They're putting a 10 core SoC in there? If so, there's no chance any of the cores will be able to ramp up for more than a femtosecond or two, unless they're going to do some significant modifications to the case.
I wonder if there's a way to work out the optimal number of cores for a given efficiency and cooling rate. Obviously, for quite a lot of algorithms still, single core performance is king, but for a theoretical process than can split evenly over all available computing resources, without doing any maths, I reckon it'll come out even - you get the same performance however many ways you split it, but in practice CPUs aren't equally efficient at all clocks, and even well behaved processes have a limit to how many ways they can be split.