDBOp. You can also download Ziff-Davis benchmarks from ZiffNet. For more information, see Appendix C. What's next Now that you have an overview of PC Bench and how it tests a PC, read Chapter 2 to install PC Bench and set up your PC to run the benchmark. Chapters 3 and 4 give details on how to run the PC Bench tests and how to view and compare results. If you would like further details on the PC Bench tests and what the results mean about your PC's performance, read Chapter 5. End of Chapter Chapter 2 Installing PC Bench This chapter tells the hardware and software requirements your PC must meet to run PC Bench successfully. It also explains how to set up your PC to run the benchmark and tells you how to install and license PC Bench. After you read this chapter and install PC Bench, you should be ready to run the PC Bench tests. NOTE: You can also learn how to install PC Bench by reading the README.TXT file that comes with PC Bench. In most cases, you will have installed PC Bench before printing a copy of this manual. However, if you are installing PC Bench on another PC or if you received a copy of this manual before you installed PC Bench, then you can follow the installation instructions in this chapter. Hardware and software requirements PC Bench version 8.0 requires the following minimum hardware and software components: + A PC running MS-DOS{SYMBOL 226 \f "Symbol"} version 3.1 (or compatible) or higher. For best results, use MS-DOS version 5.0 (or compatible) or higher. + A 386 (or compatible) CPU or better. NOTE: While you can run PC Bench on PC with a 286 CPU, you will not be able to run all the PC Bench tests. PC Bench 8.0 requires a minimum of a 386 CPU. + A minimum of 6.6 MB of free disk space to install the PC Bench files. + A minimum of 35 MB of free disk space on each drive you want to test. + A minimum of 543 KB of free conventional memory if you plan to run all the PC Bench tests. + A minimum of 2 MB of extended memory to run the DOSMark tests. + A VGA video adapter. Installing PC Bench You can request a copy of PC Bench on a diskette from ZDBOp, or you can download it from ZiffNet, Ziff-Davis's on-line service. (See Appendix C for more information.) PC Bench comes as a self-extracting file. To install PC Bench: 1. Create a directory called \PCBENCH on your PC's disk. NOTE: You do not have to create a separate PC Bench directory called \PCBENCH, but ZDBOp recommends you do so. 2. Change directories (CD) to the directory you just created. 3. Put the PC Bench diskette in your PC's floppy drive and execute the following command. PCBENCH.EXE For example, if your PC's floppy drive is the A drive, you would type: A:\PCBENCH.EXE (If you downloaded PC Bench from ZiffNet, change directories to the directory where you stored the PCBENCH.EXE file and then execute the file.) PC Bench installs its files in the current directory. 4. Type the following command from the directory where you installed PC Bench to start PC Bench's database with a set of sample comparison results: CLEARDB NOTE: You do not have to execute the CLEARDB command to run PC Bench. However, if you do not execute this command, the PC Bench database will not contain any sample comparison results when you start PC Bench. Whether the database contains sample results will not affect how PC Bench runs, but ZDBOp recommends you execute this command. Setting up your PC to run PC Bench Before you run PC Bench, you need to modify your CONFIG.SYS and AUTOEXEC.BAT files. To do this: 1. Make backup copies of your original CONFIG.SYS and AUTOEXEC.BAT files (you can use file names such as CONFIG.SAV and AUTOEXEC.SAV). 2. Modify your CONFIG.SYS file so it contains the statement FILES=20. Do not omit this statement. You can include a statement for a mouse driver or any other drivers necessary for your PC to operate. You should not load any non-essential Terminate and Stay Resident programs (TSRs) or memory managers. NOTE: While PC Bench may run if you have a memory manager loaded on your PC, ZDBOp does not recommend using PC Bench as a tool for testing or comparing the performance of memory managers. If you run PC Bench with a memory manager loaded, and you have problems running the tests, you should re-run the tests without the memory manager loaded. The following is a sample minimal CONFIG.SYS file: FILES=20 If you usually run DOS applications on your PC with HIMEM.SYS and DOS loaded high (DOS=HIGH), then you should run PC Bench the same way. For example, your CONFIG.SYS file would then be similar to the following: DEVICE=C:\WINDOWS\HIMEM.SYS FILES=20 DOS=HIGH 3. Modify your AUTOEXEC.BAT file so it contains only the essential drivers and programs necessary to operate your PC. For example: PATH=C:\DOS PROMPT $P$G C:\MOUSE\MOUSE You may also load a disk caching tool, such as SMARTDRV, from your AUTOEXEC.BAT file. NOTE: If you normally run DOS applications on your PC with a disk caching tool enabled, you should run PC Bench the same way.) Your file would then also contain a line similar to the following: C:\WINDOWS\SMARTDRV.EXE Remember, after you modify these system files you will need to reboot your PC for the changes you made to take effect. You can reboot your PC by pressing Ctrl-Alt-Del, by pressing the PC's reset button (if there is one), or by turning off the power to the PC and restarting it. Licensing PC Bench Before you run PC Bench, you must license and register it. By licensing PC Bench, you agree to the requirements and terms of its license agreement. NOTE: The first time you run PC Bench, it displays the PC Bench License Agreement. (You will find the same license agreement at the front of this manual and in the on-line README.TXT file.) If you wish to display the license agreement at any time, go to the directory where you installed PC Bench and enter the command: BENCH \L PC Bench displays its license agreement on your PC's screen. Figure 2-1: The PC Bench License Agreement {EMBED MSDraw \* mergeformat|} Use the space bar to page through the license agreement as you read it. Once you have read the entire agreement, PC Bench prompts you to register your copy of PC Bench by entering your name and your organization's name (if any). After you enter this information, press F10 or the Enter key to continue. NOTE: You only need to license your copy of PC Bench once. If you choose not to adhere to the PC Bench License Agreement, delete all copies of PC Bench from your PC. If you loaded PC Bench from a diskette rather than by downloading it from ZiffNet, return the diskette and any documentation to ZDBOp at the following address: Ziff-Davis Benchmark Operation One Copley Parkway, Suite 510 Morrisville, NC 27560 What's next Now that you have installed PC Bench, you can read Chapter 3 to learn how to run PC Bench's tests and how to change PC Bench's settings. Chapter 4 tells you how to view and compare PC Bench results. Chapter 5 discusses the different PC Bench tests and what the scores mean about your PC. End of Chapter Chapter 3 Running PC Bench This chapter explains what you need to know before you run PC Bench, how to change your PC's test settings, how to start PC Bench, and how to use functions in the PC Bench program screen. It also tells you how to run the PC Bench tests on your PC. After reading this chapter, you should know how to run PC Bench on your PC. What you should know before you run PC Bench Before you run PC Bench, make sure you: + Change your PC's CONFIG.SYS and AUTOEXEC.BAT files as described in the section "Setting up your PC to run PC Bench" in Chapter 2. + License and register your copy of PC Bench. Also, please note the following before you run any PC Bench tests: 1. ZDBOp recommends you change the Machine ID and Machine Name fields in PC Bench's Machine Information dialog box to names that are meaningful to you. (For more information on how to change this information, see the next section, "Changing your PC's test settings.") 2. PC Bench version 8.0 does not support EGA resolutions or EGA-only adapters. 3. If you wish to run all of PC Bench's tests in succession, note the following: + All disk tests will attempt to run on the same drive, because PC Bench lets you set just one drive for the disk tests. + To run either the Disk Throughput Tests or the Disk Mix requires a minimum of 32 MB of free disk space. + You will only get results for the BIOS Disk Access Tests if the drive parameter in the Disk Settings section of the Machine Information dialog box is set to a primary disk partition (usually drive C:). + The DOS File Access test will not run on a networked drive. 4. If you have disk caching tools installed and enabled on your PC, your PC Bench results for the disk tests will typically be better than results without the disk caching. If you normally use your PC with a disk caching tool enabled, then use the same tool with PC Bench and it will give you a realistic measurement of how DOS applications interact with your PC's disk. To get a true measurement of the performance of just your hard disk itself, run PC Bench without disk caching enabled. Changing your PC's test settings PC Bench lets you define information about your PC before you start a test run. You can set this information from within PC Bench using the Settings function button. By changing this information from test run to test run, you can easily identify different test runs. To change your PC's test settings for a PC Bench test run, select the Settings function button. {SYMBOL 56 \f "Wingdings"} Click on the Settings function button in the PC Bench program screen. {SYMBOL 58 \f "Wingdings"} Highlight the Settings function button using the Tab key and then press the Enter key to select the button. (The status message will say "Configure Machine ID and Disk Test settings.") You can also use either the keyboard combination Alt-e to select the Settings function button or the keyboard combination Alt-s-t to select the Settings option from the Test Suites menu. PC Bench displays the Machine Information dialog box. Figure 3-1: The Machine Information dialog box {EMBED MSDraw \* mergeformat|} From the Machine Information dialog box you can define some of the settings PC Bench uses when it runs a test. PC Bench initially defines some of the settings in this box with default values. You can change any of the default values you wish. For example, you can set a machine name to make that machine's results easy to identify when you view and compare test results. (ZDBOp recommends you change at least the Machine ID and Variant fields for each test run.) Once you have the variables in this dialog box the way you want, select the Use button. NOTE: PC Bench uses the values you specify in the Machine Information dialog box when you run tests. PC Bench does not, however, commit these values to its database until you save results. If you change your PC's settings before you save results, PC Bench updates all unsaved results. Thus, when you do save the results, PC Bench commits all the results to the database with the same settings. Note also that after you save results or restart PC Bench, PC Bench preserves all the settings from your last PC Bench session in the Machine Information dialog box except the Variant fields. PC Bench automatically changes the Variant fields to distinguish the current results from previous test runs. You do not have to enter any information in this dialog box to run PC Bench. Doing so, however, helps you distinguish between different PCs in the database and different set of results for a single PC. If you want to test the same PC under different configurations, you can use the Variant fields to identify each set of results. PC Bench uses the current time and date to initialize these fields for each new set of results. You can manually set one or more of the Variant fields using your own naming conventions. PC Bench then records the results as separate entries in its database. The Machine Information dialog box contains two sections: Machine Settings and Disk Settings. The following list summarizes the fields in the Machine Settings section. + Machine ID is an alphanumeric entry up to 12 characters long you can use to identify your PC. By default, this field is set to a string that identifies your PC's CPU type (for example, 80486). + The five Variant boxes each contain an alphanumeric entry up to three characters long. You can change these entries to make different PC Bench test runs on the same PC easily identifiable. By default, the values in these fields contain the date and time you either started this session of PC Bench or last saved PC Bench results. (The date and time format is mm dd hh mm ss.) For example, for a date and time stamp of November 28, 15:30:10, the Variant fields would be as follows: Variant 1 = 11 Variant 2 = 28 Variant 3 = 15 Variant 4 = 30 Variant 5 = 10 You can change any of these fields to any entry up to three characters long. + Machine Name is an alphanumeric entry up to 12 characters long you can use to record the brand name and model of your PC. By default, the entry in this field identifies your PC's CPU type (for example, 80486). + Project is a numeric entry up to seven characters long. If the tests you are running are part of a particular project for your organization, you may want to enter that project's name here, or you can just use this field for additional descriptive information about your PC. + Test Org is an alphanumeric entry up to 25 characters long containing the name of the organization performing the test. This field defaults to the organization name you entered when you licensed PC Bench. + Tester is an alphanumeric entry up to 25 characters long containing the name of the person performing the test. This field defaults to the licensee name you entered when you licensed PC Bench. + Machine Description is an alphanumeric string up to 224 characters that can contain any pertinent information about your PC. To enter information here, position your cursor in the text box and begin typing. The Machine Information dialog box also contains a section where you can specify settings for PC Bench's disk tests. In the Disk Settings section you can set the following fields: + Disk and Path. These are the disk and directory names where PC Bench will build its test files for the Disk Mix tests. The disk where PC Bench builds its test files is also the disk PC Bench tests. You can select the disk and directory name using drop-down menus. {SYMBOL 56 \f "Wingdings"} Click on the down-arrow beside a test box to display the list of disks and directory names available. {SYMBOL 58 \f "Wingdings"} Highlight either the disk or path text boxes using the Tab key and then press the down-arrow key to display the list of disks and directory names available. Use the up and down arrow keys to highlight your selection and then press the Enter key to select the highlighted item. + Repetitions. This tells PC Bench how many times to perform the disk test(s). By default, this field is set to one. In almost all cases, you will not need to increase the number of repetitions of the disk tests; however, if you want PC Bench to repeat the disk tests, you can type the number of repetitions you want in this text box. Starting PC Bench To run PC Bench, go to the directory where you installed PC Bench (ZDBOp recommends \PCBENCH) and enter the following command from the DOS command line: BENCH NOTE: Before you run PC Bench, you must license it. The PC Bench License Agreement and User Information screens appear the first time you start PC Bench after installing it. Once you license and register your copy of PC Bench, you do not need to do so again regardless of how many times you start PC Bench. If you would like to re-license your copy of PC Bench, you can start PC Bench with the \L option to display the license agreement. For information on how to license and register your copy of PC Bench, see the section "Licensing PC Bench" in Chapter 2. PC Bench briefly displays the Ziff-Davis benchmark stopwatch logo and PC Bench copyright information. PC Bench then displays its program screen. NOTE: If you would like to pass by these