TY - GEN
T1 - Performance modeling for systematic performance tuning
AU - Hoefler, Torsten
AU - Gropp, William
AU - Snir, Marc
AU - Kramer, William
PY - 2011
Y1 - 2011
N2 - The performance of parallel scientific applications depends on many factors which are determined by the execution environment and the parallel application. Especially on large parallel systems, it is too expensive to explore the solution space with series of experiments. Deriving analytical models for applications and platforms allow estimating and extrapolating their execution performance, bottlenecks, and the potential impact of optimization options. We propose to use such "performance modeling" techniques beginning from the application design process throughout the whole software development cycle and also during the lifetime of supercomputer systems. Such models help to guide supercomputer system design and re-engineering efforts to adopt applications to changing platforms and allow users to estimate costs to solve a particular problem. Models can often be built with the help of well-known performance profiling tools. We discuss how we successfully used modeling throughout the proposal, initial testing, and beginning deployment phase of the Blue Waters supercomputer system.
AB - The performance of parallel scientific applications depends on many factors which are determined by the execution environment and the parallel application. Especially on large parallel systems, it is too expensive to explore the solution space with series of experiments. Deriving analytical models for applications and platforms allow estimating and extrapolating their execution performance, bottlenecks, and the potential impact of optimization options. We propose to use such "performance modeling" techniques beginning from the application design process throughout the whole software development cycle and also during the lifetime of supercomputer systems. Such models help to guide supercomputer system design and re-engineering efforts to adopt applications to changing platforms and allow users to estimate costs to solve a particular problem. Models can often be built with the help of well-known performance profiling tools. We discuss how we successfully used modeling throughout the proposal, initial testing, and beginning deployment phase of the Blue Waters supercomputer system.
UR - http://www.scopus.com/inward/record.url?scp=83055184891&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=83055184891&partnerID=8YFLogxK
U2 - 10.1145/2063348.2063356
DO - 10.1145/2063348.2063356
M3 - Conference contribution
AN - SCOPUS:83055184891
SN - 9781450311397
T3 - State of the Practice Reports, SC'11
BT - State of the Practice Reports, SC'11
T2 - State of the Practice Reports, SC'11
Y2 - 12 November 2011 through 18 November 2011
ER -