home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
The World of Computer Software
/
World_Of_Computer_Software-02-386-Vol-2of3.iso
/
b
/
bnch7.zip
/
HELPME
< prev
next >
Wrap
Text File
|
1992-11-05
|
40KB
|
924 lines
README FILE
PC Labs
PC Bench
Version 7.0
Release date: November 1992
PC Bench is a registered Trademark of Ziff-Davis Publishing
Company.
Copyright (c) 1992 Ziff-Davis Publishing Company, a division
of Ziff Communications Company. All rights reserved.
The software described in this document is furnished under a
license agreement. The software may only be used or copied
in accordance with the terms of the agreement. It is
against the law to copy the software on any medium except as
specifically allowed in the license agreement. No part of
this document may be reproduced or transmitted in any form
or by any means, for any purpose other than the licensee's
personal use without the express written permission of Ziff-
Davis Publishing Company.
PRODUCT SUPPORT
You can FAX your questions to the Ziff-Davis Benchmark group
at (212) 503-5518. Or you can also mail your questions to:
PC Labs, One Park Avenue, New York, NY 10016; (212) 503-5255.
Only FAX and mail support are available at this time.
SYSTEM REQUIREMENTS
To run all tests:
MSDOS version 4 or higher
500kbyte free memory
40MB free disk space
VGA color is recommended
Less than 40MB disk space is acceptable, but
prohibits calculation of the "DOSMark"(R).
MSDOS Versions 3.x can be used but may also
prevent accurate calculation of the
"DOSMark."
The most accurate results are obtained by running PC Bench
with a newly formatted disk that contains only DOS and PC
Bench. This will ensure that consistent results are
obtained when testing different systems.
The CONFIG.SYS file should have the statement FILES=20 only.
A mouse driver may be present. No TSRs or memory managers
should be loaded in high memory. The AUTOEXEC.BAT file
should also be empty; however, a path reference and prompt
command may be present.
PC Bench may usually be run under other configurations, but
we do not recommended this for proper, valid comparison
purposes. If it is not feasible to reformat your hard disk,
the bench program will still run and produce results. The
results may not be quite as optimal as would have otherwise
been obtained, particularly with the disk tests.
If an insufficient memory message occurs, free up memory by
removing TSRs, device drivers, and memory managers from the
AUTOEXEC.BAT or CONFIG.SYS files. Generally, some of the
tests will run even with this warning message, although the
results may not be optimal.
The program has not been validated under operating systems
other than MSDOS.
OVERVIEW
PC Bench is Ziff-Davis Publishing Company's standard
benchmark tool for evaluating the relative performance of
Intel-based x86 computers. It is comprised of a set of test
suites that provides component-level measurements of CPU,
memory, disk, and video subsystems.
PC Bench is comprised of tests that are based on extensive
profiling of the functions users perform using mainstream
DOS applications as well as tests that exercise specific
subsystem components to reveal additional useful
information.
Menu Options are available to batch run the entire suite of
processor, memory, video, and disk tests or to customize a
batch run of one or more selected tests.
Additional menu options called the "Processor Harmonic,"
"Memory Harmonic," "Video Harmonic", and "Disk Harmonic"
each run selected tests from their respective subsystem test
group that reflect the application execution profiles. The
results are weighted accordingly, and a harmonic mean value
is computed.
Also available is a menu option called the "DOSMark" that
runs a select subset of tests from all the subsystem test
groups. The results are weighted, again, all according to
execution profile information, and a single overall system
score is computed and displayed.
Test results are displayed to the right of the associated
test. Options are available to compare results from
successive test runs on-line. Also available are options to
print and export results.
WHAT HAS CHANGED FROM THE PREVIOUS RELEASE
Nearly all of the tests have changed in this release of PC
Bench. There is no backwards compatibility to previous
benchmark releases due to the change of scope in the testing
philosophy.
