paper and slides It looks like we have achieved crossover, where posit hardware is getting sufficiently optimized to expose how much more efficient it is than ultra-optimized IEEE 754 float hardware.