Simon Mills (simon++at++wgs.estec.esa.nl)
Thu, 15 Jul 1999 09:35:25 +0200
Actually I've been using 1024x768_96s quadbuffer stereo happlily on an
Octance MXE also.
> Here i encountered some drawbacks:
>
> My first idea was pretty handy, we have two channels, so why
> not using one for my desktop and one to render my gfx to
> a Marquee Beamer...Ok, wanted to have 2x1280x1024_96s on
> a 2560x1024 FB, but this didn't work. Next t tried
> to configure channel 1 to display 1024x768_96s and channel 2
> the same resolution but at a different, non overlapping portion
> of the framebuffer (again, channel 1 goes to beamer,
> channel 2 to the 24" CRT). Using quadbuffer stereo, only
> monoscopic was rendered...
> But no error or warning when defining the display with ircombine
> occured, of course, because it was not to be foreseen that
> quadbuffered was to be used.
To give you an idea what's possible we're successfully driving a
Electrohome Marquee 7500 and monitor with a single 1280x1024_96s stereo
ouput from one channel fed through a video splitter. For our application
we just wanted the same output of both projector/monitor. Rendering 2
channels in stereo would have been prohibitively slow. We have a iR2
with 2 RM's. BTW, to do this I had to ditch the very nice 24" SGI
monitor, which couldn't cope with this video format (I guess because of
the fact that it's very wide), and use another monitor with higher max
horizontal freq (a Philips Brilliance 201B).
> The problem is that with ircombine, the amount of FB used for
> quadbuffered is not shown...,am i right with that?
> Does that mean if i want to use quadbuffer stereo, i have to
> reserve the doubled amount of what ircombine reserves for the
> channels on the FB?
>
> If that is the fact,
> how can i find a clue to determine the valid framebuffer-configurations?
> And how can i determine the FB size at all? (and not with try and error)
> It seems that the "extra-buffer" (for quad) is reserved from
> the framebuffer when possible, if not, well it just doesnt
> work.
>
> In fact, i really want 1280x1024_96s on both channels, and
> on different portions, and as quadbuffer. 1280x1024_96s (ok, half of it)
> was even possible with my old octane, so my expectations
> with an IR are much higher. Another point is, antialias (multisampling)
> doesn't seem to work either even when i use much lower resolutions.
> I tried it with 640x480_96s and the attributes were:
>
> fbAttrs[] = {
> PFFB_RGBA,
> PFFB_DOUBLEBUFFER,
> PFFB_STEREO,
> // PFFB_DEPTH_SIZE, 24,
I use PFFB_DEPTH_SIZE, 23 because of the iR's rather strange looking
23bit Z-buffer. Works on other systems too because you ask for at least
this amount.
> PFFB_RED_SIZE, 1,
> PFFB_STENCIL_SIZE, 1,
> PFFB_SAMPLES, 1,
If I'm on a multisampling machine (iR, RE2) I request PFFB_SAMPLES, 4.
> None,
> };
>
> That should be ok, am i right? But the output didn't show any
> effect on that. So, if u can give me a hint on any of
> these topics, an IR-newbe will be glad about it ;)
> I think i need some enlightement...
This works for me, although I have 2 RM's. You can also try the
"findvis" program to see which X visual's you have available and then
explicitly selecting one of those visuals like the perfly -t <visual_id>
option.
Hope it helps,
Simon
________________________________________________________________________
Simon C. Mills
Silicon Worlds S.A.
c/o Modelling & Simulation Section (TOS-EMM) Tel: +31 (0)71 565 3725
European Space Agency (ESA/ESTEC) Fax: +31 (0)71 565 5419
Postbus 299, 2200AG Noordwijk e-mail: simon++at++wgs.estec.esa.nl
The Netherlands http://www.estec.esa.nl/wmwww/EMM
________________________________________________________________________
This archive was generated by hypermail 2.0b2 on Thu Jul 15 1999 - 00:36:12 PDT