I'd agree with that -- I categorize SuperPi like most other synthetic benches, in that they make sense when trying to compare the exact same system between "tweakings" as a way to measure how much progress you've made between point A: and point B:. Example: here's my SuperPi result at 3Ghz / 1333 FSB / ram at DDR2-667, and here's my SuperPi result at 4.5Ghz / 2000FSB / DDR2-1000.
Comparing SuperPi results between seperate systems is iffy at best, and almost worthless when moving across architecture changes, memory subsystem changes, chipset changes, and even vendor changes.
I too would rather see something more real-world.
Comparing SuperPi results between seperate systems is iffy at best, and almost worthless when moving across architecture changes, memory subsystem changes, chipset changes, and even vendor changes.
I too would rather see something more real-world.