All of the low-level tests have been weighted through
harmonic means to generate the "DOSMark" and the four sub-
system harmonics. The weights are derived from the way in
which typical applications use various components of a
system.
FOR MORE INFORMATION
For more information about this benchmark see the related
story entitled "A Look Inside PC Magazine Labs New
Performance Benchmarks" in the Labs Notes section of "PC
Magazine," Vol.11, No. 21.
INSTALLATION
See the file README1.TXT on the PC Bench diskette. This file
contains the procedure to convert compressed files from the
installation diskette, decompress the files into DOS format,
run the licensing program, and execute the Bench program.
GETTING STARTED
Once PC Bench is installed properly and licensed, the
command BENCH from the DOS prompt executes the program and
the user ID window is displayed, followed by the Machine
Information dialog box. (For more information see Machine
Information.)
[NOTE: If a bytes available warning is displayed, it likely
means you do not have sufficient conventional RAM available
to run some of the memory and protected mode tests. The
solution is to free up memory by unloading TSRs, device
drivers, and memory managers.]
None of the fields in the Machine Information dialog box are
mandatory, although the Machine ID and Variant boxes should
not be blank, so to get started you may just want to type
any alpha-numeric Machine ID, Machine Name (brand name and
model), Machine Description (comment), and Message
(comment). (For navigation information see Navigation.)
Choose the Use button. The machine runs MScope
("Microscope") to log system information and automatically
opens the "DOSMark" dialog box.
At this point you have some choices regarding the test to
run. (For more information see "DOSMark" under View Menu.)
Either choose "DOSMark," "Processor Harmonic," "Memory
Harmonic," "Video Harmonic," or "Disk Harmonic" followed by
Execute to run the test,
-or-
To run the entire suite or a customized subset of tests,
choose Close, and then open the Batch Editor under the File
menu. (See Batch Editor for more information.) Choose
Select All to run the complete suite of tests or choose
individual tests to create a customized suite. (When a test
is selected a check mark appears to the left of the test
item.) Choose Execute to begin running the test(s). The
complete suite normally takes a little over an hour to run.
After the tests are finished running, the results are
displayed to the right of each test. Test results can also
be viewed by choosing Results under the View menu. (See View
menu for more information.)
Select Print under the File menu to print or export results.
The Machine Browser dialog box opens (note the check next to
the current run line). Choose Select and the Print Results
dialog box opens. To print results, type the printer device
name in the Filename field (example: LPT1). To export
results, create a filename (8 character limit) then choose
CSV (Comma Separated Values) or Txt (text file). The file
will be written to the directory containing the PC Bench
files. The CSV file can be imported to any spreadsheet
program that accepts CSV format.
NAVIGATION
Mouse - point & click.....= select a menu
Mouse - point & click ....= activate a selection
ALT + letter .............= select a menu or choice by
letter
ESC ......................= de-select a sub menu
Tab ......................= forward to field
Shift + Tab ..............= backward to field
Return or Enter ..........= commit choice or select test
Arrow keys left, right ...= highlight a menu item
Arrow keys up, down.......= highlight a sub menu item
Navigation is accomplished by using a mouse (point and click
on the selection). If a mouse is not available or not
configured properly, alternative keys may be used.
To navigate using the keyboard, combine the ALT key with the
underlined letter within a menu, list, or button. A gray
highlight indicates the current choice. Use the ESC key to
deselect a menu. The tab (or shift/tab to back tab) or
cursor arrow keys are used to navigate within the dialog
boxes. The Enter key commits the action unless otherwise
specified. Page Up/Page Down keys allow fast scrolling when
a scrollable box is selected.
MACHINE INFORMATION DIALOG BOX
This display contains fields for defining the type of
system to be tested, the iteration number of the test, and
other pertinent information. These fields are not
mandatory, per se, except that no field should be blank.
The Machine ID and Variant fields are the most critical
items in the Machine Information Box because they are used
to reference data being captured from the testing process.
