There are many metrics designed to assist in the performance debugging of large-scale parallel applica-tions. We describe a new technique, called True Zeroing, that permits direct quantitative comparison of the gui-dance supplied by these metrics on real applications. We apply this technique to three programs that include both numeric and symbolic applications. We compare three existing metrics: Gprof, Critical Path, and Quartz/NPT, and several new variations. Critical Path provided the best overall guidance, but it was not infallible. We also include a set of recommendations to tool builders based on the experience gained during our case study. 1
The authors discuss the important questions that benchmarking must answer and the degree to which su...
[[abstract]]©1988 North-Holland-The authors outline an approach to the design of a set of interactiv...
We analyse the inherent performance of parallel software. For this end we use a task graph to model ...
There are many metrics designed to assist in the performance debugging of large-scale parallel appli...
There are many metrics designed to assist in the performance debugging of large-scale parallel appli...
Over the past 10 years we have seen the transition from single core computer to multicore computing,...
Performance analysis of parallel programs continues to be challenging for programmers. Programmers h...
Traditional performance debugging and tuning of parallel programs is based on the "measure-modify" a...
Critical Path Profiling is a technique that provides guidance to help programmers try to improve the...
HPC applications are often very complex and their behavior depends on a wide range of factors from a...
Most performance debugging and tuning of parallel programs is based on the "measure-modify"...
Achieving a significant fraction of peak performance on a modern high-performance computer is a chal...
A new technique for estimating and understanding the speed improvement that can result from executin...
High-performance computing systems have become increasingly dynamic, complex, and unpredictable. To ...
AbstractIn software development, a metric is the measurement of some characteristic of a program's p...
The authors discuss the important questions that benchmarking must answer and the degree to which su...
[[abstract]]©1988 North-Holland-The authors outline an approach to the design of a set of interactiv...
We analyse the inherent performance of parallel software. For this end we use a task graph to model ...
There are many metrics designed to assist in the performance debugging of large-scale parallel appli...
There are many metrics designed to assist in the performance debugging of large-scale parallel appli...
Over the past 10 years we have seen the transition from single core computer to multicore computing,...
Performance analysis of parallel programs continues to be challenging for programmers. Programmers h...
Traditional performance debugging and tuning of parallel programs is based on the "measure-modify" a...
Critical Path Profiling is a technique that provides guidance to help programmers try to improve the...
HPC applications are often very complex and their behavior depends on a wide range of factors from a...
Most performance debugging and tuning of parallel programs is based on the "measure-modify"...
Achieving a significant fraction of peak performance on a modern high-performance computer is a chal...
A new technique for estimating and understanding the speed improvement that can result from executin...
High-performance computing systems have become increasingly dynamic, complex, and unpredictable. To ...
AbstractIn software development, a metric is the measurement of some characteristic of a program's p...
The authors discuss the important questions that benchmarking must answer and the degree to which su...
[[abstract]]©1988 North-Holland-The authors outline an approach to the design of a set of interactiv...
We analyse the inherent performance of parallel software. For this end we use a task graph to model ...