User's Choice Windows CD/User's Choice Windows CD (CMS Software)(1993).iso/windows5/winbn251.zip/WINBENCH.HLP

Text File | 1992-04-22 | 13KB | 284 lines

PC Magazine Labs Windows Graphics Benchmark Help
Version 2.51

Version 2.51 computes the Graphics WINMARK, the weighted geometric mean
of twelve Windows graphics operations that are the most frequently used
by major Windows applications.  In November 1991, PC Labs instrumented
three classes of leading Windows applications (text-oriented, bitmapped
graphics, and vector graphics applications), and found that several BitBlt,
line drawing, and rectangle filling operations prevailed over all others.
The specific operations are:

  Memory to Screen Source Copy
  Memory to Screen Pattern Copy
  Memory to Screen Pattern Invert
  Memory to Screen Destination Invert
  Memory to Screen Whiteness
  Draw Horizontal Single Lines
  Draw Vertical Single Lines
  Draw Diagonal Single Lines
  Draw Horizontal Polylines
  Draw Vertical Polylines
  Draw Diagonal Polylines
  Random Rectangles

When you start WINBENCH, it gives you the options of running the WINMARK
tests, running the Advanced (detailed) tests, or exiting from the program.
The Graphics WINMARK result is the weighted GEOMETRIC mean of the twelve
tests. In the case where all the weights add up to 1.0, the weighted
geometric mean is calculated as the product of each test result raised
to the power of its weight.  In more mathematical terms,
    WINMARK = (r1**w1)*(r2**w2)*...(rn**wn)
where r1...rn are the test results, and w1...wn are the corresponding
weights, and the paired asterisks (**) denote the mathematical operation
of exponentiation.

You may run the WINMARK tests and/or display the results by selecting
Options | WINMARK at any time.

To manufacturers of personal computers and graphics hardware, publishers of
Windows software drivers, and others who will publicize WINMARKs for
commercial and non-commercial purposes:

To minimize the possibility of misrepresenting the results of the PC Magazine
WINMARK test, you should make sure that any publication or distribution of these
results in printed or electronic form includes the brand name, clock speed, and
processor model of the PC on which the test was run, a concise description of the
graphics hardware used for the test, and the Windows driver name, description, and
version used for the test.  You may choose to augment this information by specifying
the graphics chip set, host PC cache memory, and any other pertinent information.
The Graphics WINMARK and the other WINBENCH tests are sensitive to all of these
factors.
A sample description follows:

A 37MHz White Label 80486 PC with a Generic Super VGA card running with
Version 3.17 of VGA.DRV (640x480x16 colors) achieved a WINMARK result of
2,366,439.

The Compaq 386/25e used as a reference PC for the WINMARK was configured with
8 megabytes of 80 nanosecond DRAM and VGA integrated onto the system board.  This
PC was tested with the standard VGA driver packaged with Windows 3.0, Version
3.00 of the driver VGA.DRV operating with 640x480x16 colors.

PC Labs is currently doing further instrumentation and analysis of major Windows
applications, using more sophisticated graphical device interface (GDI)
profiling tools recently made available by Microsoft.  Subsequent releases of
WINBENCH may reflect adjustments in both the functions used and their weights,
depending on the results of the analysis.

BENCHMARK OPTIONS

You may select the following items in the Modify Options dialog box:

  User Preference Options

     You may enable or disable the display of the timing dialog boxes at the end of
each set of tests via the DISPLAY RESULTS radio buttons.

     You may enable or disable the output of a results file and change the file
format via the FILE OUTPUT radio buttons.

     If you do select file output, then you can enter a name for the file and a
heading for the file into the edit boxes in the FILE OPTIONS group box.

  Save Checkbox

     To the left of the OK button in the Modify Options dialog box is a single
checkbox, labeled "Save".  If you select this option, WINBENCH will remember your
current option settings, and automatically use these settings each time WINBENCH
runs.  WINBENCH saves the option settings in the [Version.2.0] section of the
WINBENCH.INI file in your Windows directory.

     If you use the same file name for different runs of WINBENCH, results from the
current test are appended to the results of preceding tests.


SUGGESTED USAGE

     If this is the first time you've used the WINBENCH benchmark, you may wish to
experiment with it.  You can select and run any individual test via the menus.

     If you are timing the performance of a display card, you should run all of the
detailed tests.  The fastest way to run all of the tests is to disable the display
of results, and then enable the file options.  Running in this mode allows the
benchmark to proceed without intervention, and to record its results in a file.

     To run all detailed tests automatically:

      Select "Modify Options" from the OPTIONS menu.
      Select "No" for Display Results.
      Select "Yes" for File Output.
      Enter a name for the results file.
      Enter a file heading for the results file.
      Select OK.
      Select "Run All Tests" from the OPTIONS menu.

     You may interrupt any sequence of tests by pressing ESC.  If ESC is pressed, the
test currently running will complete, and you will be asked if you wish to cancel or
continue.


DESCRIPTIONS OF THE TESTS

BitBlt
     The BIT BLock Transfer operation (BITBLT) with all of its variations is the
single graphics operation most used within the Windows environment.

      The BitBlt Alignment tests perform a set of BitBlt copy operations of
     various sizes from one area of the client window to another.  Aligned and
     unaligned areas show the differences in throughput for video cards that perform
     best when the data is aligned on an eight-pixel horizontal boundary.

