It's absolutely wrong, though. There are plenty of factors to consider in how the programs you're comparing are written, and this is far from a fair comparison since these are 2 entirely different versions of one emulator. If you took a platform-independent version and compiled it for the GP32 and 2x with as many factors the same as possible (e.g. compilers) you could do a more fair comparison.
At any rate, it's just simple logic code runs faster on a 200 MHz ARM9 than a 133 MHz one, unless some sort of other bottleneck is introduced, which at this point, there does NOT appear to be. And I'll bet if you ran 2xMark on the GP32 you'd see quite the linear performance curve from 133->200 MHz.