screens quickly, you can press the Enter key once your PC begins to display the stopwatch logo or the copyright information and PC Bench will move to the next screen. Using PC Bench command-line switches There are several command-line switches you can use with PC Bench. These switches let you direct PC Bench's actions. For example, you can use the /B option to run PC Bench in batch mode. The PC Bench command-line switches are: /? or /H Displays a list of the switches PC Bench accepts and names the DOS environment variables it uses. /L Displays the license agreement and lets you re-register PC Bench. (PC Bench displays its license agreement automatically the first time you run the benchmark if you have not previously licensed and registered your copy of PC Bench.) /B Runs all the PC Bench tests in batch mode using the current DOS environment variables. You can set these variables from the DOS command line when you start PC Bench or from a DOS batch file. The variables you set for PC Bench when you run it in batch mode let you distinguish different sets of results. When you run PC Bench in batch mode, you can use the following DOS environment variables: + PIN (machine ID) + VARIANT1 through VARIANT5 (test variants) + TYPE (machine name) + PROJECT (project name) + TESTORG (test organization) + TESTER (tester's name) If you run PC Bench in batch mode multiple times on the same PC, you can use these variables to identify different test runs by changing variables from one test run to another. You must set the PIN and VARIANTS DOS environment variables, and these variables must be unique for each set of results. /B /Ffile Executes PC Bench in batch mode using the tests listed in the test suite file. The file name must be a legal DOS file name and must have a .STE extension. If the file is not in the PC Bench directory, you must supply the full path name for the file. You can create the test suite file from PC Bench's program screen by using the Edit and Save options in the Test Suites menu. NOTE: If you use the /B option alone, PC Bench will run all its tests. If you use the /B option with the /Ffile option, PC Bench will run only the tests specified in the test suite file. /M Prevents PC Bench from logging your PC's configuration information. You should not stop PC Bench from logging your PC's configuration information unless you are having problems running PC Bench. /Ofile Logs your PC's results to a CSV (comma separated value) format file called file. You can then use this file to view or print test results. You can supply the path name where you want PC Bench to save the file. The file name must be a legal DOS file name with a .CSV extension. /P Prevents PC Bench from automatically determining what type of processor is in your PC. This automatic determination can cause problems on future PCs with new processors or on DOS emulators. NOTE: These PC Bench command-line switches are not case sensitive. Exiting PC Bench You can exit the PC Bench program screen by: + Selecting the Exit function button. + Selecting Exit from the File menu. + Using the key sequence Alt-x. NOTE: If you have unsaved test results when you exit PC Bench, it will ask if you want to save your results. You can select Save, Delete, or Cancel in this warning box. If you choose to save your results, it may take a few moments before PC Bench completes the save and the DOS prompt appears. (For example, some laptops may take over one minute to save test results and display the DOS prompt.) Using the PC Bench program screen When you start PC Bench, its program screen appears on your PC's screen. Figure 3-2: The PC Bench program screen {EMBED MSDraw \* mergeformat|} This screen has three main parts: the menu bar at the top, the function buttons on the left side, and the results graph buttons on the right side. When you execute a test suite, a set of tests, or an individual test, PC Bench displays status messages in a fourth small area, the bar at the bottom of the screen, to let you know what it is doing. You can select the actions you want to perform using either the drop-down menus from the menu bar or the function buttons. NOTE: When you work with the PC Bench program screen, you can use either your mouse or your keyboard to select the function buttons and menu bar items in the program screen. (It is usually easier to use PC Bench with a mouse.) Note also that if you have not yet run any PC Bench tests or if you have not run enough tests to compute the DOSMark or subsystem scores, the graph buttons for the DOSMark and subsystem scores will read "(Insufficient Data to Compute Score)". In addition, the Memory Tests graph button will read "(No Results)" when there are no Memory Tests results. As PC Bench completes a test, it displays a horizontal bar graph of the result of that test. Once a test run is complete, PC Bench updates the DOSMark and subsystem scores and displays those scores on the results graph buttons in the main program screen. (For more information on how to work with and understand your PC's results, see Chapter 4.) The PC Bench program screen includes the following function buttons: + About PC Bench. Selecting this button gives you information about the PC Bench development team. You can also view this information by selecting About PC Bench from the Help menu. + System Info. Selecting this button opens the System Information dialog box that displays your PC's configuration. You can also display the System Information dialog box by selecting Display System Info from the File menu. + Run. Selecting this button starts the test suite visible in the drop-down list beside the function button. You can display a list of the available test suites by selecting the down arrow to the right of this text box. Choose the test suite you want and then select the Run button to execute it. NOTE: If you want to run a single test or set of tests without creating a test suite, go to the menu bar and select Choose Test from the Test menu. For more information, see the section "Running selected PC Bench tests" later in this chapter. + Compare Results. Selecting this button displays the Machine Browser dialog box, which contains a list of all the results in the PC Bench database. To display a set of results, choose the results you want and then select the Load button. Once you select the Load button, PC Bench returns to the main program screen and displays the set of results you chose in the results graph buttons. You can load only one machine's results at a time from this window. To add another comparison machine's results to the graph buttons, select Compare Results again. You can display up to two results in addition to your current results on the PC Bench graph buttons. You can also add comparison results to the display by selecting Add Comparison System from the Graph menu. (For complete information on how to view and compare PC Bench results, see Chapter 4.) + Settings. Selecting this button displays the Machine Information dialog box. From this box, you can change database information about your PC that PC Bench uses to identify results you save. You can also change this information by selecting Settings from the Test Suite menu. + Help. Selecting this button opens the PC Bench help system. You can also display a list of help topics by selecting the Help menu. + Exit. Selecting this button exits the PC Bench program screen and ends the current PC Bench session. You can also exit PC Bench by selecting Exit from the File menu. Selecting tests Before you run a PC Bench test, you must select which test suite or individual tests you want to run. You can select tests you want to run in two ways. + You can select a test suite or All Tests from the list of test suites in the drop-down list to the right of the Run function button. {SYMBOL 56 \f "Wingdings"} Click on the down arrow to the right of the drop-down list. Use the scroll bar beside the list to move up and down the list. Select the test suite you want by clicking on its name. {SYMBOL 58 \f "Wingdings"} Press the Tab key until PC Bench highlights the drop-down list. (The status message will say "Select group of tests to run.") Once you highlight the text box, press the down-arrow key to display the list. Then, highlight the test you want using your PC's arrow keys to move up and down the list. Once you highlight the test you want, press the Enter key to select it. NOTE: With PC Bench, you can also create custom test suites containing selected individual tests and groups of tests. For more information, see the section "Creating and running custom test suites" later in this chapter. + You can select individual tests or groups of tests from the Tests menu. {SYMBOL 56 \f "Wingdings"} Click on Choose Test from the Tests menu. PC Bench lists the individual tests in the Test Selection window. You can use the scroll bars and the Page Up and Page Down keys to move up and down the list. To select a test, click on the test's name. To select a group of tests, click on the test group title. (The individual tests names are indented under the group title.) PC Bench places a check mark ({SYMBOL 252 \f "Wingdings"}) beside each test you select. To de-select a test, click on it again. Continue with this process until you have selected all the tests you wish to run. To select all the tests in the list, click on the Select All button. To clear your current selections, click on the Clear All button. Once you have selected the test(s) you want, click on the Run button. {SYMBOL 58 \f "Wingdings"} Use the Tab key to highlight the Tests menu option. (The status message will say "Run tests quickly.") Then press the down-arrow key to display the drop-down menu. Press the Enter key to select Choose Test. (You can also use the Alt-t-t key combination to display this window.) From the Test Selection window, use your PC's arrow keys and the Page Up and Page Down keys to scroll through the list of tests. When you highlight a test you want to run, press the Enter key to select that test. To select a group of tests, highlight the test group title and then press the Enter key. (The individual tests names are indented under the group title.) PC Bench places a check mark ({SYMBOL 252 \f "Wingdings"}) beside each test you select. To de-select a test, highlight the test name again and press the Enter key. Continue with this process until you have selected all the tests you wish to run. To select all the tests in the list, use the Tab key to highlight the Select All button and then press the Enter key. To clear your current selections, highlight the Clear All button and then press the Enter key. Once you have selected the test(s) you want, highlight the Run button and press the Enter key to start the test(s). (You can also use the underlined key sequences to perform these functions.) As PC Bench runs tests, it displays status messages at the bottom of the screen. When all the tests end, PC Bench displays the scores for the group tests on the results graph buttons. To view the individual test results for a group of tests, select the appropriate graph button (Processor, Video, Disk, or Memory). (For more information on how to view results, see the section "Understanding the graph buttons" in Chapter 4.) Running the PC Bench tests With PC Bench, you can run several kinds of tests. For example, you can run: + One of the PC Bench built-in test suites: DOSMark, Processor, Video, or Disk. Each of these test suites contains a specific set of PC Bench individual tests. NOTE: A test suite is a group of tests you run at once with a single command. + Your own custom test suites. PC Bench lets you create your own test suites using the individual PC Bench tests. + One of the individual tests or a predefined group of these tests. You can run a group of individual tests without creating a test suite. + All the PC Bench tests. + PC Bench in batch mode. You can execute PC Bench from a DOS batch file using command-line options. When you run PC Bench this way, you do not have to give commands to PC Bench as it runs its tests. + PC Bench in demonstration mode. PC Bench runs continuously in demo mode until you stop it. The individual tests report unweighted, absolute results. The DOSMark and subsystem scores are weighted, harmonic means of those results. Any time you run individual tests that are a part of a previously run DOSMark or subsystem test suite, PC Bench displays the absolute results for the individual test, recalculates the weighted scores, and updates the PC Bench display to reflect the new scores. For example, if you run just the 16-bit Protected Mode Large Mix, PC Bench will automatically recalculate your DOSMark, because it includes this test in the DOSMark calculation. You can run any combination of tests during a single interactive PC Bench session. For example, you could run the DOSMark tests, run several individual tests, run a custom test suite, and then run the Disk tests. NOTE: When PC Bench is running a test, it disables your PC's mouse and keyboard. Note also that if you run PC Bench from a batch file, you can run all the tests at once, a single test suite, or PC Bench in demo mode. For more information, see the sections "Running PC Bench in batch mode" and "Running PC Bench in demonstration mode" later in this chapter. The following sections provide details on these options for running PC Bench. Running the DOSMark Test Suite To run the DOSMark Test Suite, select DOSMark from the list of test suites next to the Run function button. Then select the Run button. The DOSMark provides a general measure of your PC's performance. It includes tests from the Processor, Video, and Disk Test Suites. (For more information on these suites, see the next section.) The DOSMark tests include several disk-intensive tests that can take quite a while to run. Depending on your PC's configuration, it could take as long as 45 minutes, an hour, or even longer. A top-of-the-line, correctly configured 66-MHz 486-based PC should take less than 15 minutes to run the DOSMark Test Suite. For more information on the DOSMark tests and what this score means about your PC, see Chapter 5. Running other PC Bench built-in test suites PC Bench contains three other built-in test suites you can run: the Processor, Video, and Disk tests. Each of these tests provides a general measure of its respective subsystem. To run one of these test suites, select the appropriate test suite name from the list of test suites next to the Run button. Then select the Run button. For more information on these built-in test suites, read Chapter 5. PC Bench also includes a separate group of Memory Tests. You run the Memory Tests by selecting Memory from the list of tests next to the Run function button and then selecting the Run button. NOTE: PC Bench includes the separate Memory Tests as additional performance analysis tools, but the scores of these tests do not figure in the calculation of the DOSMark. Running all the PC Bench tests To run all the PC Bench tests, select All Tests from the list of tests. Then select the Run button. Running all the tests can take a long time. On some PCs, it can take as long as two hours to run all the tests. A top-of-the-line, correctly configured 66-MHz 486-based PC should take less than an hour to run all the PC Bench tests. Once you start the tests, you do not need to interact with PC Bench again until the tests are complete. While running all the PC Bench tests will produce many individual results for your PC, you can get a general measure of your PC's performance, without running all the tests. In most cases, the DOSMark will provide you with enough information to evaluate your PC's overall performance. NOTE: You should be aware of the following before you run all the PC Bench tests on your PC: + All disk tests will attempt to run on the same drive, because PC Bench lets you set just one drive for the disk tests. + To run either the Disk Throughput Tests or the Disk Mix requires a minimum of 32 MB of free disk space. + You will only get results for the BIOS Disk Access Tests if the drive parameter in the Disk Settings section of the Machine Information dialog box is set to a primary disk partition (usually drive C:). + The DOS File Access test will not run on a networked drive. For a list of all the PC Bench tests, see Appendix A. Running selected PC Bench tests To run specific tests or sets of tests, select Choose Test from the Tests menu. PC Bench displays the Test Selection window. Figure 3-3: Selecting tests {EMBED MSDraw \* mergeformat|} In the Test Selection window, select the test or tests you wish to run. To select a group of tests, select the test group title (such as Processor Tests). The list indents the individual tests names under each group title name. When you select a test, PC Bench marks that test with a check mark ({SYMBOL 252 \f "Wingdings"}). To de-select a test, simply choose that test again. PC Bench removes the check mark. To select all the tests in the list, select the Select All button. To de-select all of your current selections, select the Clear All button. To exit the window, select the Close button. Once you have selected the test or tests you wish to run, select the Run button at the bottom of the dialog box. PC Bench then runs the tests you marked. Running PC Bench in batch mode You can execute PC Bench from a DOS batch file using command-line options. A batch file contains a series of commands DOS will perform automatically once you execute the file. When you run PC Bench this way, you do not have to interact with PC Bench as it runs its tests, and PC Bench automatically logs the test results in its database. If you have more than one drive on your PC, your batch file should first specify the letter of the drive that contains the directory where you loaded PC Bench. Next, the batch file should change to that directory (CD) (probably \PCBENCH) and execute the BENCH command with the /B option and any other options you choose. Other options you can use on the command line in the batch file are: /Ffile, /M, /Ofile, and /P. For a description of these command-line options, see the section "Using PC Bench command-line switches" earlier in this chapter. Sample batch files The following paragraphs contain sample batch files. You must set the PIN environment variable and all the VARIANT environment variables from your batch file. (For more information on the DOS environment variables you can use with PC Bench, see the section "Using PC Bench command-line switches" earlier in this chapter.) 