I'd be irritated if the only difference is that they produce less heat and use less energy, for example.
Less heat and Less energy could be substantial. Extends overclocking abilities if they produce less heat and makes them profitable for longer if they use less energy.
I'm of two minds about that. The hacker in me loves the idea of tinkering and seeing how far we can push the hardware. Extra speed, whether from lower heat and overclocking, or from better internals, is always desired. On the other hand, the businessman in me doesn't care how payback is improved, so long as it *is* improved. What I meant by my comment was that, say, a 5% improvement in hash rate, whether from overclocking or from intrinsically better chips, is mostly useless. A 500% improvement would have us dancing with joy. Anything inbetween would have to be evaluated. Same for cost, heat, etc.
At 55nm I hope we see ~1GH/chip. If avalon comes through at 1 GH/chip we could potentially see ROI - assuming avalon offers a 1:1 trade.