PERFORMANCE ANALYSIS ON A CC-NUMA PROTOTYPE

Cache-coherent nonuniform memory access (CC-NUMA) machines have been s hown to be a promising paradigm for exploiting distributed execution. CC-NUMA systems can provide performance typically associated with para llel machines, without the high cost associated with parallel programm ing. This is because a single image of memory is provided on a CC-NUMA machine. Past research on CC-NUMA machines has focused on modificatio ns to the memory hierarchy, interconnect topology, and memory consiste ncy protocols, which are all areas critical to achieving scalable perf ormance. The research described here expands this focus to issues asso ciated with operating system structures which can increase system scal ability. We describe a hardware/software prototyping study which inves tigates how changes to the operating system of a commercia! IBM AS/400 (R) system can provide scalable performance when running transaction p rocessing workloads. The project described was a joint effort between researchers at the IBM Thomas J. Watson Research Center and a team fro m the AS/400 development laboratory in Rochester, Minnesota. This pape r describes various aspects of the project, including changes made to the operating system to enable scalable performance, and the associate d hardware and software performance tools developed to identify bottle necks in the existing operating system structures.