Wave Issue 9712 5/21/97 Article 4-01

***Ziff-Davis Response to 3D WinBench 97 Review
by David Lohse

In response to our article on 3D Winbench 97 (WAVE #710, 5/8/97), Ziff-
Davis Benchmarking Operations (ZD-BOP) sent us several clarifications and
comments.

Responding to the section discussing the 3D Triangle Tests, they respond:

"There is a test named 3D Triangle/User Defined that together with the
User Triangles Test Settings page allows the user to vary any of the
parameters you list above."

Since we indicated that the basis of the 3D WinMark score was unclear, they
describe how it is calculated:

"The 3D WinMark is a scaled unweighted arithmetic mean of the results of
the 10 tests in the 3D Winmark suite. If a test cannot be completed due
to a lack of capabilities (e.g. the mipmaps tests) it gets a score of
zero. The scale factor is chosen so that a P 200 using software
emulation only gets a 3D WinMark result of 10. The 3D WinMark is
intended as a single overall measure of the features and performance of
an adapter."

Finally, they explain why in our initial test runs we did not receive a 3D
WinMark score:

"Results for the 3D WinMark suite were not returned because you set the
rendering buffer in the Scenes Test Settings page to 'Double buffered'.
You must set it back to the default 'Front buffer' to get a result. The
'Double buffered' mode is only used to examine the quality of the
rendering, not to get a performance measurement. See note [6] in the
results output."