Re: [Update REQ 6112]: Geomview solaris -mp?

[L.Wood at surrey.ac.uk]

To: software@geom.umn.edu
Subject: Re:  [Closed REQ 6112]: Geomview solaris -mp?
Message-ID: <Pine.SOL.3.95.970218190541.11931C-100000 at petra.ee.surrey.ac.uk>

On Tue, 18 Feb 1997 daemon at geom.umn.edu (Stuart Levy) wrote:

> In SaVi's case, it appears that there's a substantially more efficient way to
> represent (in geomview) the zillions of pointlike satellites.
> Graphics speed increases by a factor of 2 on a Sparc-10 here when
> I make a mocked-up streamlined version.  I've mentioned this to Patrick;
> maybe he can incorporate this optimization at some point.

This I'd be very interested in, since console on a Sparc20 with
standard video card is the best I can hope for in terms of
CPU/graphics speed here. I'm looking at doing lots of scripting for
moving camera viewpoints, since realtime mouse control of
1500+-satellite constellations is difficult.


> So if Sun does come out with an OpenGL library for the Sparc20,
> and you have some hardware graphics accelerator on it, then
> compiling it there might help.

Yup, OpenGL would be a good excuse to get a better-than-minimum
graphics card in.


> By the way, are you using the "geomview -opengl" version or the plain one?

The plain one; Sun doesn't support OpenGL on the Sparc20 yet AFAIK.


> geomview -opengl will do texture mapping, but it's about 2x slower than
> the non-texture-mapping version.  And I think SaVi now includes two earth
> models: one texture-mapped one, and another that includes just
> continent outlines which can use either version of geomview.

I've got the continent outlines without OpenGL now that I have
geomview 1.61, although I still need to tweak SaVi a little to balance
the continent visibility against the footprint circumferences.

Thanks,

L.

<URL:http://www.sat-net.com/L.Wood/><L.Wood at ieee.org>+44-1483-300800x3435