The Variant fields are used to record configuration
differences for the system being tested, as indicated in
the Machine ID field. The Variant fields are not required
to record new test iterations of the same configuration if
the results are committed to the database for each retest.
The variant number need not be changed, because results are
identified by a date and time stamp that is created
automatically. However, if the system is to be tested
under a different configuration for comparison purposes,
changing the Variant number will record the results as
though it was an entirely different system. This optional
information is intended to assist the user in tracking test
results for multiple configurations.
Example:
Machine ID Var1 Var2 Var3 Var4 Var5
------------- ---- ---- ---- ---- ----
PC1 1 0 0 0 0
PC1 2 0 0 0 0
PC1 3 0 0 0 0
PC2 1 0 0 0 0
PC2 2 0 0 0 0
PC2 3 0 0 0 0
In the example above Var1 is used to represent testing a
machine with 3 different video boards with two different
machines.
----------------------------------------
Machine ID is a 12-character limit field for recording the
serial number or other identifier of the system being
tested.
Five Variant boxes, each providing a three-character limit
field, are available for recording the version or iteration
number of the current test. This information is especially
useful when testing a single system under a variety of
configurations.
Machine Name is 12-character limit field for recording the
brand name and model of the system being tested.
Project is a seven-character limit field for recording the
test project name, if any.
Test Org is a 25-character limit field for recording the
name of the organization performing the test.
Tester is a 25-character limit field for recording the name
of the person performing the test.
Machine Description is a 224-character limit field for
recording pertinent information about the system to be
tested.
Choosing Browse opens the Machine Browser dialog box. It
displays a list of all tests committed to the database as
well as the current unit under test. Highlighting a test
from the list and then choosing Select returns to the
Machine Information dialog box and loads the machine
information for that test into the Machine Information
fields.
Choosing the Use button will run MScope and then bring up
the "DOSMark" dialog box. (For more information about
"DOSMark" see View menu options.)
Choosing Cancel exits the program and return to the DOS
prompt.
PC BENCH MAIN MENU
The Main Menu displays seven selectable pull-down menus:
file, view, performance, compatibility, quality, set, and
help.
FILE MENU
Batch Editor
The Batch Editor is used to select the entire suite of
benchmark tests or to create a customized suite of tests to
run. The Select All or Clear All buttons perform these
functions. Individual tests are selected or deselected
using a mouse or Enter key and by using the up/down arrow
keys to designate the test choice. A check mark visually
confirms that the selected test(s) will be run when the
Execute button is chosen. Many sub-tests cannot be
individually selected. If a subtest is desired, you must
choose the next highest test group that corresponds to the
category desired.
A customized suite of tests may be saved to a batch run file
to re-run the same set of tests. This is accomplished by
selecting a set of tests and then choosing the Write dialog
box. Type a DOS filename in the field provided. The Read
box is used to recall a batch run file. Again, type the
name of the file to recall. No tests are executed until the
Execute box is chosen.
Print
Print is used to print test results and to export results.
Choosing Print displays the Machine Browser dialog box. One
test and date stamp can be selected. The Select option will
open the Print Results dialog box.
To send results to a printer, type the printer device
(example: LPT1) in the Enter Filename field. Choose the
.TXT file option and then choose OK.
A .CSV file may be printed should you desire to examine the
file contents under that format, but the principle purpose
of .CSV format is to create an file format that can be
imported into a spreadsheet program.
Any error messages will appear in the Message field.
To export the file, type a filename (8 character limit. The
format extension (.CSV or .TXT is appended automatically),
choose the desired format type, and choose OK.
Quit
Quit is the initial step toward exiting the program. Quit
brings up the Exit Program dialog box. It provides a
selectable option (checkbox) to permanently commit the
results to the database. In other words, when a test is
run, the results are written to a database file. Upon
exiting the program, choosing the commit option records
these results permanently to the database. If you do not
wish to save the results from a test session, leave the
commit field unchecked before exiting. If results are not
committed to the database, the next time the program is
executed, you will be given the option of using the
uncommitted results from the previous test run or deleting
those results from the database.
