Partial Data Traces: Efficient Generation and Representation
Binary manipulation techniques are increasing in popularity. They support program transformations tailored toward certain program inputs, and these transformations have been shown to yield performance gains beyond the scope of static code optimizations without profile-directed feedback. They even deliver moderate gains in the presence of profile-guided optimizations. In addition, transformations can be performed on the entire executable, including library routines. This work focuses on program instrumentation, yet another application of binary manipulation. This paper reports preliminary results on generating partial data traces through dynamic binary rewriting. The contributions are threefold. First, a portable method for extracting precise data traces for partial executions of arbitrary applications is developed. Second, a set of hierarchical structures for compactly representing these accesses is developed. Third, an efficient online algorithm to detect regular accesses is introduced. The authors utilize dynamic binary rewriting to selectively collect partial address traces of regions within a program. This allows partial tracing of hot paths for only a short time during program execution in contrast to static rewriting techniques that lack hot path detection and also lack facilities to limit the duration of data collection. Preliminary results show reductions of three orders of a magnitude of inline instrumentation over a dual process approach involving context switching. They also report constant size representations for regular access patters in nested loops. These efforts are part of a larger project to counter the increasing gap between processor and main memory speeds by means of software optimization and hardware enhancements.
- Research Organization:
- Lawrence Livermore National Lab. (LLNL), Livermore, CA (United States)
- Sponsoring Organization:
- US Department of Energy (US)
- DOE Contract Number:
- W-7405-ENG-48
- OSTI ID:
- 15006192
- Report Number(s):
- UCRL-JC-144405-REV-1; TRN: US200405%%329
- Resource Relation:
- Conference: Workshop on Binary Translation 2001, in conjunction with the International Conference on Parallel Architectures and Compilation Techniques, Barcelona (ES), 09/08/2001--09/12/2001; Other Information: PBD: 20 Aug 2001
- Country of Publication:
- United States
- Language:
- English
Similar Records
Data Locality Enhancement of Dynamic Simulations for Exascale Computing (Final Report)
A scalable method for run-time loop parallelization