Such a slam was rather unprofessional. Benchmarking is a complex
issue. I think Mr. Flavin should have pointed out issues about what the
Tornado Labs benchmark measures rather than trying to trash their
reputation. I've written plenty of benchmarks that taken out of context
would be "garbage" but the fact was they measured something I wanted to
measure and gave me useful numbers because I knew what I was measuring.
I have not looked at the Bench J3D code (so I don't know what it is
measuring). It could do Java 3D a disservice if others quote these
figures out of context to make it appear that Java 3D is slow. But I
think Mr. Selman's point was to give us some numbers to compare various
hardware configurations (and I thank him for that).
The frame rates of Bench J3D mean very little about overall real
application performance of Java 3D. Our application varies from 1 fps
(non accelerated video hardware) to 30 fps (low end machine) to 150 fps
(Athelon & GeForce 256 video) to 275+ fps (artificially configured to
yield incredible numbers). What this tells us is that Java 3D can do
some incredible real world things - how much it can do is very much
dependent upon the hardware you run it with.
- John Wright
Starfire Research
Evan Drumwright wrote:
>
> What's up with this? So Bench J3D may not be a good test in terms of real
> world performance. I believe the goal of it is to compare the effect of
> hardware differences on frame rate, and I think for that it does its job.
> In any case, if you are so dissatisfied with it why don't you write a better
> benchmark instead of wasting your time and ours complaining about this one?
>
> Evan Drumwright
>
> -----Original Message-----
> From: P. Flavin <[EMAIL PROTECTED]>
> To: [EMAIL PROTECTED] <[EMAIL PROTECTED]>
> Date: Monday, April 17, 2000 5:52 AM
> Subject: [JAVA3D] Benchmark Stupidty from Tornado Labs : FAST FACET FACTS
>
> > A _ POORLY DESIGNED _ Java3D Benchmark by Torando Labs.
> >
> > For _ FAST _ graphics render _ FLAT _ surfaces.
> >
> > Smart people know it, the Tornado benchmark doesn't show it.
> >
> > The dopey Tornado test only uses scenes with spheres.
> >
> > No test of what we care about : optimized, flat surfaces.
> >
> > __ You __ CAN __ GET __ GOOD __ RESULTS __
> >
> >| We are getting _ very _ satisfactory _ performance using the
> >| [ Java3d ] VRML viewer.
> >|
> >| we are getting __ 50 K+ __ polygons at realtime speed ( 25-30 fps).
> >| This is with little to no optimization except good training of modelers
> >...
> >| [ PIII 733 with a GeForce ]
>
> ... rest of nonsense deleted
>
> ===========================================================================
> To unsubscribe, send email to [EMAIL PROTECTED] and include in the body
> of the message "signoff JAVA3D-INTEREST". For general help, send email to
> [EMAIL PROTECTED] and include in the body of the message "help".
===========================================================================
To unsubscribe, send email to [EMAIL PROTECTED] and include in the body
of the message "signoff JAVA3D-INTEREST". For general help, send email to
[EMAIL PROTECTED] and include in the body of the message "help".
