From Stephen Wilson
While engineers continue to be surprised by the endurance of Moore’s Law, which says that computing power doubles roughly every 18 months, we can be certain that Silas Beane’s descendants will never get to simulate the universe on a lattice resolution of 10-27 metres (15 December 2012, p 33).
Today’s state-of-the-art memory chips use transistors a little larger than 10 nanometres (10-9 metres) across. When Beane extrapolates Moore’s Law over 500 years, he is implying transistors will continue to shrink to impractical sub-nuclear, quantum-scale dimensions.
But even if that engineering fantasy were to come true, it would defeat the purpose. The envisaged grid-based simulation involves calculating how physical properties change from one instant to another at every point in the model. You need physical memory to store the data. The circuits used to hold all the data for a universe whose points are 10-27 metres apart could not themselves fit into the universe being modelled.
Sydney, Australia