VIEW MENU
This menu is used to show test results, open the "DOSMark"
dialog box, and show the test system's component
information.
The Results option displays the Results dialog box. It
lists the tests and associated test results from the current
run. It also provides options for comparing as well as
displaying a graph of previously committed results against
the current run. Before the Compare function can be used,
however, comparison results from one or two previously
committed test runs must be selected from Comparison A
and/or Comparison B under the Set menu. (See Set menu.)
Once a comparison (A and/or B) has been set, select the
tests to compare (a check to the left of the test indicates
it is selected) in the Results dialog box, then Choose
Compare to display a comparison of the current results with
the set comparison(s). Choosing the Graph option displays a
column chart of the current run against any comparisons set.
The "DOSMark" option displays the "DOSMark" dialog box. The
"DOSMark" is an overall system performance value. A higher
score indicates better overall system performance. It is
generated from selected tests from the processor, memory,
video, and disk test groups. Not all of the tests in the
entire test suite are used to generate the "DOSMark";
instead, many tests exist to provide further detailed
information on each subsystem. The "DOSMark" uses a
weighted harmonic mean based on the tests' overall workload,
and their importance is relative to average application use.
To run the "DOSMark" suite, choose the "DOSMark" button and
then choose Execute. The test will take a little over an
hour to run. If all the tests in the "DOSMark" test suite
are successfully completed, the "DOSMark" value replaces the
NOT RUN message on the "DOSMark" button
The processor, memory, video, and disk harmonic groups test
specific aspects of each subsystem and produce an overall
value. Again, a higher score indicates a better overall
performance of the associated subsystem. If all the tests
in the harmonic test suite are successfully completed, the
harmonic value replaces the NOT RUN message on the
associated button
The "Processor Harmonic" tests the processor alone, the
CPU/memory interface, the handling of floating point
emulation, and the math co-processor. The processor is
tested both in protected and real modes using 16-bit
instructions. (32-bit instruction tests are currently under
development and will be available in subsequent releases.)
The processor tests are geared to accurately simulate
popular applications. They test bus utilization, CPU cache,
CPU cache interface, and the CPU instruction set based on
the average use of applications. Each of the tests is
weighted to produce the harmonic mean. The mean is weighted
to show the relative impact of each of the tests based on
the average use of applications. To run the "Processor
Harmonic" suite, choose the "Processor Harmonic" button and
then choose Execute.
The "Memory Harmonic" uses a series of low-level memory
reads and writes to conventional, extended, graphic, text,
and BIOS memory. The harmonic mean weights are set to
represent the average memory use of each of these areas
under popular applications. To run the "Memory Harmonic"
suite, choose the "Memory Harmonic" button and then choose
Execute.
The "Disk Harmonic" shows what effect using different file
and block size combinations has on disk performance. The
harmonic has been weighted to reflect average use of the
disk when running typical applications. To run the "Disk
Harmonic" suite, choose the "Disk Harmonic" button and then
choose Execute.
The "Video Harmonic" uses the low-level tests (from PC Bench
1.0) and weights the results to show how average
applications use the video subsystem. To run the "Video
Harmonic" suite, choose the "Video Harmonic" button and then
choose Execute.
System Info
The System Info option under the View menu displays the
machine information captured and logged when MScope was run
during the start-up phase, i.e., when the Use option is
chosen from the Machine Information dialog box, the program
automatically runs MScope and captures and logs system
information. This information includes microprocessor type,
CPU clock speed, primary video, and other important
information.
PERFORMANCE MENU
Processor Group
This test suite is designed to test the processor through an
average application-based scenario. This test group
measures performance in both protected and real mode of the
CPU instruction set, CPU cache, and the CPU/memory interface
and includes the floating point emulation and math co-
processor tests.
The 16-Bit Protected Mode Small Mix and Real Mode Small Mix
tests are designed to run within the processor's cache.
