Originally Posted by slaWter
I doubt it. It would be interesting to know for sure. I can imagine that they don't even run a real world test with a normal operating system for those measurements. The statement is coming from the hardware team after all.
The 20% improvement is the best case so that means it can't be when running at full load with more cores.
Well let's see:
We have apps for android already that run quad cores. These are already guaranteed to see battery savings over juggling all four threads in two cores, which make the processors run to the point that they lose a lot of that energy as heat.
On Android, apps have the option of running as services, such as some widgets (most lie dormant, but some things that update frequently e.g. clocks,) I myself use the samba sharing app which runs as a service, and there are many others. Each of these run in their own threads, which can operate at their own independent speeds. Again, more battery savings.