The interplay between pairs of critical factors such as performance, energy and reliability within modern computing systems has always been an interesting topic of study. However, studying the interplay of all three factors together in a many-core, multi-layer design setting has been a relatively recent undertaking. This work explores the practical problems encountered in such studies and introduces the modelling framework ArchOn, which is based on a novel resource-driven graph representation. ArchOn facilitates the analysis and potentially design and synthesis of systems whose design domains are more conveniently organized into multiple layers or levels (e.g. application, OS, hardware, etc.) and potentially large scale and diverse types of concurrency. The layer-agnostic formalism helps designers reason about cross-layer issues and the resource-driven approach is advantageous for reasoning about such issues as energy and time. Example single-and multi-core case studies help explain and illustrate the method.
The analysis for extra-functional properties like power and performance takes a critical role in the system design workflow. Hardware-software co-simulation is one of the commonly used ways to perform this type of analysis. However, with the modern development of many-core systems the problem of scalability is becoming a bottleneck for all analysis techniques including simulation, especially when a simple extrapolation from the single core results is unacceptable. This paper presents a framework aimed at the extra-functional analysis during the rapid prototyping stages of system design. The tool is based on stochastic modelling and simulation of cross-layer system representations. The concept of selective abstraction is applied to ensure a sufficient level of accuracy where it is needed, while reducing the complexity of the parts that are of less importance. A set of Networks-on-Chip topologies has been analysed and presented as a use case example. Complex (New) Platform Model Core1 Core2 Core100 ...
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.