This allows the CPU to be tested independently from the rest
of the system and thus gives a measure of relative CPU
performance. The tests use a mix of the instruction set
derived from popular applications' average use of the
processor instruction set. The real and protected mixes are
based on the average counts and execution times of
applications under DOS and Windows. Real mode tests should
run faster due to the protection checking overhead of
protected mode. These two tests should yield two numbers
per processor type independent of the machines.
The 16 Bit Protected Mode Standard Mix and 16 Bit Real Mode
Standard Mix tests are designed to fit within most second-
level caches. These tests use the bus, CPU cache, and CPU
based on average application use. The instruction set mixes
are based on the 16-bit mix. These tests indicate how
effectively the processor interacts with the rest of the
system and how efficiently a system manufacturer has
interfaced the CPU to the memory subsystem. In addition,
this measures the performance impact of the presence or
absence of an external cache subsystem.
The Prime Number Sieve test times the execution of a routine
to find the prime numbers between 0 and 8190.
The Floating Point Emulation test (also known as the
Floating Point Calculation Without Co-processor) sets up a
floating point emulation program in RAM and then exercises
the processor and tests RAM access speeds during floating
point calculations. The processes performed in the floating
point test are identical to those used in the Co-processor
Test.
The Math Co-processor test exercises the math co-processor
using the same floating point calculations that are used in
the Floating Point Test. The test can be used to analyze
the speed differences of co-processors in different systems.
Additionally, the processing speed gained by using a co-
processor can be seen by comparing its Floating Point Mix
scores with its Math Co-processor scores.
The String Sort and Move test times the execution of a
bubble sort performed on 200 random strings containing 16
characters each. This test is useful in testing systems
with a processor RAM cache, as sorting requires reading and
writing some of the same data repeatedly. Systems with
write-back caches will usually perform better on this test
than will systems with write-through caches.
Memory Timing Group
These tests give performance values for accessing the
various types of memory in the system. In general, the
results from these tests will be determined by the speed and
architecture of the system's memory, memory caches (if
present), and the type of CPU.
The extended memory tests enter protected mode to access
memory above 1 megabyte. As a result, these tests will not
execute if an extended memory driver is already installed,
such as EMS (EMM386, 386MAX, etc.), DPMI, or VCPI. The test
uses the full range of extended memory available to it;
however, during the write tests it avoids any area already
locked by an XMS (HIMEM) driver, RAM disk, or other INT 15h-
compatible software. The start and end addresses are
displayed to show what range of extended memory was used.
If HIMEM.SYS is loaded the write addresses will not be
displayed because the actual addresses accessed may not be
continuous. These results reflect the performance of
extended memory as it would be accessed by other protected-
mode applications such as Windows.
There are six tests, a read test and a write test for each
data width:
8-Bit Extended Memory Read/Write
16-Bit Extended Memory Read/Write
32-Bit Extended Memory Read/Write
Under the Graphics Memory test, the type of graphics adapter
present is detected and the appropriate buffer address is
used to read and write blocks of data directly to the
graphics adapter's memory. During memory writes, the PC
Bench screen will be overwritten, but it should be restored
at the end of the test.
There are six tests, a read and a write for each data width:
Graphics Memory 8 Bit Read/Write
Graphics Memory 16 Bit Read/Write
Graphics Memory 32 Bit Read/Write
The BIOS Memory tests only attempt to read from the BIOS
since it is stored in ROM. This test will show significant
performance differences according to whether the BIOS ROM is
shadowed and/or cached, which is a configuration choice on
many systems. There are three tests:
BIOS Memory 8 Bit Read
BIOS Memory 16 Bit Read
BIOS Memory 32 Bit Read
The Text Memory tests detect the type of display adapter and
use the appropriate screen buffer address to write directly
to the display adapter's memory.
