Re: RM4s faster than RM5s???

New Message Reply Date view Thread view Subject view Author view

Allan Schaffer (allan)
Tue, 11 Feb 1997 18:04:11 -0800

On Feb 11, 4:42pm, Daniel Stipe wrote:
> One system is faster than the other.
> Both systems are RE2s with 512 Mb of RAM.
>
> The fast system has 4 x 200MHz processors.
> The slow system has 4 x 250MHz processors.
>
> The fast system has 2 RM4s.
> The slow system has 4 RM5s.
>
> I believe both systems have comperable swap space (I'll confirm this).
>
> The difference in speed appears to happen during the draw cycle.
>
> Are there default settings for the RM5 that would impact speed?
> Texture paging perhaps? - I'm loading a 4k x 4k x 1byte image (800%
> of 4Mb). If the RM5 pages in larger chunks that might slow things
> down (particularly to 4 boards?).

I doubt the difference is in texture paging speed, they should match.

What about pixel depth? /usr/gfx/gfxinfo will tell you the current
pixel depth for each. For a given video format you may be getting
small pixels with 2 RM's and medium pixels with 4. Multisample
anti-aliasing might be turned on automagically with the medium pixels
-- this would slow down your fill rate by a few tens %.

Easy to test: pull out 2 RMs from the slow system and see if it
speeds up. Or, run a bigger video format on the slow system (one
that results in the same pixel depth as the fast system). Or
explicitly specify the same X visual config as on the fast system.

Allan

-- 
Allan Schaffer                                                allan++at++sgi.com
Silicon Graphics                               http://reality.sgi.com/allan
=======================================================================
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 - 17:54:38 PDT

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