i've run loads today, and i found that the first one will always be a little lower, with subsequent tests being in the higher ranges... perhaps this is the ondemand govenor holding off initially but then kicking in as the benchmark adds load...?? i'll run some now...
Just a note, the FPS is capped at 60, attributing to lower results on some, especially for the Game Index on Smartbench...
i used the preinstalled task manager to clear the memory and close any open apps before i open each benchmark utility, after a single clean boot.
1st: PI 3732 - GI 2194
2nd: PI 3957 - GI 2241
3rd: PI 3920 - GI 2329
4th: PI 3933 - GI 2310
5th: PI 3966 - GI 2344
MANDRO Bench (quick)
MANDRO Bench (comprehensive) - only did 3 as i started getting bored... sorry lol
I'd really be interested in seeing what a difference uncapped FPS would make to the results... Also, if Sammy cap the Tegra version, we should get a really good comparison chip-on-chip!!!
I hope this is of some use, personally for me, i don't believe its the most important factor, it's the whole experience (which i'm loving!!), but i find it interesting for comparison after rooting/modding/flashing etc...