[G4] G5 speed tests

[Joseph B. Gurman]

     Sandor Ferenczy wrote:

>*sigh*
>
>1) non of this software was optimized for the dual 2 gHz G5
>
>2) we have no idea how the processor slewing*** was set in the control 
>panel, therefore it should probably be assumed that it was at the 
>default, which is "automatic" and that setting has already been shown 
>to produce far lower benchmark numbers than the fastest/highest setting.
>
>***slewing -->
>http://developer.apple.com/documentation/Hardware/Developer_Notes/
>Macintosh_CPUs-G5/PowerMacG5/1Introduction/chapter_2_section_4.html

     I would give BareFeats a break on (2), since the G5 hardware 
developer note states that slewing is achieved "by a gradual 
transition that does not impact system or application performance..." 
A similar statement was made by the G5 product manager in the WWDC G5 
G5 architecture presentation; he said it was done to keep the fans 
barely turning when machines were idling, or you were just reading 
mail, but shouldn't be visible in performance.

      (1) I suspect is correct, and in many applications, will be 
improved considerably when Panther is a 64-bit OS and the 
applications developers have had time to implement 64-bit addressing. 
Unfortunately, my guess is that's probably at least a year away. In 
the meantime, there are certainly tweaks that can be made, and 
hopefully, developers will follow Adobe's lead and do so soon. (Note 
that the Photoshop results for the dual 2 GHz G5 were obtained before 
the G5 plug-in was available, so I bet the dual 2.0 beats the Xeon 
2.4 in the "MP-aware" Photoshop test when they re-run it.)
-- 
"I love deadlines. I love the whooshing sound they make as they go by."
                                                             - Douglas 
Adams, 1952 - 2001

Joseph B. Gurman, NASA Goddard Space Flight Center, Solar Physics
Branch, Greenbelt MD 20771 USA