Tests - Intro and Cinebench 2003 and 9.5
Where possible, all of our tests were run under Mac OS X and under Windows XP running in two ways - from its own partition created by Apple's Boot Camp Beta applet and also in a virtual machine running under OS X and created by a beta of Parallels Desktop for Mac.
We carried out all tests at least three times and show the average of those results. Each test was carried out after a reboot and without running any other tasks beforehand.
Cinebench 2003
The first results are for rendering with Maxon's Cinebench 2003 but these are probably the least significant of all our tests because Cinebench doesn't run natively under Mac OS X on Intel CPUs - causing the performance under Mac OS X to be especially poor.
You'll notice that Cinebench 2003 running in XP on the Parallels virtual machine shows no dual-core CPU results. That's because the virtual machine was emulating only a single-core CPU. We're mightily confused by the single-core test under Parallels, since the performance was clearly inferior to that of XP running in its own partition - something that appears not to be reflected by a comparison of the two XP single-core results.
Our take is simple - read nothing at all into any of the Cinebench 2003 results. About the only value they may have is when comparing the results of the own-partition XP scores with those of conventional Windows XP PCs or other Intel Macs running Boot Camp.
Cinebench 9.5
In contrast, some of the results obtained using the latest version of Cinebench, V9.5, which is available to run natively under OS X on Intel Macs, might be worth thinking about, though, again, the results under Parallels - single-core only - make no sense to us and don't chime in with the poor performance we saw compared with the program running under XP in its own partition or, indeed under OS X.
When you look at the results in XP running in its own partition and compare them with the Mac OS X results, what you can see is that the performance of the two is now very similar, though with Mac OS X just having the edge. This further underlines why it makes sense to ignore any Mac OS X results of Cinebench 2003.
