POEMS: End-to-end performance design of large parallel adaptive computational systems

The POEMS project is creating an environment for end-to-end performance mod eling of complex parallel and distributed systems, spanning the domains of application software, runtime and operating system software, and hardware a rchitecture. Toward this end. the POEMS framework supports composition of c omponent models from these different domains into an end-to-end system mode l. This composition can be specified using a generalized graph model of a p arallel system, together with interface specifications that carry informati on about component behaviors and evaluation methods. The POEMS Specificatio n Language compiler, under development, will generate an end-to-end system model automatically from such a specification. The components of the target system may be modeled using different modeling paradigms (analysis, simula tion, or direct measurement) and may be modeled at various levels of detail . As a result, evaluation of a POEMS end-to-end system model may require a variety of evaluation tools including specialized equation solvers, queuing network solvers, and discrete-event simulators. A single application repre sentation based on static and dynamic task graphs serves as a common worklo ad representation for all these modeling approaches. Sophisticated parallel izing compiler techniques allow this representation to be generated automat ically for a given parallel program. POEMS includes a library of predefined analytical and simulation component models of the different domains and a knowledge base that describes performance properties of widely used algorit hms. This paper provides an overview of the POEMS methodology and illustrat es several of its key components. The methodology and modeling capabilities are demonstrated by predicting the performance of alternative configuratio ns of Sweep3D, a complex benchmark for evaluating wavefront application tec hnologies and high-performance, parallel architectures.