Live, Runtime Phase Monitoring and Prediction on Real Systems with Application to Dynamic Power Management

Abstract

Computer architecture has experienced a major paradigm shift from focusing only on raw performance to considering power-performance efficiency as the defining factor of the emerging systems. Along with this shift has come increased interest in workload characterization. This interest fuels two closely related areas of research. First, various studies explore the properties of workload variations and develop methods to identify and track different execution behavior, commonly referred to as "phase analysis". Second, a large complementary set of research studies dynamic, on-the-fly system management techniques that can adaptively respond to these differences in application behavior. Both of these lines of work have produced very interesting and widely useful results. Thus far, however, there exists only a weak link between these conceptually related areas, especially for real-system studies.Our work aims to strengthen this link by demonstrating a real-system implementation of a runtime phase predictor that works cooperatively with on-the-fly dynamic management. We describe a fully-functional deployed system that performs accurate phase predictions on running applications. The key insight of our approach is to draw from prior branch predictor designs to create a phase history table that guides predictions. To demonstrate the value of our approach, we implement a prototype system that uses it to guide dynamic voltage and frequency scaling. Our runtime phase prediction methodology achieves above 90% prediction accuracies for many of the experimented benchmarks. For highly variable applications, our approach can reduce mispredictions by more than 6X over commonly-used statistical approaches. Dynamic frequency and voltage scaling, when guided by our runtime phase predictor, achieves energy-delay product improvements as high as 34% for benchmarks with non-negligible variability, on average 7% better than previous methods and 18% better than a baseline unmanaged system.

Phase-Based Application-Driven Hierarchical Power Management on the Single-chip Cloud Computer

“…In addition, we compare our SMRP predictor presented in Section 4 against a state-of-the-art history predictor, the Global History Table Predictor (GHTP) [13].…”

Package-Aware Scheduling of embedded workloads for temperature and Energy management on heterogeneous MPSoCs

“…This modular framework allows for an easy extension of the set of cores simulated in the heterogeneous MPSoC, and it is capable of integrating a variety of simulators or real-life experiments if needed. We assume the same three voltage settings for the XScale and [14]), and for SPARC we set the default frequency to 1.2GHz (as reported in [13]), and scale frequency using the 95% and 85% settings as in [17]. The in-order pipelines of SPARC and Xscale are modeled by modifying M5's execution engine.…”

Beyond CPU Frequency Scaling for a Fine-grained Energy Control of HPC Systems

“…Works done in [5], [6] use online techniques to detect applications execution phases, characterize them and set the appropriate CPU frequency accordingly. They rely on hardware monitoring counters to compute runtime statistics such as cache hit/miss ratio, memory access counts, retired instructions counts, etc., which are then used for phase detection and characterization.…”

“…They rely on hardware monitoring counters to compute runtime statistics such as cache hit/miss ratio, memory access counts, retired instructions counts, etc., which are then used for phase detection and characterization. Policies developed in [5], [6] tend to be designed for single task environment. We overcome this limitation by considering each node of the cluster as a black box, which means that we do not focus on any application, but instead on the platform.…”