Optimal Speedup on a Low-Degree Multi-Core Parallel Architecture (LoPRAM)
Modern microprocessor architectures have gradually incorporated support for parallelism. In the past the degree of parallelism was rather small and as such it could be best modeled as a constant speedup over the traditional RAM model, however, as a consequence of continued growth this assumption might no longer hold. For example, with the introduction of 32- and 64-bit architectures, bit-level parallelism became significant. This led to the introduction of the transdichotomous RAM model, for which many algorithms which are faster in theory and practice have been developed. Similarly, over the last five years, major microprocessor manufacturers have released road maps for the next decade predicting a rapidly increasing number of cores, with upwards of 64 cores per microprocessor by 2015.