Real-Time Program-Specific Phase Change Detection for Java Programs

Abstract

It is well-known that programs tend to have multiple phases in their execution. Because phases have impact on micro-architectural features such as caches and branch predictors, they are relevant to program performance Xian et al. [2007], Roh et al. [2009], Gu and Verbrugge [2008] and energy consumption. They are also relevant to detecting whether a program is executing as expected or is encountering unusual or exceptional conditions, a software engineering and program monitoring concern Peleg and Mendelson [2007], Singer and Kirkham [2008], Pirzadeh et al. [2011], Benomar et al. [2014]. We offer here a method for real-time phase change detection in Java programs. After applying a training protocol to a program of interest, our method can detect phase changes at run time for that program with good precision and recall (compared with a "ground truth" definition of phases) and with small performance impact (average less than 2%). We also offer improved methodology for evaluating phase change detection mechanisms. In sum, our approach offers the first known implementation of real-time phase detection for Java programs. CCS Concepts •Computer systems organization → Real-time systems; •Computing methodologies → Machine learning; •Software and its engineering → Runtime environments;

A Survey of Phase Classification Techniques for Characterizing Variable Application Behavior

“…Chiu et al [25] found that they could use variable-length intervals to improve classification accuracy. The authors proposed a technique that first traced an application's ex- Instruction type vector [86]; conditional branch execution [88]; task execution metrics frequency [41]…”

“…Variable length intervals [75] [48] [25] [108], on the other hand, allow the intervals to closely match the periodicity of the different phases and enable more accurate phase classification. However, variable length intervals are more complex, since they may require more detailed application analysis at design time or an online algorithm with a feedback loop for runtime phase classification [44].…”

“…Chiu et al [25] found that they could use variable-length intervals to improve classification accuracy. The authors proposed a technique that first traced an application's ex- 10K-15K instructions [101]; 10M [58]; 100M [89] instructions; 1s [24]; 100K branches [25]…”

“…Resource bottlenecks [101]; execution vectors [24]; cache misses [47]; metrics instruction type vectors [58]; branch edges/method entry/returns [25]; conditional branches [ Instruction type vector [86]; conditional branch execution [88]; task execution metrics frequency [41]…”