1. The following batch file changes to the \PCBENCH directory, sets the necessary environment variables, starts PC Bench in batch mode from your PC's hard disk, disables system information logging, and saves the test results in the file PC1.CSV. CD \PCBENCH SET PIN=PC25 SET VARIANT1=1 SET VARIANT2=0 SET VARIANT3=0 SET VARIANT4=0 SET VARIANT5=0 BENCH /B /M /OPC1.CSV 2. The following batch file changes to drive D:, changes to the \PCBENCH directory, sets the necessary environment variables, starts PC Bench in batch mode, and runs the tests listed in the test suite file TEST1. D: CD \PCBENCH SET PIN=PC25 SET VARIANT1=2 SET VARIANT2=0 SET VARIANT3=0 SET VARIANT4=0 SET VARIANT5=0 BENCH /B /FTEST1.STE 3. The following batch file changes to the \PCBENCH directory, sets the necessary environment variables, runs PC Bench in batch mode from your PC's hard disk, prevents PC Bench from automatically determining your processor type, and saves the test results in the file PC16.CSV. CD \PCBENCH SET PIN=PC3 SET VARIANT1=1 SET VARIANT2=2 SET VARIANT3=0 SET VARIANT4=0 SET VARIANT5=0 BENCH /B /P /OPC16.CSV Running PC Bench in demonstration mode Once you install PC Bench, you can run it in demonstration mode. To execute PC Bench in demo mode, select Run Demo from the File menu. To halt demo mode press the Esc key. NOTE: When you execute PC Bench in demo mode, it runs the current test suite or set of tests listed in the drop-down list beside the Run function button. Thus, you can select which test you want PC Bench to run in its demo by changing the current test suite. Once PC Bench completes a test run in demo mode, it displays the test results for that run briefly on the results graph buttons and then runs the same test suite again. PC Bench runs continually in demo mode until you stop it. Running the tests included as utilities PC Bench includes a set of tests you can use as utilities. These tests are the Battery Rundown, the Time the Timer, the VGA Compatibility, and the EGA/VGA Monitor Quality tests. You can not execute these tests from within PC Bench. Instead, you must execute them from the DOS command line. The tests PC Bench includes as utilities are: + The Battery Rundown Test. To run this test, execute the following command from the directory where you installed PC Bench: RUNDOWN.EXE + The Time the Timer Test. To run this test, execute the following command from the directory where you installed PC Bench: TIMER.EXE + The VGA Compatibility Test. To run this test, execute the following command from the directory where you installed PC Bench: VGAHDW.EXE + The EGA/VGA Monitor Quality Test. To run this test, execute the following command from the directory where you installed PC Bench: EGAVGA.EXE For more information on these tests, see Appendix A. Saving your test results To commit the current test results to PC Bench's database, select Save from the File menu. PC Bench automatically updates its database with the results information for your current session and stores your PC's configuration information with these results. NOTE: Because PC Bench logs your system configuration information each time you save results, it may take a while to complete the save. If you exit PC Bench without saving your current results, PC Bench displays a warning message to remind you that you have not saved your current results. You can select Save, Delete, or Cancel in this warning box. Creating and running custom test suites A test suite is a set of tests you execute with a single command. PC Bench lets you create your own test suites. This set of tests can consist of any combination of the PC Bench tests you want. You can place both individual tests and test groups in your test suite. You can also import previously created test suites. The following sections describe how to create and edit test suites and how to use previously created test suites. Creating a custom test suite To create a custom test suite: 1. Select a test suite from the list of test suites beside the Run button that is similar to the type of test suite you wish to create. If there is not a test suite similar to the type you want to create, you may wish to select All Tests. At least this way, you can begin with all the tests selected and de-select the tests you do not want in your test suite. 2. Select Edit from the Test Suites menu. PC Bench displays the Test Selection window. Figure 3-4: The Test Selection window {EMBED MSDraw \* mergeformat|} This window displays a list of the PC Bench tests you can use in your custom test suite. PC Bench marks the tests defined in the currently selected test suite with a check mark ({SYMBOL 252 \f "Wingdings"}). 3. Select the test or tests you wish to use in your custom test suite. To de-select all the tests currently marked, select the Clear All button. To select all the tests in the list, select the Select All button. To de-select a single test, choose the test's name again to remove the check mark. 4. Select the Close button once you have defined the tests for your test custom suite. The list of test suites beside the Run function button will display "," indicating that there is a new test suite you have not yet saved. NOTE: You can run the tests you selected at this point without saving the test suite by selecting the Run function button. However, if you do not save the test suite, it will not be available for future PC Bench sessions. 5. Select Save from the Test Suites menu to save the tests you defined in the Test Selection window. NOTE: When you save your custom test suite, PC Bench also saves your disk test settings. If you wish to save your test suite with different disk test settings, either select the Settings option from the Test Suite menu or use the Settings function button before you save the results. (Both options will open the Machine Information dialog box.) 6. Enter a name for your new test suite in the Save File As window and then select the Save button. The name you enter here should be a legal DOS file name with no file name extension. PC Bench automatically appends the extension .STE to the file name you enter here. PC Bench saves the information you entered and adds that test suite file name to the list of test suites in the drop-down list beside the Run button. You can use your custom test suite during your current PC Bench session or in a future PC Bench session. To run the test suite, simply select the test suite name from the list and then select the Run button. Editing a custom test suite You can change the tests that comprise an existing test suite by adding or deleting tests. To do this: 1. Select the test suite you wish to edit from the list of test suites beside the Run function button. 2. Select Edit from the Test Suites menu. PC Bench displays the Test Selection window. This window lists all the available PC Bench tests. The tests that make up the test suite you selected are marked with a check mark ({SYMBOL 252 \f "Wingdings"}). 3. Select or de-select tests. Once you have the test suite defined the way you want, select the Close button. 4. Select Save from the Test Suites menu to save the edited test suite. 5. Overwrite the old test suite in the Save File As dialog box by selecting that test suite file name. PC Bench automatically updates the test suite file. The next time you run the test suite, it will run the new set of tests you selected. Using previously created test suites To use a previously created test suite, simply copy the file you want into the PC Bench directory on your PC. If you are getting the test suite file from another PC, you can copy the file by putting it on a diskette, or if you are on a network, you can copy the file across the network. Make sure the file name has a .STE extension. Once the test suite file is in the PC Bench directory on your PC, the test suite name will appear in the list of tests the next time you run PC Bench. NOTE: Be aware that if a test suite file from another PC uses a disk drive not available on your PC, PC Bench will use your PC's current drive for the test. What's next Now that you know how to run the PC Bench tests, Chapter 4 tells you how to view and compare results. If you would like information on the different test suites that come with PC Bench or if you would like a better understanding of what the tests do, read Chapter 5. End of Chapter Chapter 4 Working with PC Bench's Results This chapter explains how you can work with PC Bench results. This chapter contains sections that tell you how to: + Read and understand the test results in the results graph buttons in the PC Bench program screen. + Compare results from different test runs on your PC or other PCs. + Print the results for a test run. + Save test results to the PC Bench database. As you read this chapter, you should understand that the current baseline system is always the set of test results to which PC Bench compares all other test results. PC Bench lets you select which set of results you want to use as your baseline system. The baseline system can be your current set of results or a set of results you saved from a different run of the tests. By default, PC Bench sets the baseline system's results to 100% and displays all other sets of results as a percentage of that PC's results. For example, since the baseline system is 100%, a slower set of results appears as less than 100% and a faster set of results appears as higher than 100%. (For complete information on setting your baseline system, see the section "Selecting a baseline system" later in this chapter.) Understanding the graph buttons When PC Bench completes a test run, it displays its results in graph format on the graph buttons in the PC Bench Results portion of the program screen. Figure 4-1: PC Bench Results {EMBED MSDraw \* mergeformat|} NOTE: If you have not yet run any PC Bench tests or if you have not run enough tests to compute the DOSMark or subsystem scores, the graph buttons for the DOSMark and subsystem scores will read "(Insufficient Data to Compute Score)". In addition, the Memory Tests graph button will read "(No Results)" when there are no Memory Tests results. Note also that if you exited a previous session of PC Bench while PC Bench was displaying a set of results, PC Bench displays that same set of results the next time you start a session. Each graph for the Processor, Video, and Disk scores is actually a large button. There is also a separate button for any Memory Tests results. (PC Bench includes the separate Memory Tests as additional performance analysis tools, but the scores of these tests do not figure in the calculation of the DOSMark.) When you select a graph button, PC Bench then displays the bar graph results for the individual tests associated with the button you selected. NOTE: If you have a color monitor, the titles for the DOSMark, Processor, Video, and Disk scores are blue. The blue titles indicate which tests PC Bench includes in the DOSMark tests. If you do not have a color monitor, the titles for these graph buttons will be slightly lighter than the title for the Memory Tests button, since the Memory Tests results are not part of the DOSMark. Each graph button for tests included in the DOSMark supplies the following information: + The name of the test PC Bench ran. + The name of your PC as set in the Machine Information dialog box. NOTE: You can change the graph so the name it displays for the PC is either the Machine ID, the Machine Name, or the Variants set when you ran the test. To change this view, select Show Name, Show ID, or Show Variants from the Graph menu. + The absolute score your PC received on the test. (If you use the Compare Results feature, you will also see a score for each set of results you choose to add to the graph.) The absolute score is the number to the left of the bar graph. + A bar graph displaying the score as a percentage. If you are displaying only one set of results, those results are set to 100%. If you are displaying more than one set of results and if you have not selected a baseline system, PC Bench sets the largest set of results for any given test to 100%. If you have chosen a baseline system, the bar graph score for each set of results is relative to the bar graph score for the baseline system. By default, the set of results for the first tests you run become the baseline system. (For more information on the baseline system, see the section "Selecting a baseline system" later in this chapter.) NOTE: To display the bar graph according to the absolute numbers instead of a percentage, select Value from the Graph menu. To see results for the individual tests PC Bench ran as part of a group of tests, select the button for the appropriate test group. For example, to view the results for the Processor Tests, select the graph button for the Processor Score. {SYMBOL 56 \f "Wingdings"} Click directly on the graph button for the appropriate group of tests whose results you want to view. For example, if you want to view your PC's scores for the individual video tests, click on the Video Score graph button. {SYMBOL 58 \f "Wingdings"} Use the Tab key to highlight the appropriate graph button and then press the Enter key to select that graph's button. For example, if you want to view your PC's scores for the individual processor tests, use the Tab key to highlight the Processor Score graph button and then press the Enter key. You can also use the key combination Alt-v to move directly to the graph buttons and to toggle between the graph buttons and the graphs for the individual tests. Once you select the appropriate graph button, PC Bench displays that group's individual tests results. If there is more than one page of results, you will see a Next Page button at the bottom