PC WEEK LABS SYSTEM BENCHMARKS -- VERSION 1.20

Type BENCH at the DOS prompt to run the benchmarks.


GENERAL INFORMATION

These system benchmarks are a series of more than two dozen
tests grouped into six modules. Each module measures the
time required to perform typical elements of an application.

Text Scrolling displays short, medium, and long lines
of text using DOS, BIOS, and C library routines.

Text Windows scrolls text through windows.

Graphics Video draws random-sized areas, lines,
ellipses, and 15 small graphs.

CPU Speed generates and sorts 30,000 integers.

Floating Point performs a Fast Fourier Transform on
internally generated data.

Disk Speed generates data, indexes, sorts, and
generates reports.

To run an individual test, move the cursor bar to the test
description on the main menu and press Enter. Select
"Run all above tests" to run the entire test suite.

Select "Display and compare results" to compare test results
with those stored in the benchmark library.

To attach a description to the benchmark results select
"Describe this machine."

To save the benchmark results select "Save data file."

"Return to DOS" terminates the benchmark suite.

"About these benchmarks" displays this information.

Pressing Esc returns control to the previous menu; if
pressed when using the main menu, it returns control to DOS.


INTERPRETING THE RESULTS

Benchmark results are presented in tabular and graphical
form. Times presented are in seconds. In each case the
first column of numbers are the benchmark test times for
the reference machine, the second are times for the
comparison machine, and the third is a performance factor.
The performance factor is calculated by dividing the
results in the first column by those in the second, and
expressing the result as a percentage. Thus, if the
reference machine runs a particular test in 10 seconds,
and the comparison machine runs it in 5, the performance
factor is 200%.


ADDITIONAL INFORMATION

The benchmark results file, and the default benchmark
reference file can be specified on the HLBENCH command
line by entering:

BENCH bfile rfile

where bfile and rfile are the respective file names.

The default headings built into the benchmark result
description record can be overwritten if desired.

By default, Graphics Video tests the the highest resolution
video mode available. A lower-resolution video mode can be
tested if desired (or if the benchmark suite is unable to
detect the proper video mode), by entering the DOS command:

SET HLGRAPH=n

The possible values for n, and associated video modes are:

4 CGA Medium-res (320x200, 4 color)
6 CGA "high" res (640x200, 2 color)
8 Hercules (720x348, 2 color)
15 EGA Monochrome (640x350, 2 color)
16 EGA Color (640x350, 16 color)
17 VGA Monochrome (640x480, 2 color)
18 VGA Color (640x480, 16 color)
19 VGA 256-color (320x200, 256 color)

Disk Speed tests drive C: by default. A different drive can
be tested by entering the DOS command:

SET HLDISK=x:

before running the test, where x is replaced by the letter
of the drive to be tested. The drive should have at least
100KB available storage.

The name of the default benchmark reference file used by
"Display and compare results" is stored in XREF.TXT. A
different file can be specified by editing XREF.TXT.

Print data from the "Display and compare results" print
option is written to PRN by default. It can be redirected
to either a serial or parallel printer using the DOS MODE
command. The printer output can be directed to a file or
different device by entering the DOS command:

SET PRINTER=filename

where filename is the desired destination of the data.
Print data is always appended to the file selected.

Additional information about the program and utility files
that make up the benchmark suite are contained in the
READ.ME file.

These benchmark programs and additional TIM files can be
downloaded from PC Week EXTRA!

The design of the benchmark tests, the test methods, and
how to run the tests and interpret the results is described
in "High Level Benchmarks," Kent Quirk, PC Tech Journal,
September 1988, p. 54. This article can be found on PC
Week Extra!

Last Help Update: 10/24/90

PC WEEK LABS SYSTEM BENCHMARKS -- VERSION 1.20

Type BENCH at the DOS prompt to run the benchmarks.


GENERAL INFORMATION

These system benchmarks are a series of more than two dozen
tests grouped into six modules. Each module measures the
time required to perform typical elements of an application.

Text Scrolling displays short, medium, and long lines
of text using DOS, BIOS, and C library routines.

Text Windows scrolls text through windows.

Graphics Video draws random-sized areas, lines,
ellipses, and 15 small graphs.

CPU Speed generates and sorts 30,000 integers.

Floating Point performs a Fast Fourier Transform on
internally generated data.