There are four tests, a read and a write test for each data
width:
Text Memory 8-Bit Read/Write
Text Memory 16-Bit Read/Write
The Conventional Memory tests access memory in the 640-Kb
area managed by DOS. These tests reflect the memory
performance of programs running in Real mode.
There are four tests, a read and a write test for each data
width:
Conventional Memory 8 Bit Read/Write
Conventional Memory 16 Bit Read/Write
EGA/VGA Text Group
The Unscrolled BIOS Write test times the writing of data to
the screen using interrupt 10h BIOS calls. Fast times are
advantageous for programs that display large amounts of data
in a non-scrolling fashion.
The Scrolled BIOS Write test times the writing of data to
the screen using interrupt 10h BIOS calls. Once the screen
is filled, the test scrolls through an entire screen by
adding one new line at a time. This test writes the same
amount of data as the non-scrolling test. The difference
between the two times indicates the overhead attributable to
scrolling the screen. Fast times are advantageous for
applications that frequently cause the screen to scroll.
The pair of BIOS write tests gives a good indication of
video BIOS speed. If the video BIOS is shadowed (loaded in
fast system memory), throughput will be higher for these
tests.
The Direct Screen Write tests write data to the video
adapter screen memory using the MOVSB (8-bit) and MOVSW(16-
bit) instructions. For a properly installed 16-bit VGA
card, the 16-bit test will yield about twice as much
throughput as the 8-bit test. These tests provide the best
measure of the text throughput of a display adapter.
EGA/VGA Graphics Group
The Write Mode 0 Fill tests fill the video adapter's memory
with varying colors using the STOSB (8 bit), STOSW (16 bit),
and STOSD (32 bit, for systems with at least an 80386
processor) instructions with EGA/VGA write mode 0 and all
bit planes enabled. Commonly used by graphics software
applications, it is one of the best measures of video
adapter throughput in graphics mode.
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. Memory to Screen BITBLT
functions are used when updating the images on a display.
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. Since system memory
is being updated, there is no visible activity for the
duration of this test. Screen to Memory BITBLT functions
are used when modifying or saving screen images.
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. Screen to Screen BITBLT
functions are used when moving objects from one area of the
screen to another.
Disk Access Group
The BIOS Disk Seek tests (Sequential and Random) measure
mechanical track to track disk drive access times. Both
sequential and random access are tested. The test uses the
BIOS Interrupt 13h. It cannot be used to test floppy
diskettes or mass storage media that do not have an
Interrupt 13h interface. The test involves 1,000 sequential
seeks (alternating between cylinders 0 and 1) 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. The
result is the average access time per track.
The DOS Disk Access test measures the time necessary to
perform 1,000 read requests at random locations on the disk
using the DOS Interrupt 25h. This test should work with any
device that DOS recognizes as a disk. The test reports the
total time required to complete the requests in seconds.
NOTE: If a seek error occurs in either of the above tests,
a message will appear for a few seconds and then a
replacement sector will be chosen and the test will
continue. This is not necessarily an indication of a bad
disk drive. These tests access sectors which 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.
Disk Throughput Group
This test suite is designed to show what effects different
file and block size combinations have on disk performance.
The disk tests use 256-KB, 1-MB, 2-MB, 4-MB, 8-MB, 16-MB,
and 32-MB file sizes to test the disk. For each file size
the test suite uses 200-byte, 512-byte, 2-KB, and 4-KB block
sizes. Sequential reads, sequential writes, random reads,
and random writes are done for each block size. All
together there are 105 tests.
Three main aspects of disk performance become clear through
this test suite. The results show the variation in DOS
overhead caused by various block sizes. The 200-byte block
size causes the 512-byte sector boundaries to be crossed
during reads and writes. The results reflect the overhead
due to the inefficiency of crossing the sector boundaries.
In-cache versus out-of-cache, throughput differences can be
seen by comparing the results for a file size that fits in
the cache against results for a file size that exceeds the
cache.