of the screen. You can use the Next Page (and Previous Page) buttons to page through the results screens for the individual tests. Figure 4-2: The Next Page and Previous Page buttons {EMBED MSDraw \* mergeformat|} For more information on what the PC Bench results mean about your PC's performance, see Chapter 5. For more information on how to use the graph buttons to compare results, see the next section. Comparing results To compare different PC Bench results, select the Compare Results function button from the PC Bench program screen. PC Bench displays the Machine Browser window. Figure 4-3: The Machine Browser window {EMBED MSDraw \* mergeformat|} From this window, you can select which set of results you want to display in the PC Bench Results portion of the screen. PC Bench lists all the results available in its database in this window. NOTE: If you run PC Bench repeatedly on the same machine, the Machine ID or Variants for each test run must be unique. PC Bench automatically makes each test run unique by changing the Variant fields. You can, however, modify these fields from the Machine Information dialog box and use your own naming scheme or convention. (For more information, see "Changing your PC's test settings" in Chapter 3.) For example, if you are testing a PC with different software disk caches, you could use one variant field to identify the different types of disk cache and another variant field to identify the size of the disk cache. The following sections explain how to compare, view, and manipulate the results PC Bench displays. Selecting a baseline system You can compare all the results to one set of results. PC Bench considers this set of results the baseline system. PC Bench lets you choose which set of results to use as your baseline system. PC Bench sets those results to 100% and displays all other results in the graphs according to how they compare with this baseline system's results. For example, the bar graph for a set of results that are faster than the baseline system will be more than 100%, while the bar graph for a set of results that are slower than the baseline system will be less than 100%. To specify which set of results you want PC Bench to use as the baseline system or to change the current baseline system, select Select Baseline System from the Graph menu. Figure 4-4: Selecting the baseline system {EMBED MSDraw \* mergeformat|} PC Bench displays a submenu listing the machine names for your current test run and for all comparison results you have added to the display. (For more information on how to add a set of results to the display, see the next section, "Adding a set of results to the display.") At this point, you can either select the machine name whose set of results you wish to use as your baseline system, or you can select "Largest." {SYMBOL 56 \f "Wingdings"} Click on Graph in the menu bar and then click on Select Baseline System. When PC Bench displays the list of available machine names in the cascading menu, select the one you want by clicking on it. {SYMBOL 58 \f "Wingdings"} Use the Alt-g-s key combination to display the cascading menu. Use your PC's up and down arrow keys to highlight the machine name you want and then press the Enter key to select it. If you select "Largest," PC Bench sets the highest results for any given individual test as the baseline system for that test. Using "Largest" is a quick and easy way to see which results for which tests are the highest. NOTE: Because PC Bench lets you select the baseline system, you can change the baseline system to a different set of results at any time. You may wish to change the baseline system to compare results for different tests. You can also display the results graphs relative to the absolute numbers instead of a percentage by selecting Value from the Graph menu. NOTE: To see the differences in machine results better, you can use the Zoom In and Zoom Out options in the Graph menu. The zoom feature is useful when viewing absolute numbers. When you zoom in, PC Bench displays a more detailed view of the bar graph results. For example, if PC Bench is displaying the results in a percentage and the scale numbers are 100% and 200%, when you zoom in, PC Bench displays the scale numbers as 50% and 100%. By zooming in, you can read the bar graph results more easily when two tests have results that are very close. When you zoom out, PC Bench compresses the size of the bar graph so more of the graph displays in the same amount of space. For example, if PC Bench is displaying the results in a percentage and the scale numbers are 100% and 200%, when you zoom out PC Bench displays the scale numbers as 250% and 500%. Adding a set of results to the display You can add a set of results to the display for comparison with your current results or with other results. You can display up to three sets of results: your current results and two other comparison systems. To add a comparison system: 1. Select either the Compare Results function button or Add Comparison System from the Graph menu. (Selecting either will give you the same dialog box.) {SYMBOL 56 \f "Wingdings"} Click on the Compare Results function button or click on Add Comparison System from the Graph menu. {SYMBOL 58 \f "Wingdings"} Use either the Alt-c or the Alt-g-a key combination. PC Bench displays the Machine Browser dialog box, which lists all the results available in the database. 2. Select the results you want to add to the results graph buttons. {SYMBOL 56 \f "Wingdings"} Click on the machine name and then click on the Load button. {SYMBOL 58 \f "Wingdings"} Use your PC's up and down arrow keys to highlight the machine name and then use the Alt-l key combination to load that PC's results. (You can also use the Tab key to highlight the Load button and then press the Enter key to select it.) PC Bench adds the results you selected to the results displayed on the graph buttons. Importing PC Bench results If you have access to PC Bench results for other PCs, you can import those results into your PC Bench database and then use them as comparison points with your PC's results. PC Bench includes batch files you can use to set up an import database in your PC Bench directory. Before you execute these batch files, you should put the results you want to import in a directory accessible from your PC. NOTE: You can import only one PC Bench database of results at a time. To import results: 1. Execute the following command from the directory where you installed PC Bench (probably \PCBENCH): SETUPIMP DIR where DIR is the pathname for the database files you wish to import. NOTE: The pathname should end with a backslash (\). For example, if the database files you want to import are in C:\PC1\RESULTS, you would enter the following command: SETUPIMP C:\PC1\RESULTS\ Or if the import database files are on a floppy diskette, and your PC's floppy drive is drive A, you would enter the following command: SETUPIMP A:\ The directory you specify in the pathname should contain the following PC Bench database files: MACHINE_.DBF RESULT_.DBF SYSINFO_.DBF MACHINE_.MDX RESULT_.MDX SYSINFO_.MDX NOTE: DO NOT copy the files you want to import directly into your PC Bench directory. If you do, you will overwrite your database. PC Bench automatically renames these files as it copies them to your PC Bench directory so that you do not overwrite your own existing database files. 2. Start PC Bench and select Import from the File menu. {SYMBOL 56 \f "Wingdings"} Click on Import from the File menu. {SYMBOL 58 \f "Wingdings"} Use the Alt-f-i key combination. NOTE: If the import database is very large, you will need to free as much conventional memory as possible to load the list of results into the Import Machine Browser window. If you get a memory error while trying to import results, exit PC Bench, free up memory, and try importing the results again. PC Bench displays the Import Machine Browser window showing you the different sets of results from the other PC. Figure 4-5: The Import Machine Browser window {EMBED MSDraw \* mergeformat|} 3. Select the set of results you wish to import into the PC Bench database on your PC. {SYMBOL 56 \f "Wingdings"} Click on the set of results and then click on the Import button. {SYMBOL 58 \f "Wingdings"} Use your PC's up and down arrow keys to highlight the set of results, press the Enter key to mark those results, and then use the Alt-i key combination to import that PC's results. You can also use the Tab key to highlight the Import button and then press the Enter key. You can import all the results in the other PC's database by selecting the All button at the bottom of the screen. NOTE: Depending on the number of results in the database you are importing, it may take a while to import them. You can repeat the above steps for each database of results you want to import. Once you import the results you want, you can save disk space by deleting the imported database files. To delete these files, execute the following command from the PC Bench directory: CLEARIMP This command deletes the database files created by the SETUPIMP command. Removing a set of results from the displayed results To remove a set of comparison results from the PC Bench Results portion of the screen: 1. Select Remove Comparison System from the Graph menu. {SYMBOL 56 \f "Wingdings"} Click on Graph in the menu bar and then click on Remove Comparison System. {SYMBOL 58 \f "Wingdings"} Use the Alt-g-r key combination. PC Bench displays a cascading menu that lists all the results PC Bench is currently displaying. Figure 4-6: Removing a set of results from the display {EMBED MSDraw \* mergeformat|} 2. Select the set of results you want to remove from the display. {SYMBOL 56 \f "Wingdings"} Click on the machine name, machine ID, or variants. {SYMBOL 58 \f "Wingdings"} Use your PC's up and down arrow keys to highlight the machine name, machine ID, or variants, and then press the Enter key. NOTE: The type of display you have chosen from the Graph menu (Show Name, Show ID, or Show Variants) determines what type of identifier appears in this list. PC Bench removes that set of results from the graph buttons and re-displays the results. It does not, however, delete these results from the PC Bench database. You can display these results again if you choose to do so. Deleting results from the PC Bench database To delete results from the PC Bench database permanently: 1. Select Delete Results from the File menu. {SYMBOL 56 \f "Wingdings"} Click on File in the menu bar and then click on Delete Results. {SYMBOL 58 \f "Wingdings"} Use the Alt-f-d key combination. PC Bench displays the Machine Browser window. In it are the sets of results in the database that do not currently appear in the PC Bench Results portion of the screen. NOTE: You cannot delete a set of results you are viewing from the PC Bench database. To delete a set of results you are viewing, first remove that set of results from the display using the Remove Comparison System option from the Graph menu. 2. Select the set of results you want to remove from the database. {SYMBOL 56 \f "Wingdings"} Click on the set of results and then click on the Delete button. {SYMBOL 58 \f "Wingdings"} Use your PC's up and down arrow keys to highlight the set of results, press the Enter key to mark those results, and then use the Alt-d key combination to delete the results. You can also use the Tab key to highlight the Delete button and then press the Enter key. PC Bench deletes that set of results from the database. Viewing your PC's configuration information Because your PC's configuration affects your PC Bench scores, PC Bench automatically logs configuration information when you save results. To view the configuration information, select the System Info function button from the PC Bench program screen. NOTE: You can also view your PC's configuration information by selecting the Display System Info option from the File menu (Alt-f-i). PC Bench displays your PC's configuration information in the System Information dialog box. Figure 4-7: The System Information dialog box {EMBED MSDraw \* mergeformat|} NOTE: When you select the System Info function button for the first time during your current PC Bench session, PC Bench must gather all the information about your PC. Thus, it may take PC Bench a while before it displays the System Information window. The System Information window contains data on the configuration that produced each set of results currently loaded in the PC Bench Results portion of the screen. You can also view the CONFIG.SYS and AUTOEXEC.BAT files for any of these sets of results. NOTE: When you view information in this window, you can use the scroll bar on the right to page up and down through the information in each screen. The following list explains briefly the information you can get from each of the buttons in this window. + The Hardware button. By default, when you open the System Information window, PC Bench displays your PC's hardware information. This screen tells you information about your PC's microprocessor type and CPU clock speed, plus other information about your PC's hardware. + The System button. When you select this button, PC Bench displays information about your PC's operating system type and version as well as information about any disk caching software you have loaded on your system. You can also find the current date and time in this screen plus other information about your PC's software. + The Memory button. Selecting this button displays information about your PC's memory subsystem. For example, this screen tells you the amount (in kilobytes) of your PC's conventional, expanded, and extended memory. + The Disk button. When you select this button, PC Bench displays information on your PC's disk subsystem. For example, in this screen you find the size (in bytes) of your PC's fixed disk. The screen also provides details on the disk partitions of the fixed disk (such as the size, in megabytes, and type of partitions). + The Config.sys and Autoexec.bat buttons. These buttons display your PC's current CONFIG.SYS and AUTOEXEC.BAT files. (To edit these files, you must exit PC Bench and use a DOS file editor.) Printing results Once you have added comparison results to the PC Bench Results portion of the screen, you can print those results to a file or to your local printer. To print test results, select Print from the File menu. PC Bench displays the Print Results dialog box. Figure 4-8: Printing results {EMBED MSDraw \* mergeformat|} This box lists all the sets of results currently loaded in the PC Bench Results portion of the screen. You can print only one set of results at a time. To print a set of results: 1. Select which set of results you want to print from the list at the top of the Print Results dialog box. An "X" appears beside the results you select. {SYMBOL 56 \f "Wingdings"} Click on the box beside the set of results. {SYMBOL 58 \f "Wingdings"} Use the Tab key to highlight the box beside the set of results and then press the Enter key. 2. Mark whether you want to print the set of results to a file or to a local device (such as a printer). An "X" marks your selection. The list of devices in the Device drop-down list are PRN, LPT1, LPT2, COM1, and COM2. Which device you select depends on which of your PC's ports the printer is connected to. To print the results to a file, specify a file name in the box to the right of File. The file name must be a legal DOS file name. PC Bench saves the file in the current directory. To enter this information: {SYMBOL 56 \f "Wingdings"} Click on the box beside either Device or File. You can display a list of devices by clicking on the down arrow to the right of the drop-down list beside Device. {SYMBOL 58 \f "Wingdings"} Use the Tab key to highlight the drop-down list to the right of either Device or File and then press the Enter key. To display a list of devices, use your PC's right arrow key to move the focus to the list and then press the down arrow key to display the list. 3. Select the Print button to print the set of results. If you choose to print your PC's