Re: [info-performer] Texture/fill-rate performance

New Message Reply Date view Thread view Subject view Author view

From: Marcin Romaszewicz (marcin++at++sgi.com)
Date: 07/18/2002 11:26:10


On Wed, 17 Jul 2002, Krajcarski, Rob wrote:

> Hello,
>
> I'm trying to improve the performance of an application that is being run on
> an InfiniteReality Onyx machine. Right now the current bottleneck seems to
> be the fill rate, at least thats what the frame stats indicate (its a stereo
> application, 2xVGA resolution, draw times are about 5ms each channel, with
> the fill time looking like 12ms).
>
> I've made sure that I've done the steps that I could find for improving fill
> performance, like clearing the buffer right away. Now the next thing I
> tried was to look at expensive fill modes (pfAntialias/pfTransparency).
> What I was suprised to see was that the fill time completely disappeared
> when I got rid of textures. Now I was under the impression that texturing
> was 'free' on InifiniteReality machines....so then I got confused..
>

Texturing on IR is practically free for 16 bit internal formats. 24 and 32
bit formats have a hit, though I don't know how much of a hit it is. See
if setting a 16 bit format on your pfTextures helps.

> My question is really what factors/situations does texturing lead degraded
> fill performance. Any help on this matter would be greatly appreciated..
>
> Thanks
>
> Rob Krajcarski
>
> -----------------------------------------------------------------------
> SIGGRAPH 2002 Friends of Performer Meeting July 24 6:00pm-8:00pm
> Salon I, Marriott Rivercenter Hotel, San Antonio, TX USA
> -----------------------------------------------------------------------
> List Archives, Info, FAQ: http://www.sgi.com/software/performer/
> Open Development Project: http://oss.sgi.com/projects/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 2b29 : Thu Jul 18 2002 - 11:26:17 PDT

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