Software energy estimation based on statistical characterization of intermediate compilation code

Brandolese,; Corbetta,; Fornaciari, William

doi:10.1109/islped.2011.5993659

Cited by 21 publications

(17 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…At the source or intermediate levels, existing power estimation approaches employ coarse-grain models that assume a constant or statistical energy consumption model at the granularity of complete instructions or source-level opera- tions [2,3,7]. They thus largely focus on predicting the execution time correctly to arrive at an estimate of overall power consumed.…”

Section: Related Workmentioning

confidence: 99%

Automated, retargetable back-annotation for host compiled performance and power modeling

Chakravarty

Zhao

Gerstlauer

2013

2013 International Conference on Hardware/Software Codesign and System Synthesis (CODES+ISSS)

View full text Add to dashboard Cite

With traditional cycle-accurate or instruction-set simulations of processors often being too slow, host-compiled or source-level software execution approaches have recently become popular. Such high-level simulations can achieve order of magnitude speedups, but approaches that can achieve highly accurate characterization of both power and performance metrics are lacking. In this paper, we propose a novel host-compiled simulation approach that provides close to cycle-accurate estimation of energy and timing metrics in a retargetable manner, using flexible, architecture description language (ADL) based reference models. Our automated flow considers typical front-and back-end optimizations by working at the compiler-generated intermediate representation (IR). Path-dependent execution effects are accurately captured through pairwise characterization and backannotation of basic code blocks with all possible predecessors. Results from applying our approach to PowerPC targets running various benchmark suites show that close to native average speeds of 2000 MIPS at more than 98% timing and energy accuracy can be achieved.

show abstract

Section: Related Workmentioning

confidence: 99%

Automated, retargetable back-annotation for host compiled performance and power modeling

Chakravarty

Zhao

Gerstlauer

2013

2013 International Conference on Hardware/Software Codesign and System Synthesis (CODES+ISSS)

View full text Add to dashboard Cite

show abstract

“…[13], statistical analysis has been employed to build an energy model for LLVM IR for the purpose of fast and accurate early-stage prediction of the energy consumption of embedded software at the function level to enable compile-time energy optimisation. Modelling starts with instrumenting and compiling the source code of a large set of benchmarks into architecture-independent LLVM IR to extract block-level statistics capturing the structural features of the source code in terms of LLVM instructions.…”

Section: Energy Modelling At Higher Levels Of Software Abstractionmentioning

confidence: 99%

Energy-Aware Software Engineering

Eder¹,

Gallagher

2017

ICT - Energy Concepts for Energy Efficiency and Sustainability

View full text Add to dashboard Cite

A great deal of energy in Information and Communication Technology (ICT) systems can be wasted by software, regardless of how energy-efficient the underlying hardware is. To avoid such waste, programmers need to understand the energy consumption of programs during the development process rather than waiting to measure energy after deployment. Such understanding is hindered by the large conceptual gap from hardware, where energy is consumed, to high-level languages and programming abstractions. The approaches described in this chapter involve two main topics: energy modelling and energy analysis. The purpose of modelling is to attribute energy values to programming constructs, whether at the level of machine instructions, intermediate code or source code. Energy analysis involves inferring the energy consumption of a program from the program semantics along with an energy model. Finally, the chapter discusses how energy analysis and modelling techniques can be incorporated in software engineering tools, including existing compilers, to assist the energy-aware programmer to optimise the energy consumption of code.

show abstract

“…The basic ideas supporting the proposed approach have been explored by the authors in previous works [3] and finalized in [4]. The new approach described in this paper uses an intermediate representation based on the assembly-level LLVM code produced by the open-source project LLVM Compiler Infrastructure [2].…”

Section: Estimation Modelmentioning

confidence: 99%

“…The new approach described in this paper uses an intermediate representation based on the assembly-level LLVM code produced by the open-source project LLVM Compiler Infrastructure [2]. The modeling approach described in the following mimics the methodology proposed in [4], but uses the concept of effective capacitance to express the basic energy figures on which the estimation is founded. The estimation is decomposed in three subsequent phases:…”

Section: Estimation Modelmentioning

confidence: 99%

“…More precisely, the input C code is promoted to C++ and overloaded instrumented versions of all the operators are provided. The estimation approach proposed in this this paper is an evolution of a methodology and its related toolchain already presented by the authors in a previous work [4]. Two are the major novelties of the models and of the toolchain described in this paper:…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Software energy optimization through fine-grained function-level voltage and frequency scaling

Brandolese

Fornaciari

2012

Proceedings of the Eighth IEEE/ACM/IFIP International Conference on Hardware/Software Codesign and System Synthesis

Self Cite

View full text Add to dashboard Cite

This paper presents a methodology and a toolchain to perform estimation and optimization of the energy consumption associated to software execution on tiny embedded systems. The estimation phase is based on an ISA-level characterization of the target processor, while the optimization phase is made combining the estimation process with design space exploration in order to exploit fine-grained dynamic voltage and frequency scaling. The proposed approach operates at compile-time, with the granularity of single C functions and almost automatically augments the source code.

show abstract

Software energy estimation based on statistical characterization of intermediate compilation code

Cited by 21 publications

References 3 publications

Automated, retargetable back-annotation for host compiled performance and power modeling

Automated, retargetable back-annotation for host compiled performance and power modeling

Energy-Aware Software Engineering

Software energy optimization through fine-grained function-level voltage and frequency scaling

Contact Info

Product

Resources

About