results to a file, you have two choices for your file's format: ASCII text (.TXT) or comma separated values (.CSV) format. Enter a legal DOS file name without an extension in the text box beside the File option. PC Bench automatically adds the correct extension to the file name you enter (either .TXT for ASCII text or .CSV for comma separated value). NOTE: If you print your PC's results to a file using CSV format, you can read that CSV file into many database or spreadsheet programs to view or reformat your results. Note also that when you print a set of results, the print output and the file will contain a "Not Run" entry for each individual test PC Bench did not run. For example, if you print a set of results that contains results for only the Processor tests, PC Bench lists all the other tests as "Not Run." What's next Now that you understand how to view and compare PC Bench results, Chapter 5 provides details on what operations the PC Bench tests perform and what your PC's scores mean about system performance. Chapter 6 lists some of the factors that might influence your PC's results and gives tips on how to configure your PC to obtain the best possible PC Bench results. If you would like to know more about the concepts behind PC Bench's development, see Chapter 7. End of Chapter Chapter 5 Understanding PC Bench's Results This chapter discusses the different PC Bench tests and how you can use their results. After reading this chapter, you should know what the PC Bench results mean about your PC's performance. For a list of the tests that come with PC Bench and a complete description of each test, see Appendix A. The PC Bench scores Which PC Bench tests you should choose to run depends on how you want to test your PC. The following sections describe the different types of PC Bench tests and how these tests measure your PC's performance. (For information on how to run the PC Bench tests, see the section "Running the PC Bench tests" in Chapter 3.) The DOSMark The DOSMark provides an overall measure of your PC's performance. It consists of tests that reflect how your PC performs when running DOS applications. To produce a DOSMark, PC Bench runs the tests necessary to produce the Processor, Video, and Disk scores and then calculates the DOSMark as a weighted harmonic mean of those three scores. You can get an overall measure of your PC's performance, and you can also consult the individual test results that make up the DOSMark to understand better how your PC's subsystems perform. NOTE: If you have a color monitor, you will notice the titles for the DOSMark, Processor, Video, and Disk graph buttons are in blue. The blue titles indicate which tests PC Bench includes in the DOSMark. If you do not have a color monitor, the titles for these graph buttons will be slightly lighter than the title for the Memory Tests button. PC Bench uses tests based on application profiling to gauge your PC's performance, so the DOSMark represents a very close portrayal of application performance without forcing you to run a large set of applications. NOTE: Because PC Bench's profiling used a set of representative applications, individual applications you use on your PC may not reflect that same performance behavior as the DOSMark. For example, if your PC has a DOSMark twice that of another PC, it does not mean every individual application will run twice as fast on your PC as on the other PC. However, your overall DOS application performance will be twice as fast. You can use the DOSMark to compare your PC with other DOS-compatible PCs. You can also use the DOSMark to compare the effect of hardware or software changes or upgrades on your PC's overall performance. The Processor Score The Processor Score is a measure of how your PC's processor and memory subsystems perform. The core of PC Bench's Processor Score is the 16-bit Protected Mode Large Instruction Mix. This mix reflects the larger working-set sizes that today's applications use. The 16-bit Protected Mode Large Instruction Mix stresses a PC's bus, CPU cache, CPU cache interface, external cache, main memory, and CPU instruction set in the same way as top-selling DOS applications. The Processor Test Suite also includes a test of the performance of your PC's floating point unit (FPU). When you run the Processor Test Suite, PC Bench takes the results for the CPU and FPU tests and calculates a weighted harmonic mean to produce the overall Processor Score. By weighting these results, PC Bench can reflect the relative importance of the CPU and FPU to profiled applications. NOTE: PC Bench includes a Math Coprocessor test that requires an FPU. If your PC does not have an FPU, PC Bench uses FPU-emulation for this particular test. Results with FPU-emulation will be significantly slower than those with a hardware FPU. The Video Score The Video Score provides a measure of your PC's VGA video subsystem performance. The Video Score is a combination of several different individual video tests. PC Bench uses low-level video tests to measure how your PC's VGA subsystem performs when running applications under DOS. When you run the Video Test Suite, PC Bench takes the results for the individual video tests in the test suite and calculates a weighted harmonic mean of those results. These individual tests include 8- and 16-bit VGA write operations involving both text and graphics. The Video results give you a breakdown of the different types of DOS video calls that applications make. The Disk Score The Disk Score lets you gauge the performance of your PC's disk subsystem, which includes the hard disk, disk controller, and any hardware or software disk caches. PC Bench 8.0 uses a new disk testing methodology, its Disk Mix. ZDBOp based the Disk Mix on the disk usage of a set of profiled, leading DOS applications. The Disk Mix works with your PC's disk subsystem the way actual applications do. The Disk Mix performs a range of disk activities involving a large set of files. It works with over 100 files spread across six directories. The Disk Mix works with multiple files concurrently, creates and deletes files, reads and writes file data in different sizes, intersperses write and read operations, and generally uses the disk subsystem the way applications do. The Disk Mix works with a PC's disk subsystem via the standard DOS Int 21H interface, so you can use the Disk Score as an accurate measure regardless of whether you are using a software or hardware disk cache. The Memory Tests results The Memory Tests measure the speed with which your PC can access various types of memory. In general, the Memory Tests results reflect the speed and architecture of your PC's memory, memory caches (if any), and CPU type. These tests use a series of low-level memory read and write operations to conventional, extended, graphic, text, and BIOS memory. PC Bench does not weight the Memory test results as part of its DOSMark. These tests are, however, are useful tools for evaluating memory subsystem speed. NOTE Although PC Bench includes a separate set of Memory Tests, ZDBOp recommends you use the Processor Test Suite to measure memory performance in conjunction with the processor. ZDBOp includes the Memory Tests with PC Bench version 8.0 as additional informational and diagnostic tools. Other PC Bench tests The four PC Bench overall weighted scores (the DOSMark and the Processor, Video, and Disk Scores) use only a subset of the tests available with PC Bench. PC Bench contains a large number of individual tests that measure the performance of different aspects of your PC's subsystems. NOTE: For a complete list of the individual tests included with PC Bench and a brief description of each test, see Appendix A. For instructions on how to run a PC Bench test, see the section "Running the PC Bench tests" in Chapter 3. You can combine the individual tests in any way you want and run them. You can also create your own test suite using these individual tests. Some of the individual PC Bench tests are: + The Disk Throughput Tests. If you want to measure your PC's disk subsystem performance when manipulating files with specific sizes in specific block sizes, use the Disk Throughput Tests. You should note, however, that due to the low-level approach of these tests, they may not give the best reflection of a PC's disk performance when you are using a disk cache. + The Disk Access Tests. The BIOS Disk Seek Tests (sequential and random) measure mechanical track-to-track disk drive access times. The tests use the BIOS Int 13H. You cannot use this test to check floppy diskettes or mass storage media that do not support the standard Int 13H interface. The tests involve 1,000 sequential seeks (alternating between cylinders 0 and 11) and 1,000 random seeks. Caching disk controllers and certain SCSI adapters that do not force the disk mechanism to move for disk seeks will yield unrealistically fast times on this test. The DOS Disk Access Test measures the time necessary to perform 1,000 read requests at random locations on the disk using the DOS Int 25H. This test should work with any device that DOS recognizes as a disk except CD-ROM or network drives. The test reports the total time (in seconds) required to complete the requests. Why results differ If you run the same PC Bench test on your PC multiple times and use proper testing procedures, you will get essentially the same results every time. (The margin of error in the PC Bench results is about one percent.) Proper testing procedures include defragmenting your PC's disk and rebooting your PC before each test run. If you modify any of your PC's components between test runs or deviate from proper testing procedures, you probably will see a difference in the results. A fragmented disk, for example, can affect the PC Bench results for your PC. Naturally, there are several hardware and software components that can affect your PC's performance. For a list of some of the factors that can influence PC Bench results, see Chapter 6. Publishing PC Bench results The PC Bench License Agreement requires that you include certain information about your PC when you publish any PC Bench results. Thus, if you publish your PC's results, you would need to say something similar to: Ziff-Davis' PC Bench (TM) 8.0 produced a DOSMark (TM) score of X on WXY PC Corp. Model 466 with a 66-MHz Intel (R) 486DX2-66 CPU, 8MB of RAM, 64 KB RAM cache, 200 MB hard disk, IDE controller, and no hardware disk cache. The test PC used XYZ Corp. XYZ Video adapter with a refresh rate of 72 Hz. The test PC was running MS-DOS (R) 5.0, SMARTDRV version X.Y with 2 MB cache, and disk compression utility ABC version 1.0 enabled. All products used in the test were shipping versions available to the general public, and the test and its results were not verified by Ziff-Davis. PC Bench (TM) and DOSMark (TM) are trademarks of Ziff-Davis Publishing Company, L.P. To find out exactly what you must include when you publish PC Bench results, consult the PC Bench License Agreement at the front of this manual and in the README.TXT file. You can also view the license agreement from the Help menu in the PC Bench program screen. Using PC Bench results You can run PC Bench as many times as you wish on your PC and save the results for each test run in the PC Bench database. You can then use PC Bench results to compare different configurations on the same PC or to compare the performance of different PCs. For example, you can: + Run PC Bench multiple times on a single PC with multiple configurations and then compare the results for each test run. + Compare PC Bench results for your PC with the results of other PCs using results published in various Ziff-Davis publications. (For a list of these publications, see Chapter 8.) + Run PC Bench on different machines within your organization to compare their performance. Running PC Bench on different machines may be helpful when your organization has more than one type of PC or has PCs with different configurations. If you have access to PC Bench results for different PCs, you can import those results into the PC Bench database on one PC for comparison. (For more information, see the section "Importing PC Bench results" in Chapter 4.) You may also be able to download PC Bench results from ZiffNet, or you can use the small set of comparison results that come with PC Bench. NOTE: PC Bench includes sets of sample results in its database you can use as comparison systems. To view PC Bench results, use the Compare Results function button from the PC Bench main screen. For more information, see Chapter 4. Using PC Bench with other benchmarks Although PC Bench measures the speed of the processor, memory, disk, and video subsystems using the same operations that DOS applications do, it does not provide you with a measure of the graphics and disk performance you will see should you run Microsoft Windows and Windows-based applications on your PC. To get a picture of your PC's performance under Windows, you can use PC Bench in combination with two other Ziff-Davis benchmarks: WinBench and Winstone 94. + Use WinBench to measure the performance of your PC's graphics and disk subsystems in a Windows environment. ZDBOp based the WinBench tests on application profiling. Thus, these tests reflect the actual operations applications perform. WinBench uses synthetic tests and does not run applications. + Use Winstone 94 to measure the overall performance of your PC as it runs actual Windows-based applications. With Winstone 94 you can also measure the effect that changing individual system components has on your PC's overall performance when running Windows and Windows-based applications. Winstone 94 is the sole Microsoft-certified benchmark for the testing of industry-standard PCs running Windows 3.1-based applications. You can use the form at the back of this manual to request individual copies of WinBench, Winstone 94, and other Ziff-Davis benchmarks from ZDBOp. You can also download Ziff-Davis benchmarks from ZiffNet. What's next Chapter 6 lists some of the factors that can influence PC Bench results and gives tips on how to configure your PC to get the best PC Bench results possible. Chapter 7 explains the concepts behind the development of PC Bench. If you would like to know where to find published PC Bench results, read Chapter 8. If you need to run the PC Bench tests, see Chapter 3. End of Chapter Chapter 6 What Affects PC Bench's Results This chapter lists some of the factors that can influence the PC Bench results for your PC and offers some tips on ways to get the best PC Bench results possible. After you read this chapter, you should understand how to run PC Bench to obtain optimal results. Software and hardware factors that influence PC Bench's results Many factors, both hardware and software, can influence PC Bench results. For example, CPU speed obviously affects results. Perhaps less obvious is that if your PC has a fragmented disk, its Disk Score will often be lower than if the disk is freshly formatted or defragmented. NOTE: There are facets of your computer's architecture you may not normally consider that can affect your PC Bench results. Some PCs in the ZDBOp lab showed improved performance if single SIMMs were replaced with multiple SIMMs. For example, one PC ran tests faster with four 4 MB SIMMs instead of one 16 MB SIMM. The following list gives some of the software and hardware factors that can influence your PC's score. Processor subsystem CPU model and clock speed Internal and external CPU RAM caches Main memory speed Memory/CPU interface speed and width Wait states Power-saving settings Disk subsystem Hard disk drive type Hard disk controller and size of cache (if any) Hard disk driver software Software disk cache and size (if any) Fragmentation level of disk and remaining space on disk Compression Bus interface (such as local bus) Video subsystem Graphics adapter Video interface, such as Local Bus Software Background activity (especially network and TSRs) Version of DOS Hardware bus types ISA bus, EISA bus, Micro Channel (TM) bus Local bus presence and type (for example, VESA or PCI) Tips for improving your PC Bench results The following sections provide tips on ways to get the best and most accurate PC Bench results for your PC. Following proper testing procedures The way to get the best and most accurate PC Bench results for your PC is to do the following each time you run PC Bench: 1. Reformat or defragment your hard disk. 2. Install the operating system (DOS) from scratch. 