> At 06:37 PM 2/15/99 -0800, Raab, Andreas wrote:
> >ObSqueak: I just got about 20,000 triangles per second on a P166 with
> >Balloon 3D ;^)
> 
> Can you say more about the test conditions. Are these texture mapped,
> z-buffered, perspective correct, 3D triangles?
> 
No, yes, no ;-) The entire test was just to see if the primitive level code
is working as expected. No performance measures intended.

> To define a bit of 3-D performance perspective, the low cost ($150) 3D
> accelerator (NVidia Rage 128) in the PC I'm on is rated at about 5
> million,
> texture mapped, z-buffered, perspective correct, 3D triangles per second.
> 
These are *theoretical* throughputs. You'll never get them in practice.
Usually, it's quite a good idea to divide these numbers by ten and take this
value for average applications (but 500k is still a lot more than 20k).

> I'm not sure the processor (Pentium II-266) can come up with 15 million
> vertex values/sec though, so actual milage may vary.
> 
BTW, there ain't no need for the processor to transfer 15 million vertices,
12 bytes each. Given indexed triangle meshes you may have something between
3-5 million vertices plus 15M 32bit words.

  Andreas
--
PLEASE NOTE NEW EMAIL ADRESS!

+===== Andreas Raab ========= (andreasr at wdi.disney.com) ==+
| Walt Disney Imagineering        Phone: +1 818 544 5016  I
I Glendale, CA                    Fax:   +1 818 544 4544  I
+======< http://isgwww.cs.uni-magdeburg.de/~raab >========+