      The BitBlt Overlap tests exercise the ability of a video card and Windows
     driver to perform BitBlt copy operations where the source and destination
     overlap in the window.

      The BitBlt ROPS tests are a series of 48 tests of the most frequently used
     BitBlt operations for three combinations of source and destination: memory to
     screen, screen to screen, and memory to memory.

      The BitBlt Screen/Memory tests copy source images in memory and on screen
     to random destinations in the client window.

Curves
     The Curves tests show the difference in performance between video cards that
have hardware curve drawing operations and those cards that do not.

      The Partial Arcs tests draw arc shaped lines.

      The Ellipses test draw and fill ellipses.

Lines
     The Lines tests are the second most frequently used group Windows graphics
operations.  They show the difference in performance between video cards that have
hardware line drawing operations and those cards that do not.

      The Single Lines tests draw single horizontal, vertical, and diagonal lines
     across the window.  Drawing lines in different directions shows that various
     video cards and drivers draw lines faster in one direction than another.  With
     many video cards, horizontal line drawing is fastest, followed by vertical, then
     diagonal.

      The PolyLines tests draw many horizontal, vertical, and diagonal lines
     across the window with a single Windows graphics operation.  Interpretation of
     results is similar to the Single Lines tests.

      The All Directions test draws lines at every possible angle.  The results
     represent a composite of the line drawing capabilities of a video card and
     driver.

      The Wide Lines tests draw lines of varying width in random directions and
     random lengths.

Polygons
     The Polygons tests exercise operations used by business graphics and drawing
software.

      The Alternate Fill test fills the area between the odd-numbered and the
     even-numbered sides of a polygon.

      The Winding Fill test draws a border that encloses a polygon but does not
     overlap, then fills the shape defined by the border.

Rectangles
     The Rectangles tests are the third most frequently used group Windows graphics
operations.  They show the difference in performance between video cards that have
rectangular area filling operations and those cards that do not.

      The Boxes tests fill concentric rectangles of various sizes.  One test
     fills square rectangles and the other fills rectangles with rounded corners.

      The Random test draws and fills rectangles of random size and placement on
     the screen.  It uses different colors and fill patterns for each rectangle.

      The Patterns test fills the client window with rectangles using various
     shades from black to white, and using the standard Windows cross-hatching
     patterns.

Text
     The Text tests exercise Windows operations that are used by Windows itself and
by word processors, desktop publishing software, spreadsheets, and other software
that makes extensive use of different fonts in different sizes.

      The Image Fonts tests repeatedly fills the client window with the system
     font, and two sizes each of the Roman and Helvetica fonts.

      The Scroll Fonts tests repeatedly scroll the client window upward and fill
     the bottom of the screen with text.  There are five tests using the system font,
     and two sizes each of the Roman and Helvetica fonts.  These tests are typical of
     the scrolling operations done by word processors and desktop publishing
     software.

      The Background Mix tests display the system font against opaque and
     transparent backgrounds.

      The Greyed Text test displays the system font in grey at random positions
     and angles on the screen.

      The Mixed Fonts test displays the Courier font in normal, italic,
     underline, and bold faces.

Misc
     The Misc tests perform miscellaneous graphics operations done by Windows.

      The Stretch tests display the PC Magazine logo in different sizes,
     stretching or shrinking it to eleven different dimensions.

      The Image Data test displays the PC Magazine logo in 256 combinations of
     foreground and background colors.

User Interface
     The User Interface tests exercise high-level functions that are part of the
Windows user interface.

      The Dialogue Boxes test draws a dialogue box on screen, then restores the
     part of the client window covered by the dialog box.

      The Rubber Banding test varies the size of a rectangle, an operation used
     in drawing packages and by software that supports the Windows clipboard

      The Scroll Window tests scroll text inside a window, an operation used by
     list boxes and combo boxes.  One test scrolls with the window aligned on an
     eight-pixel horizontal boundary, and the other uses an unaligned boundary.
     Aligned and unaligned scrolling show the differences in throughput for video
     cards that perform best when the data is aligned on an eight-pixel horizontal
     boundary.

      The Clear Window test clears the client window.

Clipping
     The Clipping tests show the effect of graphics operations performed through a
combination of four rectangular clipping areas.

      The BitBlt Clipping test displays the PC Magazine logo at random points in
     the client window.

      The Line Clipping test draws lines in all direction.

      The Text Clipping test repeatedly fills the client window with text.

Options
     The Options menu allows you to select different options and to view test
information.

      The System Information menu choice displays information about the current
     combination of video card, video driver, and PC being tested.

      The Modify Options menu choice allows you to change the options for running
     tests.  It is described above.

      The Run All Tests menu choice runs all of the tests using the current
     options.

      The WINMARK menu choice runs the tests used to calculate the Graphics
     WINMARK value, the weighted geometric mean for the combination of graphics
     adapter and PC system on which the tests are run.

      The Demo Mode runs all of the tests but neither displays test results nor
     produces a file containing test results.

      The Compare Results menu choice allows you to select another set of test
     results to compare with the results run to date for the current test
     configuration.  You may select test results only from files written by WINBENCH
     in XLS format into the directory from which you have launched WINBENCH.