3. Install PC Bench. 4. Change your AUTOEXE.BAT and CONFIG.SYS files to meet PC Bench's requirements. 5. Reboot your PC. These steps ensure that your PC is "clean" each time you run PC Bench. A clean configuration is a freshly rebooted PC with DOS and a minimum of other software. Because reformatting your hard disk and reinstalling your operating system may not always be convenient or even possible, at a minimum you should always do the following each time you run PC Bench: + Defragment your hard disk. + Free as much RAM as reasonably possible by removing all non-essential items from your PC's AUTOEXEC.BAT and CONFIG.SYS files. (See Chapter 2 for the recommended AUTOEXEC.BAT and CONFIG.SYS files.) + Reboot your PC. Typical PC activities, such as creating, saving, and deleting files, fragment your hard disk. Running the PC Bench Disk Test Suite numerous times will also fragment your hard disk for the same reason. Fragmentation can affect your PC's performance and, thus, its PC Bench results. If you run PC Bench, work on your PC, and then run PC Bench again without defragmenting your hard disk, your score may not be the same. When you defragment your hard disk, you return it to a relatively stable initial state. To defragment your hard disk, be sure to use a utility that both defragments the disk and resets the DOS file pointer. Other ways to improve your PC Bench results You may see an improvement in your PC Bench results if you: + Halt TSRs or network software you normally run and run only PC Bench. If you do have such tasks running, they could affect your PC Bench results as well as cause your PC to run out of memory during some of the PC Bench tests. + Use disk caching. + Enable any processor cache(s). Using PC Bench to best advantage To use PC Bench to best advantage, consider how you normally use your PC. In most cases, the DOSMark will provide an adequate measure of your PC's overall performance. In addition to the DOSMark Test Suite, however, you can also run other individual tests according to your specific needs. For example, if in your day-to-day work you move lots of large files on your PC's disk, you may want to run the 32 MB File Disk Throughput Test. You should run PC Bench on your PC the way you run DOS applications on your PC. For example, if you normally do not run applications on your PC with a disk cache in effect, then run PC Bench without any disk caching. The resulting score will reflect how you normally use your PC. If, on the other hand, you normally use disk caching software, use that same software when you run PC Bench. Do not run PC Bench tests when you have a memory manager other than EMM386 or HIMEM.SYS loaded on your PC. Memory managers can interfere with the proper operation of protected mode tests. ZDBOp tested PC Bench 8.0 with the EMM386 and HIMEM.SYS memory managers that ship with DOS. While PC Bench may run if you have a memory manager loaded on your PC, ZDBOp does not recommend using PC Bench as a tool for testing or comparing the performance of memory managers. What's next Chapter 7 provides a look into the concepts behind PC Bench. If you would like to know more about the development of the PC Bench tests, read this chapter. Chapter 8 tells you where to find published PC Bench results for use in comparing your PC with other PCs. If you need to run the PC Bench tests, see Chapter 3. End of Chapter Chapter 7 The Concepts Behind PC Bench This chapter discusses some of the factors that went into creating the PC Bench tests and making them reflect the way popular applications actually use PC subsystems. After reading this chapter, you should understand the key concepts underlying PC Bench. The design goal The key design goal behind PC Bench is to provide a benchmark that lets you accurately measure the performance of the processor, memory, video, and disk subsystems of a PC. The results should reflect the way real, top-selling DOS applications perform. PC Bench provides such a set of synthetic performance measurement tests. Its results tell you why a PC responds a certain way to certain types of operations. PC Bench does not run actual applications. Instead, it performs many of the typical operations DOS applications perform. The DOSMark gives you a single number representing the overall performance of your PC when running DOS applications. The DOSMark is a weighted harmonic mean of the scores for the Processor, Video, and Disk subsystem tests. PC Bench also returns scores for these three subsystem test suites. PC Bench weights the DOSMark and subsystem scores to make them reflect as accurately as possible the way popular DOS applications use PCs. (For more information on these tests, see Chapter 5.) You can use the DOSMark and subsystem scores to get a general picture of your PC's performance. PC Bench also computes individual test results you can use to see more details about a particular subsystem's performance. The individual test results provide a way for you to explore further your PC's subsystem-level performance. Some of the other factors that went into the design of PC Bench include: + Profiling applications to determine the most commonly used operations. PC Bench follows these application profiles in the mixes it runs. + Using weighted harmonic means to combine groups of results based on the way applications actually use PC subsystems. Profiling applications PC Bench uses tests based on extensive profiling of such popular DOS applications as the following: + Borland's dBASE (R) IV + Borland's Paradox (R) + WordPerfect Corporation's WordPerfect (R) + Lotus (R) 1-2-3 (R) + Microsoft (R) C + Harvard Graphics (R) Profiling is a method of monitoring an application to see what types of operations it performs, how often it performs these operations, and how long it takes each operation to complete. By using the profiling information helps PC Bench's tests accurately reflect what applications really do. By following the profiles of popular DOS application, PC Bench's tests provide a market-centered measure of performance. The weights PC Bench uses to calculate its scores reflect the importance of individual operations in popular applications. Understanding the "market-centered" concept ZDBOp used a market-centered approach to develop the PC Bench tests. The DOS applications ZDBOp profiled were standard, best-selling applications performing typical user tasks. Thus, the application profiles enable PC Bench's tests to perform operations much like those the profiled applications perform. This market-centered approach lets PC Bench produce an accurate, overall picture of your PC's performance when running DOS applications. Using a weighted harmonic mean PC Bench generates the DOSMark, Processor, Video, and Disk scores using weighted harmonic means. Each harmonic mean combines several individual test results to create a single, representative score for a test suite. PC Bench places weights on the different tests to indicate the importance of an operation. Each weight is based on the operations actual applications perform. An operation an application performs frequently has a higher weight than an operation an application performs rarely. For example, integer CPU instructions are far more important than floating point calculations in the vast majority of DOS applications, so PC Bench weights its CPU instruction mixes more heavily than its Floating Point Emulation Test in calculating the overall Processor Score. Understanding the Disk Mix PC Bench 8.0 includes a new testing methodology: the Disk Mix. Today, PCs have both hardware and software disk cache tools available, and PC Bench's disk tests must accurately reflect the way applications work with these tools. The Disk Mix performs a range of disk activities over a large set of files. It uses over 100 files spread across six directories. It works with multiple files concurrently, creates and deletes files, moves data in different sizes, intersperses write and read operations, and generally uses disk subsystems the way applications do. PC Bench's Disk Mix reflects the actual disk usage patterns of the profiled applications. The Disk Mix works with a PC's disk subsystem via the standard DOS Int 21H interface, so you can use the Disk Score as an accurate measurement regardless of whether you are using a software or hardware disk cache. PC Bench's Protected Mode Large Instruction Mix PC Bench 8.0's 16-bit Protected Mode Large Instruction Mix reflects the larger working-set sizes that today's applications use. This mix accurately simulates the way popular applications running under DOS use a PC's processor and memory subsystems. The 16-bit Protected Mode Large Instruction Mix reflects typical application performance in such areas as bus utilization and use of the CPU internal cache, CPU cache interface, external cache, main memory, and the CPU instruction set. Correlating the score to applications PC Bench's DOSMark represents a cumulative effort of profiling today's popular DOS applications and extensive testing of PC Bench and is an extremely accurate measurement of overall PC performance under DOS. During one large test in the development process, the DOSMark yielded a correlation with performance tests involving profiled applications in excess of 99.5 percent on over 70 configurations with and without SMARTDRV enabled. What's next Chapter 8 tells you where to find published PC Bench results for comparing your PC with other PCs. For a list of all the PC Bench tests and a brief description of each, see Appendix A. To run PC Bench, read Chapter 3. To view or compare PC Bench results, see Chapter 4. End of Chapter Chapter 8 How Your PC Compares This chapter tells you where to find published results and other results to use in comparing PC Bench results for your PC with results for other PCs. Checking published results The easiest way to compare your PC to other PCs is to check articles that publish PC Bench results. Many Ziff-Davis publications provide PC reviews that include PC Bench performance scores. You can find published results in "First Looks{SYMBOL 228 \f "Symbol"}" articles, comparative reviews, and other stories in such Ziff-Davis publications as: United States Computer Shopper (R) PC/Computing (R) PC Magazine (R) PC Week (R) Windows (TM) Sources (TM) United Kingdom PC Direct (TM) PC Magazine (TM) France PC Direct (R) PC Expert (R) Germany PC Direkt (TM) PC Professionell (TM) Windows (TM) Activ (TM) You can also find articles about Ziff-Davis benchmarks on Computer Library's Computer Select{SYMBOL 226 \f "Symbol"} CD-ROM, a database of computer-related product information. You may also be able to download PC Bench results from various forums on ZiffNet. Using PCs within your organization If you have access to other PCs within your organization, you can run PC Bench on those PCs, import their results to a single PC, and then use those results from within PC Bench as comparison points. (For more information on how to import PC Bench results, see the section "Importing PC Bench results" in Chapter 4.) You can also merge PC Bench results from different PCs into one file by: 1. Printing your current results to an ASCII text file or CSV format file using the Print option from the File menu in the PC Bench program screen. Once you have your results in ASCII text file format, you can then read the text file into a database or spreadsheet program. 2. Moving the PC Bench results for different PCs to one PC running a database or spreadsheet program. 3. Importing the ASCII text file containing the results into one database or spreadsheet. Once you have all the results in one file, you can view and print those results from your database or spreadsheet program. Publishing PC Bench scores The PC Bench License Agreement requires that you include certain information about your PC with any PC Bench results you publish. To find out exactly what you must include when you publish PC Bench results, consult the license agreement at the front of this manual and in the PC Bench README.TXT file. What's next The appendices that follow provide information you may find useful when running PC Bench. + Appendix A lists all the PC Bench tests and provides a description of each test. + Appendix B lists error messages you may receive while running PC Bench and, where possible, tells you what to do to deal with these errors. This appendix also explains what to do if you must abort PC Bench while it is running its tests. + Appendix C tells you how to get technical support for PC Bench. End of Chapter Appendix A The PC Bench Tests This appendix lists the tests that PC Bench includes and gives you a brief description of each. For information on how to run these tests, see Chapter 2. Processor tests The Processor tests measure the processor subsystem performance by running instruction mixes that reflect the way applications use a PC's processor subsystem. These tests exercise both the protected and real modes of the CPU, the CPU's internal cache (if any), and the CPU/memory interface. The individual processor tests are: + 16-Bit Protected Mode Small Mix and 16-Bit Real Mode Small Mix. The 16-Bit Protected Mode Small Mix and 16-Bit Real Mode Small Mix tests are small instruction mixes that will run within the internal cache of almost all PC CPU chips that contain such a cache. These mixes let PC Bench test CPU speed largely independently from the rest of the PC. Thus, the results give a good, relative measure of basic CPU performance. These tests use a mix of instructions based on popular DOS applications' average instruction use. Both the real and protected mixes reflect the instruction frequencies of DOS applications. The real mode test generally should run faster than the protected mode test due to the protection-checking overhead of protected mode. + 16-Bit Protected Mode Standard Mix and 16-Bit Real Mode Standard Mix. The 16-Bit Protected Mode Standard Mix and 16-Bit Real Mode Standard Mix tests use medium-sized instruction mixes that should fit entirely in most second-level 64 KB or larger caches. These tests use a PC's CPU, CPU cache, and bus the way average DOS applications do. The instruction set mixes use the same basic technology as the 16-bit small mix. These tests indicate how effectively the processor interacts with the rest of the processor subsystem and how efficiently a system manufacturer has interfaced the CPU to the cache components. In addition, these tests serve as a good way to measure the performance effect of the presence or absence of an external cache subsystem. + 16-Bit Protected Mode Large Mix. The core of PC Bench's Processor Score is the 16-bit Protected Mode Large Instruction Mix. This new large instruction mix reflects the larger working-set sizes that today's applications use. This mix accurately simulates the way leading DOS applications use a PC's processor and memory subsystems. The 16-bit Protected Mode Large Instruction Mix uses a PC's bus, CPU internal cache, CPU cache interface, external cache, main memory, and instruction set in the same way as leading DOS applications. + String Sort and Move. The String Sort and Move test times how long it takes the processor subsystem to execute a bubble sort on 200 random strings containing 16 characters each. This test can be useful in gauging the speed of systems with a RAM cache, because sorting requires reading and writing some of the same data repeatedly. + Prime Number Sieve. The Prime Number Sieve test measures how long it takes to execute a routine that finds the prime numbers between 0 and 8190. This is a common processor test that PC Bench includes for that reason. + Floating Point Emulation. The Floating Point Emulation test sets up a floating point emulation