***** Smartbench 2010 Benchmark Thread *****

HTC Desire on LeeDroid 2.3d @ 1190Mhz

screenshot.png
 
simulatorman said:
I read this at the start of the thread:

Is it and how?
Click to expand...

I'm not sure if Smartbench is better optimized but if there is one thing I was trying to get right from the get go was to get the weightings of CPU, GPU and I/O performances realistic in comparison to the real applications we use every day. This is another reason why I ended up with 2 indexes - one for productivity users (who benefit more from faster CPUs) and gaming users (who benefit more from raw GPU power).

Have a look at the Quadrant Advanced chart for any phones. And look at the length of each sub-bars. You will find that on some phones, a single bar totally dominates, especially in the area of I/O. If your app relies on I/O 70% of the time, then this is accurate but typical apps don't have this kind of usage pattern. Quadrant weighting made sense back when HTC Magic was the norm, but right now, its a bit unrealistic. Smartbench 2010 uses Nexus One as the baseline rather than HTC Magic. Smartbench 2011 (to be released in few months) will use higher spec phone as the baseline, once again to remove any potential weighting issues.

EDITED: Meant to say "Smartbench 2011"
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom