Search

With the diverging improvements in CPU speeds and memory access latencies, detecting and removing memory access bottlenecks becomes increasingly important. In this work we present METRIC, a software framework for isolating and understanding such bottlenecks using partial access traces. METRIC extracts access traces from executing programs without special compiler or linker support. We make four primary contributions. First, we present a framework for extracting partial access traces based on dynamic binary rewriting of the executing application. Second, we introduce a novel algorithm for compressing these traces. The algorithm generates constant space representations for regular accesses occurring in nested loop structures. Third, we use these traces for offline incremental memory hierarchy simulation. We extract symbolic information from the application executable and use this to generate detailed source-code correlated statistics including per-reference metrics, cache evictor information, and stream metrics. Finally, we demonstrate how this information can be used to isolate and understand memory access inefficiencies. This illustrates a potential advantage of METRIC over compile-time analysis for sample codes, particularly when interprocedural analysis is required. Categories and Subject Descriptors: D.3.4 [Programming Languages]: Processors---Compilers; optimization; B.3.2 [Memory Structures]: Design Styles--Cache memories General Terms: Algorithms, Languages, Performance Additional Key Words and Phrases: Dynamic binary rewriting, program instrumentation, data trace generation, data trace compression, cache analysis

With the rapid replacement of closed, homogeneous, proprietary HPC systems by heterogeneous, Linux-MPI cluster systems, the state of performance monitoring and analysis tools has become a cause for concern. Proprietary systems, despite their drawbacks, provided consistent tools of high quality. Modern Linux cluster systems, on the other hand, benefit from a wide variety of Open Source tools in differing stages of evolution. Recognizing that Linux clusters are here to stay, SiCortex has taken a unique approach of integrating and enhancing Open Source tools into a production-quality suite. Further, as a tribute to the unrewarded Open Source community developers and for more pragmatic reasons, such as long-term sustainability, changes made to the tools are fed upstream to the original tool developers. In this paper, we present an overview of the SiCortex tools' suite, and some of the challenges and successes we had in the process of realizing it. Copyright © 2009 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]