Is this good?(1.6Ghz becnhmarks)

wireless

Well I just looked at the bench marks for the 1.6GHz model at http://www.chaosmint.com/powermac-g5-16/



and it says Xbench is now ready to work on the G5. Then it posts all these numbers. And I have NO clue how to read those, so guys help me understand.

Look:







CPU Test\t 118.69 \t126.12\t

GCD Recursion\t73.95 2.89 Mops/sec\t82.57\t3.22 Mops/sec

Floating Point Basic\t198.78 706.93 Mflop/sec\t207.84\t700.71 Mflop/sec

AltiVec Basic\t78.23 1.14 Gflop/sec\t95.27\t5.17 Gflop/sec

vecLib FFT\t143.08 2.20 Gflop/sec

Floating Point Library\t263.18 10.53 Mops/sec\t232.79\t10.45 Mops/sec



The first set of numbers is the Xbench ver 1.1 the 2nd set of numbers is the one doesn't even know its a G5 proc (ver 1.0). Is it good to have lower numbers?

zapchud

It just proves that Xbench sucks as a CPU-benchmarking tool, between architectures. Xbench is specially tuned for G4 execution, and since the author do not have access to a G5, it will stay so, even when recompiled. Also, the code Xbench benches is far too trivial to reflect anything about real-world-application performance.



Anyways, there are a ton of threads on this subject already, I just posted to get the first reply!

markiv

Xbench some how has unoptimized there benchmarking software for both G4's and G5's. There has been a big drop in the evaluation result of Altivec's performance. You can check out the benchmarks performed on various G4 machines and G5 1.6 GHZ and dual 2.0. The results are a sham, the results have been posted on macrumors website. Hope this helps Xbench Results

locomotive

Before anyone jumps to hasty conclusions from benchmarks, perhaps we need to learn from our sunstroked brethren in Florida. We should test, tweak, and test again until we achieve some results that we can all be proud of!