Javier Castellar (javier++at++sixty.asd.sgi.com)
Fri, 26 Apr 1996 23:49:28 -0700
In my opinion your test suit will show *YOUR* software perfomance across the
different IGs, NOT the perfomance of the IGs themselves.
I know that your company have very strong skills on SGI and this makes your SGI
OpenGVS implementation very interesting but if you are using OpenGL instead of
Performer for the SGI-OpenGVS it means that you are taking a trade off of easy
portability versus perfomance.
If your SGI-OpenGVS were running on top of libpr i could accept that the pure
rendering speed (after substracting all the culling-LOD-App host load due to
OpenGVS) will be a good representation of the SGI IG speed but since you are
using OpenGL i may say that is not 100% sure that you are getting the top
rendering perfomance for most of the databases.
If you are running the benchmarks in single CPU (as it seems to be after
speaking with Judith Pafford) you are adding a distorting factor since VisSim
perfomance is not only graphics perfomance (you are adding your software
propietary culling-lod perfomance on top of the IG measure).
> The test suite as proposed will include a flight simulation stress test, a
>race car test, and a tank simulation environment test. The paper itself will
>shortly be published on the WEB at http://www.gemtech.com. Gemini has
received >favorable comments from some hardware manufacturers who are
interested in >officially endorsing the test suite. I have incorporated good
commments as >well from SGI (Judith Pafford). The results of tests will be
published in June >or July on our WEB site.
My PERSONAL opinion (and vote) is that SGI should not take your test suit as a
valid comparison of SGI versus other IGs since is based on OpenGL GVS
implementation instead of Performer. Beside this fact your test suit is a very
interesting reference as an example of OpenGVS application perfomance on
OpenGL that could be in some way be extrapolated to a well written OpenGL
implementation in VisSim databases.
Another issue is the support for SGI specific features for iR that are only
available using performer, some of this features help a lot in order to fully
use the "fast path" of our IGs.
>...
>
>
>-- End of excerpt from John Archdeacon
-- ************************************************************************* * Javier Castellar Arribas * Email: javier++at++asd.sgi.com * * * Vmail: 3-1589 * * Member of Technical Staff * Phone: (415)-933-1589 | 933-2108 (lab) * * Applied Engineering * Fax: (415)-964-8671 * * Advanced Systems Division * MailStop: 8L-800 * ************************************************************************* * Silicon Graphics Inc. * * 2011 N. Shoreline Boulevard, * * Mountain View, California 94043-1386, USA * ************************************************************************* "Violence is the last refuge of the incompetent" Hari Seldon
This archive was generated by hypermail 2.0b2 on Mon Aug 10 1998 - 17:52:47 PDT