Re: Antialiasing Problem on iR

New Message Reply Date view Thread view Subject view Author view

Hill_Brian++at++amelnx.advmar.com
Thu, 08 Aug 96 08:36:38 EST

     Shawn,
      
     I just went through digging up information on the IR and here is what
     I found out.
      
     Your frame buffer configuration will be dependent on your color
     representation, multisamples, zbuffer depth, and stencil depth. Here
     is an example of how to figure it out:
      
     Front Buffer 10 Bit RGB 30
     Back Buffer 10 Bit RGB 30
     Pixel IDs... 32 Bit 32
      
     Multisample Buffer
        Color 8 Bit RGB 96
        Z-buffer 23 Bit 92
        Stencil 8 Bit 32
     =======================================
               Total Bits per pixel 312
      
     Pixel sizes supported are small 256 bits, medium 512 bits, and large
     1024 bits. This configuration would require a medium depth of 512
     bits.
      
     2 RMs have 160 Mbytes frame buffer memory.
      
     The above configuration will require:
      
     (3 chans * 1280 * 1024 * 512 bits/pixel)
     ---------------------------------------- = 240 Mbytes
         (8*1024*1024 bits/Mbyte)
      
     Which won't fit within the available 160 Mbyte frame buffer.
      
     You also have to check the DG throughput which is:
      
        12 bit RGB 221 Mpix/sec.
        10 bit RGBA 221 Mpix/sec
        10 bit RGB 300 Mpix/sec
      
     Your configuration is:
      
     3 chans * 1280 * 1024 * 60 Hz = 236 Mpix
      
     Which works for the 10 bit RGB color which was used in the frame
     buffer calc too.
      
     So your problem is frame buffer size. If you are using a rack you can
     increase the number of RMs to 4 and get 320 Mbytes frame buffer
     memory.
      
     I hope this is correct and that it shines some light on the situation.
     I still don't know the details about the RGB8 representation in the
     multisample buffer (why not RGB10 or RGBA12) or the DG throughput with
     RGBA12. There was just some traffic on this list on the 23 bit Zbuffer
     which explains it pretty well.
      
     Any further insight/clarification would be appreciated.
      
     Brian Hill
     hill_brian++at++advmar.com
      
      
      
      
______________________________ Reply Separator _________________________________
Subject: Antialiasing Problem on iR
Author: "Shawn Soeder" <shawn_soeder++at++esplt14m-s.grumman.com> at Internet
Date: 8/7/96 02:46 PM
      
      
Performers,
      
I am having difficulty with antialiasing on a Performer app on iR.
      
I have an Onyx/iR with (2) RM6s and Performer 2.1. When configured with (3)
1280x1024 output channels, the irconfig utility tells me the pixel depth is
"small". My understanding was that this would allow (4) subsamples, and this
seems to be verified by the pfQueryFeature(PFQFTR_MULTISAMPLE) which returns
PFQFTR_FAST.
      
However, running this app shows substantial aliasing, and no perceptable
difference between antialias on/off.
      
What am I doing wrong?
      
Any info would be appreciated.
Thanks,
Shawn
      
=======================================================================
List Archives, FAQ, FTP: http://www.sgi.com/Technology/Performer/
            Submissions: info-performer++at++sgi.com
        Admin. requests: info-performer-request++at++sgi.com

=======================================================================
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 - 17:53:19 PDT

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