• Open Menu Close Menu
  • Apple
  • Shopping Bag
  • Apple
  • Mac
  • iPad
  • iPhone
  • Watch
  • TV
  • Music
  • Support
  • Search apple.com
  • Shopping Bag

Lists

Open Menu Close Menu
  • Terms and Conditions
  • Lists hosted on this site
  • Email the Postmaster
  • Tips for posting to public mailing lists
Re: OpenGL and Tiger
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: OpenGL and Tiger


  • Subject: Re: OpenGL and Tiger
  • From: Shaun Wexler <email@hidden>
  • Date: Mon, 16 May 2005 23:40:54 -0700

On May 16, 2005, at 8:12 AM, Scott Ahten wrote:

I'm working on a video mixer application which uses the Core Video display link thread to render Quicktime movie frames as OpenGL textures. This problem doesn't seem to be effecting my application.

I can set a prefs value and MacFOH will utilize CVDisplayLink instead of SKWDisplay VBL sync.  The results of profiling using the same setup and realtime analyzer document are:

    CVDisplayLink
    Tiger 10.4.1 
    CGS Beam Sync enabled
    4.93% OpenGL
    glFlush 410 µS
    
    SKWDisplay
    Tiger 10.4.1 
    CGS Beam Sync enabled
    3.88% OpenGL
    glFlush 256 µS

    CVDisplayLink
    Tiger 10.4.1 
    CGS Beam Sync disabled
    3.00% OpenGL
    glFlush 128 µS

    SKWDisplay
    Tiger 10.4.1 
    CGS Beam Sync disabled
    2.87% OpenGL
    glFlush 107 µS

    SKWDisplay
    Panther 10.3.9
    1.32% OpenGL
    glFlush 62 µS

Are you saying that auto-beam-sync doesn't work as advertised

So far, it does not improve performance, in fact it is quite the opposite.  Its implementation needs to be rethought.

or you think that blocking in flush() when drawing faster than the screen refresh rate is a "Bad Idea"?

I think introducing blocking to anything (especially something that didn't previously block) is generally a bad idea, unless it's a specific API designed to block.  Completion callbacks would be better, and allow developers to elect their use.

I designed an API for rendering management that is superior IMO, and it would be great to incorporate it into OpenGL/CGS/CV so everything would enjoy the same high performance and low latency as MacFOH (under Panther).  Until these glaring Tiger/CGS bugs are resolved, I'm staying with Panther, and will continue to recommend that my users resist upgrading.  Dual-monitor window dragging on Tiger is currently so pathetic that I can't even read the window contents because it jitters so badly when windows cross the screen boundary.

Previous seeds in the range 393-414 provided substantially better performance, then it was broken again.

-- 

Shaun Wexler

MacFOH

http://www.macfoh.com


"Intellectuals solve problems, geniuses prevent them." - Albert Einstein



 _______________________________________________
Do not post admin requests to the list. They will be ignored.
Cocoa-dev mailing list      (email@hidden)
Help/Unsubscribe/Update your Subscription:

This email sent to email@hidden

References: 
 >Re: OpenGL and Tiger (From: Lorenzo <email@hidden>)
 >Re: OpenGL and Tiger (From: Shaun Wexler <email@hidden>)
 >Re: OpenGL and Tiger (From: Steve Christensen <email@hidden>)
 >Re: OpenGL and Tiger (From: Shaun Wexler <email@hidden>)
 >Re: OpenGL and Tiger (From: Scott Thompson <email@hidden>)
 >Re: OpenGL and Tiger (From: Shaun Wexler <email@hidden>)
 >Re: OpenGL and Tiger (From: Scott Ahten <email@hidden>)

  • Prev by Date: Building with XCode 2 for Jag and Panther
  • Next by Date: Re: Building with XCode 2 for Jag and Panther
  • Previous by thread: Re: OpenGL and Tiger
  • Next by thread: Re: OpenGL and Tiger
  • Index(es):
    • Date
    • Thread