Counterfit said:
I wonder if the guy at Barefeats remembered to set the CPU to Highest. I don't think it would have such a large impact on the scores, but you never know...
Did you read his reviews? It's obvious he's doing a thorough job.
No, he didn't forget to do the CPU setting. This tester is VERY meticulous, and has done several passes and multiple updates and a rigorous set of comparison machines.
He's even gotten some communication with apple's performance marketing or some such division.
Remaining variables (that he knows) that could affect the game benchmarks (from
http://www.apple.com/imac/graphics.html ) are:
* RAM-- not only might the g5 experience more FPS benefits than the G4 above 1G (where G4s tend to level off in gaming gains), but matched-dimms in the G5 allow for a 128bit path.
* patch version of ut2004-- currently at 3323, which was not yet released at the time of the imac release-- also the original retail patch was very unoptimized, to the extent that audio alone took up to 10% of the CPU load.
*graphics detail settings of the game (Apple may have used lowest)
*resolution for game: 17" LCD native of 1440x900, or a non-native standard resolution such as 1024x768. Word is that apple and barefeats (rightfully) used the LCD native version of the benchmark.
To do your own UT2004, get the
Santaduck benchmark and do it yourself, if you have both a g4 & g5 17" imac with different dimms of RAM laying about.