TRACE FACTORY - GENERATING WORKLOADS FOR TRACE-DRIVEN SIMULATION OF SHARED-BUS MULTIPROCESSORS

A major concern with high-performance general-purpose workstations is to speed up the execution of commands, uniprocess applications, and mu ltiprocess applications with coarse-to medium-grain parallelism, To th at end, a simple extension of a uniprocessor machine such as a shared- bus, shared-memory architecture can be employed, Both kinds of machine s generally use the same OS model, and die same application can execut e on these machines without recoding. However, an intrinsic limitation of the shared-bus architecture is the low number of processors that c an be connected to the shared bus. When this number exceeds a critical value, the system's global performance drops drastically because of b us saturation, When two or more processors store a copy of the same me mory block in their respective caches and one of them performs a write operation on a location in that block, a set of bus actions is necess ary to guarantee that every subsequent read operation by any processor can get the up-to-date value of the modified location. Typically, res earchers use simulation to investigate how to improve the performance of such machines. In particular, trace-driven simulation offers a good trade-off between speed, accuracy, and flexibility. A key point of th is methodology is to find traces that both represent typical operating conditions and include all information potentially needed fur an accu rate simulation of the system, The authors have developed a methodolog y and a set of tools (called Trace Factory) to generate traces for the performance evaluation of shared-bus, shared-memory multiprocessor sy stems. Trace Factory is particularly useful for evaluating a multiproc essor architecture's performance related to different workloads and to most of the influencing activities of the operating system. The desig ner can evaluate and tune architectural solutions for coherence protoc ol, cache structure, bus, and memory.