Multiple isect passes per frame

New Message Reply Date view Thread view Subject view Author view

Scott McMillan (scott++at++er)
Mon, 6 May 1996 13:46:25 -0400 (EDT)


I am interested to see if anyone else is doing anything unusual with
intersection processes...

We are performing dynamic simulations and using Performer to visualize the
results. The dynamics/accelerations must be computed and numerically
integrated many times per second (600 or more is not unusual which translates
to about 40 or more per frame). To compute the dynamics each time, we need
contact forces that are a function many Performer intersection tests.

We have to perform intersection tests more than one time a frame because the
pfSegs are updated each time through the dynamics computation based on the
results of the previous step. We have a 4 processor Onyx so we thought about
forking the isect off, but it is my belief that using PFMP_FORK_ISECT would
not work because that would only allow me to update the pfSegs once per
frame. Can someone verify this?

Anyway, the way we get around this is to sproc a process off that
continually does intersection testing with a set of pfSegs that are updated
by the APP. It works kind of like a ring buffer. The APP extracts
intersection results from the previous dynamics cycle and stuffs in a new
pfSeg for the next cycle, and our ISECT process cycles through the same list
looking for new pfSegs to test against the scene (note that the part of the
scene graph that intersections are tested against is static).

I was just wondering if anyone else has come up against this limitation with
PFMP_FORK_ISECT? If so, has any other novel solutions been found? OR do I
underestimate what I am able to do by using PFMP_FORK_ISECT?

scott

-- 
Scott McMillan / scott++at++ht.com / (301)984-3706 x250 / FAX (301)984-2104
                      High Techsplanations Inc. 
       6001 Montrose Road, Suite 902 / Rockville, MD 20852-4874 

New Message Reply Date view Thread view Subject view Author view

This archive was generated by hypermail 2.0b2 on Mon Aug 10 1998 - 17:52:52 PDT

This message has been cleansed for anti-spam protection. Replace '++at++' in any mail addresses with the '@' symbol.