program in RAM and then performs a mix of floating-point calculations. The processes this test performs are identical to those in the Math Co-processor test. + Math Co-processor. The Math Co-processor test exercises the math co-processor using the same floating-point calculations the Floating Point Emulation test uses. You can use the test to analyze the speed differences of floating point co-processors in different systems. Additionally, you can use the test to gauge the processing speed increase a co-processor provides by comparing a PC's Floating Point Emulation results with its Math Co-processor results. EGA/VGA text mode tests The EGA/VGA text mode tests include both BIOS tests and direct-screen tests. NOTE: Some BIOS manufacturers purposely slow scrolled writes so you will have time to read the scrolling text. If the video BIOS is loaded and copied from ROM into RAM (for example, if it is shadowed), the throughput for these tests will be higher. + Unscrolled BIOS Write. The Unscrolled BIOS Write test times the speed of writing data to the screen using interrupt 10h BIOS calls. + Scrolled BIOS Write. The Scrolled BIOS Write test times the speed of writing data to the screen using interrupt 10h BIOS calls. Once the test fills the screen, it scrolls through an entire screen by adding one new line at a time. This test writes the same amount of data to the screen as the Unscrolled BIOS Write test. + Direct Screen Write (8-bit) and Direct Screen Write (16-bit). The Direct Screen Write tests write data to the video adapter screen memory using the MOVSB (8-bit) and MOVSW(16-bit) instructions. In general, one can reasonably expect the 16-bit test to yield about twice as much throughput as the 8-bit test. EGA/VGA graphics mode tests + Write Mode 0 Fill (8-bit), Write Mode 0 Fill (16-bit), and Write Mode 0 Fill (32-bit). The Write Mode 0 Fill tests fill the video adapter's VGA memory with varying colors using the STOSB (8-bit), STOSW (16-bit), and STOSD (32-bit) instructions with EGA/VGA write mode 0 and all bit planes enabled. (The STOSD instruction is only applicable on PCs with at least an 80386 processor.) + Mem to Scr BitBlt (16-bit) and Mem to Scr BitBlt (32-bit). The Memory to Screen BitBlt tests create screen segments and then save these screen images to system memory one video plane at a time. The timed portion of the test copies the screen segments in system memory back to screen memory using the MOVSW (16-bit) and MOVSD (32-bit for systems with at least an 80386 processor) instructions, one video plane at a time. Very fast video cards will show the text in a "marquee" effect on the screen. + Scr to Mem BitBlt (16-bit) and Scr to Mem BitBlt (32-bit). The Screen to Memory BitBlt tests are similar to the Memory to Screen BitBlt tests except that the timed portion of the test measures the throughput of copying information in screen memory to system memory using the MOVSW (16-bit) and MOVSD (32-bit for systems with at least an 80386 processor) instructions one video plane at a time. Because these tests update system memory, there is no visible activity during this test. + Scr to Scr BitBlt (16-bit) and Scr to Scr BitBlt (32-bit). The Screen to Screen BitBlt tests divide screen memory into four quadrants and fill the first three quadrants with uniform fill patterns of different colors. The test repeatedly moves screen data from the first three screen quadrants to the fourth one using the MOVSW (16-bit) and MOVSD (32-bit for systems with at least an 80386 processor) instructions with write mode 1. Disk tests The Disk Mix performs a broad range of disk activities over a large set of files. It uses over 100 files spread across six directories. It works with multiple files concurrently, creates and deletes files, moves data in different sizes, intersperses write and read operations, and generally uses the disk subsystem the way applications do. The Disk Mix works with a PC's disk subsystem via the standard DOS Int 21H interface, so you can use the Disk Score as an accurate measurement regardless of whether you are using a software or hardware disk cache. PC Bench's Disk Mix reflects the disk usage of a set of profiled, leading DOS applications. Thus, the Disk Mix works with your PC's disk subsystem the way actual applications do. Disk throughput tests The disk throughput tests show the effect different file and block size combinations have on raw disk performance. The disk tests use 32 MB, 16 MB, 8 MB, 4 MB, 2 MB, 1 MB, and 256 KB file sizes to test disk throughput. For each file size, these tests move blocks of many different sizes: 200 bytes, 512 bytes, 2 KB, and 4 KB. For each block size, the tests perform sequential reads, sequential writes, random reads, and random writes. All together, there are 105 disk throughput tests. These tests are most useful for gauging three main aspects of disk performance: + The variation in DOS overhead caused by various block sizes. + The overhead due to the inefficiency of crossing sector boundaries. The 200-byte block size causes disks to cross their 512-byte sector boundaries during reads and writes. + The differences between in-cache and out-of-cache throughput. You can get an idea of these differences by comparing the results for a file size that fits in the cache against the results for a file size that exceeds the cache's size. If you want to measure your PC's disk subsystem performance when manipulating files with specific sizes in specific block sizes, use the Disk Throughput Tests. Note, however, that due to the low-level approach of these tests, they may not give a realistic reflection of disk performance when a cache is active. The following is a list of all the Disk Throughput Tests: 32 M File 32M-200 Byte Records 32M/200 Sequential Read 32M/200 Sequential Write 32M/200 Random Read 32M/200 Random Write 32M-512 Byte Records 32M/512 Sequential Read 32M/512 Sequential Write 32M/512 Random Read 32M/512 Random Write 32M-2K Byte Records 32M/2K Sequential Read 32M/2K Sequential Write 32M/2K Random Read 32M/2K Random Write 32M-4K Byte Records 32M/4K Sequential Read 32M/4K Sequential Write 32M/4K Random Read 32M/4K Random Write 16 M File 16M-200 Byte Records 16M/200 Sequential Read 16M/200 Sequential Write 16M/200 Random Read 16M/200 Random Write 16M-512 Byte Records 16M/512 Sequential Read 16M/512 Sequential Write 16M/512 Random Read 16M/512 Random Write 16M-2K Byte Records 16M/2K Sequential Read 16M/2K Sequential Write 16M/2K Random Read 16M/2K Random Write 16M-4K Byte Records 16M/4K Sequential Read 16M/4K Sequential Write 16M/4K Random Read 16M/4K Random Write 8 M File 8M-200 Byte Records 8M/200 Sequential Read 8M/200 Sequential Write 8M/200 Random Read 8M/200 Random Write 8M-512 Byte Records 8M/512 Sequential Read 8M/512 Sequential Write 8M/512 Random Read 8M/512 Random Write 8M-2K Byte Records 8M/2K Sequential Read 8M/2K Sequential Write 8M/2K Random Read 8M/2K Random Write 8M-4K Byte Records 8M/4K Sequential Read 8M/4K Sequential Write 8M/4K Random Read 8M/4K Random Write 4 M File 4M-200 Byte Records 4M/200 Sequential Read 4M/200 Sequential Write 4M/200 Random Read 4M/200 Random Write 4M-512 Byte Records 4M/512 Sequential Read 4M/512 Sequential Write 4M/512 Random Read 4M/512 Random Write 4M-2K Byte Records 4M/2K Sequential Read 4M/2K Sequential Write 4M/2K Random Read 4M/2K Random Write 4M-4K Byte Records 4M/4K Sequential Read 4M/4K Sequential Write 4M/4K Random Read 4M/4K Random Write 2 M File 2M-200 Byte Records 2M/200 Sequential Read 2M/200 Sequential Write 2M/200 Random Read 2M/200 Random Write 2M-512 Byte Records 2M/512 Sequential Read 2M/512 Sequential Write 2M/512 Random Read 2M/512 Random Write 2M-2K Byte Records 2M/2K Sequential Read 2M/2K Sequential Write 2M/2K Random Read 2M/2K Random Write 2M-4K Byte Records 2M/4K Sequential Read 2M/4K Sequential Write 2M/4K Random Read 2M/4K Random Write 1 M File 1M-200 Byte Records 1M/200 Sequential Read 1M/200 Sequential Write 1M/200 Random Read 1M/200 Random Write 1M-512 Byte Records 1M/512 Sequential Read 1M/512 Sequential Write 1M/512 Random Read 1M/512 Random Write 1M-2K Byte Records 1M/2K Sequential Read 1M/2K Sequential Write 1M/2K Random Read 1M/2K Random Write 1M-4K Byte Records 1M/4K Sequential Read 1M/4K Sequential Write 1M/4K Random Read 1M/4K Random Write 256 K File 256K-200 Byte Records 256K/200 Sequential Read 256K/200 Sequential Write 256K/200 Random Read 256K/200 Random Write 256K-512 Byte Records 256K/512 Sequential Read 256K/512 Sequential Write 256K/512 Random Read 256K/512 Random Write 256K-2K Byte Records 256K/2K Sequential Read 256K/2K Sequential Write 256K/2K Random Read 256K/2K Random Write 256K-4K Byte Records 256K/4K Sequential Read 256K/4K Sequential Write 256K/4K Random Read 256K/4K Random Write DOS disk access The DOS disk access group of tests includes the BIOS Disk Seek (sequential and random) and the DOS Disk Access tests. NOTE: If a seek error occurs in either of the tests, PC Bench will display a message for a few seconds, choose a replacement sector, and then continue the test. Such a failure is not necessarily an indication of a bad disk drive. These tests access sectors that may have been legitimately locked out during the formatting of the drive. If you suspect a problem with your drive, use a diagnostic program to test the drive. + BIOS Disk Seek - Sequential and BIOS Disk Seek - Random. The BIOS Disk Seek tests (sequential and random) measure mechanical track-to-track disk drive access times. The tests use the standard BIOS interrupt 13h interface. You cannot use this test to check floppy diskettes or mass storage media that do not have an Interrupt 13h interface. The tests uses 1,000 sequential seeks (alternating between cylinders 0 and 11) and 1,000 random seeks. Caching disk controllers and certain SCSI adapters that do not force the disk mechanism to move for disk seeks may yield unrealistically fast times. + DOS Disk Access. The DOS Disk Access Test measures the time necessary to perform 1,000 read requests at random locations on the disk using the DOS Int 25H interface. This test should generally work with any device DOS recognizes as a disk except CD-ROM and network drives. The test reports the total time (in seconds) required to complete the requests. Memory timing tests The memory timing tests gauge a PC's speed at accessing various types of memory. A PC's CPU speed and the architecture of its memory, memory caches (if present), and CPU can affect the results of these tests. The memory timing tests include a set of Extended Memory tests. These tests enter protected mode to access memory above 1 MB. As a result, these tests will generally not execute properly or fail to execute at all if an extended memory driver (such as EMS, DPMI, or VCPI) is already installed on your PC. The memory tests basically use the full range of extended memory available. The write tests avoid any area already locked by an XMS (HIMEM) driver, RAM disk, or other DOS Int 15H-compatible software. There are six Extended Memory tests, a read test and a write test for each data width. Those tests are: + 8-bit Extended Memory Read and Write + 16-bit Extended Memory Read and Write + 32-bit Extended Memory Read and Write The memory timing tests also include a set of Graphics Memory tests. These tests detect the type of VGA graphics adapter present and use the appropriate buffer addresses to read and write blocks of data directly to the graphics adapter's VGA memory. During memory writes, PC Bench will overwrite the screen's contents and then restore them at the end of the test. There are six Graphics Memory tests, a read and a write for each data width. These tests are: + Graphics Memory 8-bit Read and Write + Graphics Memory 16-bit Read and Write + Graphics Memory 32-bit Read and Write The memory timing tests also include BIOS Memory tests. These tests read only from the BIOS. These tests generally shows significant performance differences when your PC's BIOS ROM is shadowed or cached, a common configuration choice on many systems. There are three BIOS memory tests: + BIOS Memory 8-bit Read + BIOS Memory 16-bit Read + BIOS Memory 32-bit Read The memory timing tests also include a set of Text Memory tests. These tests detect the type of VGA display adapter and use the appropriate screen buffer addresses to write directly to the display adapter's memory. There are four Text Memory tests, a read and a write test for each data width. These tests are: + Text Memory 8-bit Read and Write + Text Memory 16-bit Read and Write Finally, the memory timing tests also include a set of Conventional Memory tests. These tests access memory in the 640 KB area that DOS manages. There are four Conventional Memory tests, a read and a write test for each data width. These tests are: + Conventional Memory 8-bit Read and Write + Conventional Memory 16-bit Read and Write Tests included as utilities PC Bench includes the following tests you can run as utilities. You execute these tests from the directory where you installed PC Bench. Battery Rundown Test The Battery Rundown Test exercises a portable computer's battery system in a worst-case scenario. You should start the test after fully charging the computer's battery according to the manufacturer's specifications and disabling all power-conservation features. The test types 10 screens full of information to the computer's display and then writes a 15-KB file to disk with a time stamp of elapsed time. PC Bench repeats this operation until the battery runs out of power. Time the Timer Test The Time the Timer Test calculates elapsed time using DOS calls. Some systems may not report a correct time when running a version of DOS that is not properly configured for the system. In this case, the results for the Time the Timer Test may be skewed. This test lets you measure the accuracy of the system's timer by comparing it to some external source (such as a stopwatch). VGA Compatibility Test The VGA Compatibility Test performs several VGA register-level functions that exercise the different subsystems (including the VGA controller chip, video memory, and RAMDAC) of a VGA or VGA-compatible display subsystem. Because the test is an older test designed with the original VGA in IBM PS/2 systems in mind, most boards do not pass every test but can still run application software without any problems. Thus, a failure of a few of the tests does not usually indicate serious compatibility problems for most software. EGA/VGA Monitor Quality Test The EGA/VGA Monitor Quality Test displays images in various modes from black-and-white text to VGA color graphics. You can use these tests to determine what display modes a monitor or display adapter combination can support. Also, you can use these tests to evaluate monochrome portable computer screens to see how well they map colors to gray scales. End of Appendix Appendix B Troubleshooting Normally, PC Bench runs without problems. Occasionally, however, you may encounter a problem when you run PC Bench. This appendix gives you some basic information about what to do if you have problems with PC Bench. It also lists messages you may receive when running PC Bench and tells you what to do when you receive those messages. If you have a problem and cannot resolve it after reading this appendix, write down any message (including any directory or file names) that appears on your PC's screen, note what PC Bench was doing at the time you received the message (if you know), and then contact ZDBOp's PC Bench technical support as described in Appendix C of this manual. What to do if you have problems PC Bench may encounter a fatal error if it runs out of memory, or it may stop running (hang) during its tests on some PCs due to compatibility issues. In most cases, PC Bench will exit as gracefully as it can if there is a serious problem. If you have problems, you should follow a few basic steps before you contact technical support: + Verify that your PC meets the hardware and software requirements of PC Bench. Also, check to be sure that you have set up your PC correctly. (For more information on both areas, see Chapter 2.) + If PC Bench issues