Dynamic redundancy allocation for reliable and high-performance nanocomputing.