Re: Multi-Channel Option(MCO) problem

New Message Reply Date view Thread view Subject view Author view

Tom Fuke (tom++at++war.reading.sgi.com)
Thu, 17 Oct 1996 12:52:52 +0100

On Oct 17, 11:52, Mario Veraart wrote:
> Subject: Re: Multi-Channel Option(MCO) problem
> > This is Ren-Jye Yu. I have some problems when I use Multi-Channel
> > Option(MCO). I use Performer to write the code and use MCO to show the
scene in
> > three different monitors. When I start the MCO, the average video
frame rate
> > will drop down seriously. At least 70% below running the same
Performer code on
> > the default monitor. Why does the MCO destroy the refresh rate time?
 The MCO
> > is running three channels (2 ++at++ 640x480 and 1 ++at++ 1280x1024.) We have a
reality
> > engine 2, R4400 processor, 256 Mb RAM, and 2 video raster boards.

Assuming your non-MCO "default" case is a single channel at 1280x1024, it
should not surprise you to see a performance drop when you add two more
640x480 channels. The pipeline not only has to process the extra geometry
in the two new channels (Geometry Engines), but also deal with filling
more pixels (Raster Managers). Your graphics application is likely to be
either "geometry" OR "pixel-fill" limited at any one time (geometry
operations and pixel-fill operations can work in parallel due to beefy
FIFOs in the graphics pipe). I think you are seeing pixel-fill limits
(especially at those resolutions with 2RMs)...

Pixel fill rate is a function of Depth Complexity and Total Pixels, i.e.
if you add two more 640x480 channels at the same DC (same scene in each)
but increase the TP by 50% then you will see a performance drop in a
pixel-fill limited scene by about 33%. To reduce pixel-fill limitations
you either reduce Depth Complexity or Total Pixels.

Your MCO is not impeding the performance in itself - the MCO is purely a
digital video-splitter that works at the back-end of the pipeline (after
geometry/pixel processing) - it is the increase in Total Pixels/Geometry
that causes a hit.

tom

---



-- 
--------------------------------------------------------------
Tom Fuke                  email: tom++at++reading.sgi.com         -
The Reality Centre        vmail: 59101 (0800 896020 in U.K)  -
Silicon Graphics Ltd      vox: +44 (1734) 257604             -
Theale, Reading           fax: +44 (1734) 257608    \   o  _
RG7 4SB, UK               mailstop: IUK-311         | [][] |
                                                    | (  ) |
--------------------------------------------------------------
Silicon Surf:   www.sgi.com
Reality Centre: www.sgi.com/International/UK/centre/index.html
                (use www-europe.sgi.com for European mirror)
Tom:            reality.sgi.com/employees/tom_reading/ 
--------------------------------------------------------------
=======================================================================
List Archives, FAQ, FTP:  http://www.sgi.com/Technology/Performer/
            Submissions:  info-performer++at++sgi.com
        Admin. requests:  info-performer-request++at++sgi.com















New Message
Reply
Date view
Thread view
Subject view
Author view











This archive was generated by hypermail 2.0b2 
on Mon Aug 10 1998 - 17:53:46 PDT





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