the error "Unable to allocate enough memory," you will need to free more conventional memory by removing TSRs or device drivers or by loading DOS high with HIMEM.SYS. (For more information, see the section "Setting up your PC to run PC Bench" in Chapter 2.) + If you get a PC Bench database error or a message suggesting you restore your database files, you should exit PC Bench. You should then restore PC Bench's database files by typing the following command from the directory where you installed PC Bench: RESTDB This command restores the backup copy of your database files and may fix any corruption that may have occurred. The backup files are usually those that were present the last time you started PC Bench. NOTE: Even though you restore PC Bench's database file, you will lose any data that you added to the database since the last backup. + If PC Bench seems to hang during a test or some operation and is not responding to any input, you may need to reboot your PC. To reboot, you can: * Press Ctrl-Alt-Del * Press your PC's reset button if it has one * Turn the PC's power off NOTE: Several PC Bench tests may take a while to run. One way to tell if PC Bench is still running its tests is to check to see if your PC's disk light is active. If it is, then PC Bench is still running its test. Do not abort PC Bench unless you are very sure there is a problem. Restart PC Bench and see if the problem persists. If so, you can contact PC Bench technical support at ZDBOp. Running PC Bench with memory managers PC Bench 8.0 works with the EMM386 and HIMEM.SYS memory managers that ship with DOS. Although PC Bench may run if you have a different memory manager loaded on your PC, ZDBOp does not guarantee your results. Also, ZDBOp does not recommend using PC Bench as a tool for testing or comparing the performance of memory managers. NOTE: If you are running the Above DISK (R) memory manager (either by running PC Bench on a 286 CPU or running Above DISK on a 386 or higher CPU), PC Bench may hang or may not execute its tests correctly. PC Bench requires a 386 or higher CPU to successfully run its tests. Running PC Bench in batch mode If you have problems running PC Bench from a DOS batch file, try running it with the /M option. This option prevents PC Bench from logging your PC's configuration information. You should not stop PC Bench from logging your PC's configuration information unless you are having problems running PC Bench. You should also note that using the /P option when running PC Bench in batch made can cause problems on future PCs with new processors or on DOS emulators . This option prevents PC Bench from automatically determining what type of processor is in your PC. PC Bench error messages This section lists most of the error messages you might receive while running PC Bench and explains what you should do if you get one of these messages. You may need to free up XX bytes on your system before PC Bench can successfully run all its tests at once. PC Bench issues this warning message if you try to run PC Bench with less than 540 KB of free conventional memory. You need a minimum of 540 KB of free conventional memory to run all the PC Bench tests at once. When you get this error, you can continue if you do not plan to run all the tests at once, or you can exit PC Bench, free conventional memory by removing TSRs, drivers, etc., and then restart PC Bench. Remember, if you change your CONFIG.SYS or AUTOEXEC.BAT files, you must reboot your PC before you restart PC Bench. PC Bench supports only VGA (640x480) resolution or greater. PC Bench may run on your display, but text in the PC Bench screens will be truncated. ZDBOp designed PC Bench to run on VGA systems. If you run PC Bench on a non-VGA system, it will try to continue, but the display and positioning of its windows and buttons will not be correct. Generally, you should not run PC Bench without VGA support. PC Bench does not support processors less than 80386. If you continue, you will not be able to run all the PC Bench tests. You get this error message if you attempt to run PC Bench on a PC with a CPU less than an 80386 (or 80386-compatible). PC Bench 8.0 requires a minimum of a 386 (or compatible) CPU. PC Bench can only produce a DOSMark score and the Protected Mode Large Mix Processor results for PCs with more than 1 Mb of Extended Memory. You will receive this error message if PC Bench detects that there is not enough extended memory on your PC to run the Protected Mode Large Mix. This test requires just over 1 MB of extended memory. PC Bench is only able to calculate the DOSMark score on PCs with a total of at least 2 MB of RAM. Unable to allocate enough memory. This message indicates that PC Bench has completely run out of conventional memory and cannot continue its current operation. PC Bench will quit after it displays this message. You can exit PC Bench and free up conventional memory by removing TSRs, drivers, etc. Even if you do not change your CONFIG.SYS or AUTOEXEC.BAT files, you should reboot your PC before you restart PC Bench to ensure the tests run correctly. Note that memory fragmentation can also cause low memory conditions. Try repeating the operation that caused this error message after you restart PC Bench before doing any other operations. Insufficient memory to run test. Not enough conventional memory is available to provide the currently running test with the buffers it needs to run. The test will cancel, and if there is another test still to run, that test will begin. That test, however, may also cause the same error. This error message is not fatal and does not keep PC Bench from continuing. Insufficient memory to load test group. This message is similar to the one above, but it indicates that there was not enough conventional memory to hold the test code itself. In this case, PC Bench skips the whole test group (for example, processor tests) you are trying to run. If you get this error message, you need to free more conventional memory to run the PC Bench tests. This machine's settings are not unique. Please change its Machine ID or Variants. You will get this error when you are using the Machine Information dialog box if you set the Machine ID and Variants to the same settings as an existing set of results. You should change one or more of these fields. The Machine ID and Variants you specified are not unique. You will get this error in batch mode when you specify the DOS environment variables Machine ID or Variants that are not unique. You should re-enter these settings. The Machine ID and Variants are not unique for the new machine. You will receive this message when PC Bench tries to automatically create variants for your PC but none of them are unique. This might happen if you name a series of PCs with variants that just happen to coincide with the current date and time. If you wish to use the variants that conflict with the current date and time, you should try again after some time has passed, or you can change the Variants using batch mode. Invalid file name. You receive this fatal PC Bench error if you use the /Ffile option and file is not a valid file name. You should verify that the file name you supplied is correct. Could not find the environment variables: PIN VARIANT1 VARIANT2 VARIANT3 VARIANT4 VARIANT5 You get this message when you use the /B option without setting the minimum necessary DOS environment variables. To correct the problem, make sure you set the appropriate variables when you execute PC Bench from the DOS command line or from a batch file. You cannot execute this program directly--Use BENCH. You will receive this message if you try to run a PC Bench test module from the DOS command line. You cannot run these modules individually; they will run only when PC Bench starts them. Internal Error--Please contact ZDBOp. If you receive this fatal PC Bench error, it means PC Bench has encountered an internal error. Write down what PC Bench was doing at the time you received the error, if possible. Then contact ZDBOp as described in Appendix C. The PC Bench licensing information is not set up correctly. You should re-license PC Bench. Before you can run PC Bench, you must first license it. You will receive one of these error messages if PC Bench cannot find the license file it creates when you license and register the program. To license PC Bench, use the /L option when you start PC Bench. The licensing procedure failed. You receive this error message when the PC Bench licensing procedure aborts (possibly due to a file error). Try licensing PC Bench again with the /L option. PC Bench cannot find a resource file. Check to be sure you are in the PC Bench directory. You may receive this error message if you attempt to run PC Bench and you are not in the directory where you installed it. Check to make sure you are in the correct directory. Open Error: FILENAME.DBF You may receive this message when PC Bench encounters an error trying to open a database file. FILENAME is the name of the database file. Your database of results may be corrupt. You should restore your PC Bench database files and restart PC Bench. If you still get this error, contact ZDBOp. General database errors When PC Bench encounters general database errors, it will automatically issue a warning message and quit. You should reboot your PC and restore PC Bench's database files. NOTE: When you restore PC Bench's database files, you will lose any data you have added to the database since PC Bench backed up its files. General database errors are of the format: Unique Key error (Error XX) Restore your database files Where XX is an internal error number. You can record this number and contact ZDBOp. End of Appendix Appendix C Technical Support This appendix tells you how to get copies of PC Bench and other Ziff-Davis benchmarks and tells you how to order printed copies of this manual. This appendix also explains the different ways you can contact the Ziff-Davis Benchmark Operation (ZDBOp). Requesting a benchmark PC Bench is available free of charge on 3.5" diskettes. You can get copies of PC Bench and other Ziff-Davis benchmarks free of charge by contacting ZDBOp or by downloading the benchmark you want from ZiffNet. NOTE: The Winstone 94 CD-ROM includes both PC Bench and WinBench. Thus if you get the Winstone 94 CD-ROM, you actually get all three benchmarks. To request a benchmark from ZDBOp, use the form that comes after this appendix. You can fax the form to the ZDBOp dedicated benchmark request fax number (919-380-2879), or you can mail the form to ZDBOp at the address listed at the top of the form. The benchmark will arrive via third-class US mail in four to six weeks. If you want to receive a benchmark sooner and you have a Federal Express account, include your Federal Express account number and shipping instructions on the fax form. The benchmark will arrive via Federal Express and ZDBOp will charge the cost of sending the benchmark to your Federal Express account. Ordering a printed manual You can also order a perfect-bound, printed copy of this manual for a nominal fee. To order a copy of this manual or other Ziff-Davis benchmarks' manuals, use the Ziff-Davis Benchmarks' Manual Order Form included at the back of this manual. Contacting ZDBOp If you have comments or questions about PC Bench, please contact ZDBOp. You can contact ZDBOp in several ways: + If you have a modem and communications software, you can reach ZDBOp via ZiffNet, the Ziff-Davis on-line service. One way to access ZiffNet is by using CompuServe (R). ZiffNet's ZDBOp forum (GO ZDBENCH) is a way for you to communicate your technical questions and comments to Ziff-Davis. (Access to CompuServe is available for a fee.) + You can fax questions and comments directly to PC Bench Technical Support at the ZDBOp fax number (919-380-2879). + You can mail questions and comments to ZDBOp at the following address: Ziff-Davis Benchmark Operation One Copley Parkway, Suite 510 Morrisville, NC 27560 Attn: PC Bench Technical Support End of Appendix Benchmark Request Form FAX OR MAIL THIS FORM TO: Ziff-Davis Benchmark Operation One Copley Parkway, Suite 510 Morrisville, NC 27560 fax: (919) 380-2879 WHO I AM: Name: ____________________________________________________________ Company: _________________________________________________________ Address: __________________________________________________________ City: _________________________ State: _____________ Zip: ______________ Country: __________________________________________________________ Fax: ___________________________ Phone: _____________________________ We answer requests in the order we receive them. We ship all benchmarks via third-class US mail unless you supply a Federal Express account number. Please allow four to six weeks for delivery via regular mail. Shipping information if you choose to use Federal Express: Your Federal Express account number: ______________________________ Check one: ___ priority overnight ___standard overnight WHAT I WANT: ___ PC Bench (TM) 8.0 ___ NetBench (TM) 2.1 ___ WinBench (R) 4.0 ___ ServerBench (TM) 1.0 SCO (R) UNIX (R)* ___ Winstone (TM) 94 Version 1.0 (on CD-ROM) ___ MacBench (TM) 1.0 (All benchmarks EXCEPT Winstone 94 are available only on 3.5" diskettes.) * Other versions to follow Ziff-Davis Benchmarks' Manual Order Form To Order: Fax this form to (919) 380-2879 or mail to: Ziff-Davis Benchmark Operation One Copley Parkway, Suite 510 Morrisville, NC 27560 Attn.: Manual Orders Title Price Qty. Total Understanding and Using WinBench (R) 4.0 $10.00 ____ _______ Understanding and Using Winstone (TM) 94 Version 1.0 $10.00 ____ _______ Understanding and Using PC Bench (TM) 8.0 $10.00 ____ _______ Understanding and Using MacBench (TM) 1.0 $10.00 ____ _______ Understanding and Using NetBench (TM) 2.1 $10.00 ____ _______ Understanding and Using ServerBench (TM) 1.0 for SCO (R) UNIX (R) $10.00 ____ _______ Shipping and handling included. TOTAL ____ _______ Ship to: (please print) Name: ________________________________________________________________ Company: _____________________________________________________________ Address:______________________________________________________________ City / State / Zip: ________________________________________________________ Daytime Phone: _________________________________________________________ Payment form: __Check to Ziff-Davis Publishing Co. __Visa __MasterCard __American Express Credit Card #: _____________________________________ Exp. Date: ___________ Signature: ______________________________________________________________ We do not accept Purchase Orders or cash payments. For fastest shipment supply your Federal Express or UPS account number: Your account number:_____________________________________________________ Check one: __ Priority overnight __Standard overnight Acknowledgments The Ziff-Davis Benchmark Operation (ZDBOp) developed version 8.0 of PC Bench. Numerous people worked together to create this product. Members of the primary PC Bench development team are: Debbie Willmschen PC Bench Technical Documentation Thomas E. Siering PC Bench Development Leader Michael Agerbak PC Bench Developer David J. Arendash PC Bench Developer Bradley G. Epranian PC Bench Developer In addition, the PC Bench benchmark would not be possible without the use of software from Phar Lap Software, Inc. Other contributors to the contents of the PC Bench diskette include: Richard Butner Gabriel DeBacker Israel Ehrisman Dan Ghidali Homi Kapadia Kasey Lee The ZDBOp support staff who helped make PC Bench possible are: Betsy Bates Jennie Faries Rae Fernandez Gina Massel-Castater Olivia Ongpin Maria Rowan Many people in different parts of Ziff contributed to the design, testing, and production of PC Bench, including: Tim Baltrusch Adam Baum Jessica Beyer Michael Blakeley Brain Booher Dean Bushik Stephen Chan Michael Edelhart Robert Ferguson Jim Galley Edward O. Griffin Edward Henning Alex Ho Bert Jensen Bob Kane Amy Leung William Machrone Cynthia Meade Jim Nell Huy Nguyen Ron Panus Harold Poskanzer Dave Salvator Kern Schireson Nick Stam Tim Stefanini Jim Wolf Joe Virzi