Summary
As a result of increased core count both on Intel Xeon Phi and nVidia Tesla platforms, global interest in SMP (shared memory parallelism) and SIMD (single instruction multiple data) computations has rekindled. For complex computations, however, development of data structures and algorithms has lagged behind serial methods. The goal of this PhD is therefore to build systematically towards data structures, tools and libraries that effectively exploit the massive parallelism now available.
