This collection is exclusively made up of theses, dissertations, and project reports submitted as a requirement for completing a graduate degree at UPR-Mayagüez.
If you are a UPRM graduate student and you are looking for information related to the deposit process, please refer to https://libguides.uprm.edu/repositorioUPRM/tesis
Browsing Theses & Dissertations by Author "Acevedo-Patiño, Oscar"
Acevedo-Patiño, Oscar; Jiménez-Cedeño, Manuel; College of Engineering; Palomera-Garcia, Rogelio; Couvertier, Isidoro; Department of Electrical and Computer Engineering; Maldonado, Francisco
Current methodologies for software-level estimation of power and energy consumption use a power model developed for the microprocessor along with specialized tools that profile the program under study to extract the model ́s parameters. These tools commonly rely on real-time execution or simulations of the analyzed program which require real run-time data. This work presents an alternative methodology for power and energy estimation, in which the usage of such specialized tools is deemphasized. It is proposed instead the use of static code analysis to study and predict a program behavior. This, in combination with the microprocessor power model, allows the developed methodology to estimate power and energy for a program execution with only a small amount of run-time data. The new methodology first performs a static analysis of the program and then computes time, power, and energy costs for each node of its control flow graph (CFG) representation. Subsequently, these costs are combined with statistical program path information, obtained by analyzing the CFG, arriving at estimated power and energy costs for the program. We present power and energy estimation results for a set of 5 representative benchmark embedded-system programs. Results show that the new methodology, besides the advantages of static code analysis and reduced run-time data requirements, can produce competitive results, with errors less than 20% for energy estimates and 12% for power estimates, with respect to experimentally measured values.