Abstract

Abstract machines provide a certain separation between platform-dependent and platform-independent concerns in compilation. Many of the differences between architectures are encapsulated in the specific abstract machine implementation and the bytecode is left largely architecture independent. Taking advantage of this fact, we present a framework for estimating upper and lower bounds on the execution times of logic programs running on a bytecode-based abstract machine. Our approach includes a one-time, program-independent profiling stage which calculates constants or functions bounding the execution time of each abstract machine instruction. Then, a compiletime cost estimation phase, using the instruction timing information, infers expressions giving platform-dependent upper and lower bounds on actual execution time as functions of input data sizes for each program. Working at the abstract machine level allows taking into account lowlevel issues without having to tailor the analysis for each architecture and platform, and instead only having to redo the calibration step. Applications of such predicted execution times include debugging/verification of time properties, granularity control in parallel/distributed computing, and resource-oriented specialization.

Citations

474	An Abstract Prolog Instruction Set - Warren - 1983
231	Warren’s Abstract Machine: A Tutorial Reconstruction. Logic Programming - Aït-Kaci - 1991
110	Non-failure Analysis for Logic Programs - Debray, López-García, et al. - 1997
63	Task Granularity Analysis in Logic Programs - Debray, Lin, et al. - 1990
55	Cost Analysis of Java Bytecode - Albert, Arenas, et al. - 2007
50	Lower Bound Cost Estimation for Logic Programs - Debray, Lopez-Garcia, et al. - 1997
42	Integrated program debugging, verification, and optimization using abstract interpretation (and the Ciao system preprocessor - Hermenegildo, Puebla, et al. - 2005
25	User-definable resource bounds analysis for logic programs - Navas, Mera, et al. - 2010
24	Automatic time-bound analysis for a higher-order language - G’omez, Liu - 1999
17	M.: Self-tuning resource aware specialisation for prolog - Craig, Leuschel - 2005
16	Automated higher-order complexity analysis - Benzinger - 2004
16	Abstract machines for programming language implementation - Diehl, Sestoft - 2000
16	Combining static analysis and profiling for estimating execution times - Mera, López-García, et al. - 2007
14	Low-level analysis of a portable Java byte code WCET analysis framework - Bate, Bernat, et al.
11	Timing Analysis and Timing Predictability - Wilhelm - 2004
8	Java virtual-machine support for portable worst-case execution-time analysis - Bate, Bernat, et al. - 2002
6	PURRS: The Parma University’s recurrence relation solver. http: //www.cs.unipr.it/purrs - Bagnara, Pescetti, et al.
5	Experiences from Industrial WCET Analysis Case Studies - Ermedahl, Gustafsson, et al. - 2005
4	Deriving java virtual machine timing models for portable worst-case execution time analysis - Hu, Wellings, et al. - 2003
4	Virtual Environment Architecture - PowerPC
4	Non-failure Analysis and Granularity Control in Parallel Execution of Logic Programs - López-García - 2000
2	Warren’s Abstract Machine - A Java Implementation. http://www.stefan.buettcher.org/cs/wam/index.html - Buettcher
1	Application Notes. Using the RDTSC Instruction for Performance Monitoring - Intel’s - 1997