| This dissertation describes complete machine simulation, a novel approach to understanding the behavior of modern computer systems. Complete machine simulation differs from existing simulation tools by modeling all of the hardware found in a typical computer system. As a result, it can help investigators better understand the behavior of machines running commercial operating systems as well as any applications designed for these operating systems. In contrast, most existing simulation tools model only limited portions of a computer's hardware and cannot support the accurate investigation of many workloads. In addition to extensive workload support, the complete machine simulation approach permits significant hardware modeling flexibility and provides detailed information regarding the behavior of this hardware. This combination of features has widespread applicability, providing benefits to such research domains as hardware design, operating system development, and application performance tuning.; Complete machine simulation extends the applicability of traditional machine simulation techniques by addressing two difficult challenges. The first challenge is to achieve the speed needed to investigate complex, long-running workloads. To address this challenge, complete machine simulation allows an investigator to dynamically adjust the characteristics of its hardware simulation. There is an inherent trade-off between the level of detail that a hardware simulator models and the speed at which it runs, and complete machine simulation provides users with explicit control over this trade-off. An investigator can select a high-speed, low-detail simulation setting to quickly pass through uninteresting portions of a workload's execution. Once the workload has reached a more interesting execution state, an investigator can switch to slower, more detailed simulation to obtain behavioral information. The second challenge is to efficiently organize low-level hardware simulation data into information that is more meaningful to an investigation. Complete machine simulation addresses this challenge by providing mechanisms that allow a user to easily incorporate higher-level workload knowledge into the data management process. These mechanisms are efficient and further improve simulation speed by customizing all data collection and reporting to the specific needs of an investigation. |
