|
Descripción
|
|
|---|---|
| The static estimation of the energy consumed by program executions is an important challenge, which has applications in program optimization and verification, and enables energy-aware software development. An approach has been proposed previously that estimates such energy as functions on a program's input data sizes. This approach has been further adapted to the LLVM IR level, making use of techniques developed in our theses. The work in the first thesis deals with the propagation of an existing ISA level energy model to the LLVM-IR level, through a mapping technique, to assign energy values to each instruction at that layer. Using these propagated energy values, the work in the second thesis deals with the conversion from the LLVM-IR format to a Horn clause-based representation, so that existing analyses can be used to infer energy cost functions on the input data sizes. Experimental results show that our LLVM layer analysis is reasonably accurate and more powerful than our previous analysis at the ISA layer | |
|
Internacional
|
Si |
|
Entidad
|
ICT-Energy (Nanoenergy) Letters |
|
Lugar
|
|
|
Páginas
|
16-17 |
|
Referencia/URL
|
http://www.nanoenergyletters.com/files/nel/ICT-Energy_Letters_8.pdf |
|
Tipo de publicación
|
Newsletter |