Battery Rundown Test
The Battery Rundown test is designed to exercise a laptop
computer's battery system in a worst-case scenario. The
test is started after charging the computer's battery
according to the manufacturer's specifications and disabling
all power-conservation features. The test teletypes 10
screens full of information to the display and then writes a
15K file to disk with a time stamp of elapsed time. This
operation continues until the battery runs out of power.
Time the Timer Test
This test calculates elapsed time using DOS calls. Some
systems may not report a correct time when running a version
of DOS that is not configured for the system. In this case,
the results obtained from using these tests may be skewed.
The Time the Timer test allows measuring the accuracy of the
system's timer by comparing it to some external source (such
as a stopwatch).
COMPATIBILITY MENU
VGA Group
The compatibility test for VGA hardware performs several VGA
register-level functions which exercise the different
subsystems of a VGA or compatible display board (including
the VGA controller chip, video memory, and RAMDAC). Since
the test was designed with the original VGA in IBM PS/2
systems in mind, most boards don't pass every test.
However, a failure of one or two of the tests doesn't
usually indicate serious compatibility problems for most
software.
QUALITY MENU
EGA/VGA Monitor Group
The quality test permits images to be displayed in various
modes from black and white text to VGA color graphics.
These tests are used to determine what display modes a
monitor/display adapter combination can support. Also, they
are used for evaluating laptop screens to see how well they
can map colors to gray scale.
SET MENU
The Set Menu contains Test Machine, Comparison A, and
Comparison B selections.
Test Machine
Choosing this option opens the Machine Information dialog
box. Modifying this information and choosing Use will update
the database with the modifications for the current test.
Comparison A
Choosing this option opens the Machine Browser dialog box
and displays a list of the test runs committed to the
database. Selecting a test run and date stamp sets up a
comparison to the current run. Results comparisons can be
viewed in the Results dialog box (see Results under the View
menu for more information) and under the "DOSMark" option
(see "DOSMark" under the View menu for more information)
Comparison B
Choosing this option opens the Machine Browser dialog box
and displays a list of the test runs committed to the
database. Selecting a test run and date stamp sets up a
comparison to the current run. Results comparisons can be
viewed in the Results dialog box (see Results under the View
menu for more information) and under the "DOSMark" option
(see "DOSMark" under the View menu for more information)
HELP MENU
The Help Menu provides options for General and About.
Selecting General opens a help information text file.
Selecting About displays a list of the personnel involved
with the development of the benchmark program.
REGARDING RESULTS
After the benchmarks have run, the results are displayed to
the right of the associated test in the Results dialog box
under the View menu. If other results have previously been
obtained and committed to the database, they can be compared
with the current test results. Use Comparison A and/or
Comparison B under the Set menu to select data sets to use
as Test A or Test B. Choosing Compare in the Results dialog
box will display the results. The menu permits up to two
comparisons with the current test. These can also be
displayed as a column chart with the Graph button.
The "DOSMARK" result is an overall composite score based on
specific test selections from the full test suite. The
results are aggregated with weighting factors assigned to
account for different units of measurement and standard
application-specific data on component or subsystem use.
"DOSMark" is intended to be a consistent scoring procedure
comparing DOS-compatible systems for business use.
The performance of system components can also be analyzed as
discussed below:
After running the tests but before exiting the menu, choose
the Print dialog box under the File menu to export the data.
There will be a dialog box to select the test data followed
by a dialog box to export the data (*.CSV or *.TXT). Comma-
delimited files (*.CSV) are best suited for spreadsheet
analysis, and text format is preferable for databases and
word processors. When supplying the file name, do not enter
the period or a three-letter file extension. This is
automatically supplied depending on the format choice you
make.
Once the results have been exported, choose Quit under the
File menu. Use "Commit results to database" by selecting the
box (an X appears in the box). The file is automatically
written to the directory containing the benchmark files.
The benchmark directory also contains the internal database
files named Test.DBF and Result.DBF. that can be used to
prepare a working spreadsheet using database software (such
as dBASE).
