Inference and declaration of independence

Abstract

We present a set of static techniques that reduce runtime overheads in task-parallel programs with implicit synchronization. We use a static dependence analysis to detect nonconflicting tasks and remove unnecessary runtime checks. We further reduce overheads by statically optimizing task creation and management of runtime metadata. We implemented these optimizations in SCOOP, a source-to-source compiler for such a programming model and runtime system. We evaluate SCOOP on 10 representative benchmarks and show that our approach can improve performance by 12% on average.

Optimization of program execution using computational significance

“…The compiler for the programming model is implemented based on a source-tosource compiler infrastructure [108]. It recognizes the pragmas introduced by the programmer and lowers them to corresponding calls of the respective runtime system (supporting approximate or unreliable computing).…”

“…Additionally, I contributed to the process of benchmarking, as well as the analysis of the results. I also modified an existing source-to-source compiler [108] to augment it with significance-related extensions.…”

System software techniques to enhance reliability of modern platforms

“…A user-level library implements the runtime system and runs on top of the Linux operating system. A source-to-source compiler, which we developed based on [148], lowers programs that use the primitives of our programming model to code with calls to the runtime system API. Finally, the produced source code is compiled into machine code using the standard gcc tool chain.…”