Disk Speed generates data, indexes, sorts, and
generates reports.

To run an individual test, move the cursor bar to the test
description on the main menu and press Enter. Select
"Run all above tests" to run the entire test suite.

Select "Display and compare results" to compare test results
with those stored in the benchmark library.

To attach a description to the benchmark results select
"Describe this machine."

To save the benchmark results select "Save data file."

"Return to DOS" terminates the benchmark suite.

"About these benchmarks" displays this information.

Pressing Esc returns control to the previous menu; if
pressed when using the main menu, it returns control to DOS.


INTERPRETING THE RESULTS

Benchmark results are presented in tabular and graphical
form. Times presented are in seconds. In each case the
first column of numbers are the benchmark test times for
the reference machine, the second are times for the
comparison machine, and the third is a performance factor.
The performance factor is calculated by dividing the
results in the first column by those in the second, and
expressing the result as a percentage. Thus, if the
reference machine runs a particular test in 10 seconds,
and the comparison machine runs it in 5, the performance
factor is 200%.


ADDITIONAL INFORMATION

The benchmark results file, and the default benchmark
reference file can be specified on the HLBENCH command
line by entering:

BENCH bfile rfile

where bfile and rfile are the respective file names.

The default headings built into the benchmark result
description record can be overwritten if desired.

By default, Graphics Video tests the the highest resolution
video mode available. A lower-resolution video mode can be
tested if desired (or if the benchmark suite is unable to
detect the proper video mode), by entering the DOS command:

SET HLGRAPH=n

The possible values for n, and associated video modes are:

4 CGA Medium-res (320x200, 4 color)
6 CGA "high" res (640x200, 2 color)
8 Hercules (720x348, 2 color)
15 EGA Monochrome (640x350, 2 color)
16 EGA Color (640x350, 16 color)
17 VGA Monochrome (640x480, 2 color)
18 VGA Color (640x480, 16 color)
19 VGA 256-color (320x200, 256 color)

Disk Speed tests drive C: by default. A different drive can
be tested by entering the DOS command:

SET HLDISK=x:

before running the test, where x is replaced by the letter
of the drive to be tested. The drive should have at least
100KB available storage.

The name of the default benchmark reference file used by
"Display and compare results" is stored in XREF.TXT. A
different file can be specified by editing XREF.TXT.

Print data from the "Display and compare results" print
option is written to PRN by default. It can be redirected
to either a serial or parallel printer using the DOS MODE
command. The printer output can be directed to a file or
different device by entering the DOS command:

SET PRINTER=filename

where filename is the desired destination of the data.
Print data is always appended to the file selected.

Additional information about the program and utility files
that make up the benchmark suite are contained in the
READ.ME file.

These benchmark programs and additional TIM files can be
downloaded from PC Week EXTRA!

The design of the benchmark tests, the test methods, and
how to run the tests and interpret the results is described
in "High Level Benchmarks," Kent Quirk, PC Tech Journal,
September 1988, p. 54. This article can be found on PC
Week Extra!

Last Help Update: 10/24/90
PC WEEK LABS SYSTEM BENCHMARKS -- VERSION 1.20 09/05/90

With this release of our high level benchmarks and lan performance
benchmarks, we at PC Week are proud to announce our new 6500 square
foot multiplatform, interoperability testing facility in Medford MA.

Included on this disk are two seperate benchmarks. It is recommended that
each is placed in a seperate directory prior to running the self-extracting
files. The file INSTALL.EXE self-extracts a suite of single-user tests
that measure memory, video, disk, and CPU performance. The file LANSTALL.COM
provides a load-generation and load-measurement tool for LANs.

In addition, we have included the file SPENCER.BMP, a Windows desktop
bit-mapped display. To use Spencer as your Windows desktop display,
you will have to copy SPENCER.BMP into your Windows subdirectory and
then use the Windows Control Panel and Desktop to select it as your
wallpaper. It is recommended that it be tiled rather than placed in the
center. SPENCER.BMP was created by Chris Stetson.

If you have any specific questions or suggestions, please
send them along with your registration form or in a separate letter
addressed to: PC Week Labs Benchmark Registration, 800 Boylston Street,
Boston, MA, 02199. The PC Week Extra Forum off of Compuserve is also an
excellent source of information as well as a place to post questions.


Abe Rosner
Director, PC Week Labs