Four subsystems are tested: CPU, memory, VGA, and hard disk.
CPU tests are broken down into instructions performed in the
CPU and in the memory cache, for real and protected modes in
both cases, and numerical operations including floating
point calculations and math co-processor operations.
Memory results are all expressed in kilobytes per second
(kbps), so they all can be aggregated into a single
performance result.
VGA results for text mode are expressed as thousands of
characters per second. Graphics mode results are expressed
as thousands of pixels per second. These must be aggregated
separately and normalized before they are weighted and
combined into a single VGA performance score.
Disk results come in two forms: disk access in elapsed time
and disk throughput expressed in kbps.
Before comparing products, a standard set of results must be
obtained and analyzed so that they can be used to normalize
the test system results. If no standard has been previously
adopted, the sample averages can be used for normalization.
TESTING METHODOLOGY
PC Bench can be used to test systems' performance in any of
the following approaches: single system comparisons, single
system/multiple configurations comparisons, multiple system
comparisons, and multiple systems/multiple configurations
comparisons.
Single-system comparisons are the most simple. In this
scenario, PC Bench is run a few times to establish an
average. When testing a single system under multiple
configurations, changing the Variant numbers in the Machine
Information display will identify the results by their test
iteration number. Test results committed to the database
will reflect the different data recorded in the Variant
fields of the Machine Information display.
If PC Bench is to be run on multiple systems, then the
program must be installed on each test system. To compare
the results of the first system with those of the second
system within the Bench program, the results files must also
be copied into the second system. In this scenario, system
1 is tested under PC Bench, and the results are committed to
the database. The Bench program is then loaded onto the
second system, and the results files from the first system
are then copied onto the second system. The files to be
copied are MACHINE_.MDX, MACHINE_.DBF, RESULT_.MDX,
RESULT_.DBF, SYSINFO_.MDX, and SYSINFO_.DBF.
The information in the Machine Information display is used
to identify the specific test system and configuration and
thus prevent the results files from being over-written. As
before, multiple configurations are identified by using the
Variant fields in the Machine Information box.
ERRORS and ERROR RECOVERY PROCEDURE
If a File Error -70 occurs, it is probable that the database
files have become corrupted. There are two kinds of database
files, those used by the bench program and those which store
user data.
Try this first.
LEVEL 1 Severity. This should not impact user data.
Go to the DOS prompt.
Execute the program REINITDB by typing REINITDB at the DOS
prompt.
Try running the BENCH program again.
If the error re-occurs proceed to the next level.
LEVEL 2 Severity. This will delete user data.
Go to the DOS prompt.
Delete *.DBF
Delete *.MDX
{If user data has been previously saved using a backup or
save activity, perform the restore function at this time.}
Execute the program REINITDB by typing REINITDB at the DOS
prompt.
Try running the BENCH program again.
The error should not re-occur.
If Insufficient Memory or Insufficient Disk Space error
messages appear, refer to the SYSTEM REQUIREMENTS section of
this file.
If the Insufficient Disk Space on Drive <X> or Insufficient
Disk Space to Create File messages occur, it means that the
Bench program does not have enough room to run the disk
performance tests on your system for the particular test
required.
=========
Protected Memory
Unexplained Lockups
There is a potential for unexplained lockups to occur when
performing protected mode memory tests on 286, 386, or 486
machines. This is due to the differences between the way the
low-level instructions interact with the CPU chip.
Therefore, if you experience a lockup while memory tests are
executing, take the following steps:
1. Reboot the machine.
2. From DOS, run the alternative batch file below by
typing:
MKMEM386.BAT
(NOTE: MKMEM286.BAT will restore the default.)
3. Start the BENCH program.
4. Do not commit the data from the previous run. Delete
the data.
5. Rerun the tests for the machine configuration you
select.
Remember that when you execute one of the batch programs
mentioned above, it permanently renames the test file that
is executed. To reset it for a different machine, you must
execute the other batch file.