This report outlines the design of a performance analysis environment for SVM-Fortran programs. SVM-Fortran is a shared memory parallel programming language developed at KFA for distributed memory multiprocessors. The environment allows to analyze the data locality of a given program via runtime tracing and supports the identification of critical code regions to guide the user or an optimization tool in tuning the program. To reduce the amount of runtime data, the environment combines trace data with static program information and supports an incremental analysis cycle. In addition to the overall design, we present the trace format of runtime events. The trace format includes symbolic information to relate runtime information to the program...
Modern supercomputers with multi-core nodes enhanced by accelerators, as well as hybrid programming ...
SVM-Fortran is a language designed to program highly parallel systems with a global address space. A...
Most performance debugging and tuning of parallel programs is based on the "measure-modify"...
Programming distributed memory multiprocessors requires program parallelization as well as program o...
Programming distributed memory parallel computers with message passing is often considered to be a d...
By using the shared memory programming model it is possible to parallelize applications step by step...
Programming distributed memory multiprocessors requires program parallelization as well as program o...
SVM-Fortran is a language extension of Fortran 77 developed by KFA for shared memory parallel progra...
For massively parallel systems, the well-known global address space is realized by the programming m...
This paper describes a performance tuning tool, named SVMview, for DSM-based parallel computers. SVM...
P 3 T is an interactive performance estimator that assists users in performance tuning of scientif...
Vienna Fortran, a machine-independent language extension to Fortran which allows the user to write p...
This paper presents a profiling tool that allows the programmer to identify the regions of the progr...
This article outlines the parallelization of an irregular grid application with SVM-Fortran. It desc...
This article outlines the parallelization of an irregular grid application based on shared virtual m...
Modern supercomputers with multi-core nodes enhanced by accelerators, as well as hybrid programming ...
SVM-Fortran is a language designed to program highly parallel systems with a global address space. A...
Most performance debugging and tuning of parallel programs is based on the "measure-modify"...
Programming distributed memory multiprocessors requires program parallelization as well as program o...
Programming distributed memory parallel computers with message passing is often considered to be a d...
By using the shared memory programming model it is possible to parallelize applications step by step...
Programming distributed memory multiprocessors requires program parallelization as well as program o...
SVM-Fortran is a language extension of Fortran 77 developed by KFA for shared memory parallel progra...
For massively parallel systems, the well-known global address space is realized by the programming m...
This paper describes a performance tuning tool, named SVMview, for DSM-based parallel computers. SVM...
P 3 T is an interactive performance estimator that assists users in performance tuning of scientif...
Vienna Fortran, a machine-independent language extension to Fortran which allows the user to write p...
This paper presents a profiling tool that allows the programmer to identify the regions of the progr...
This article outlines the parallelization of an irregular grid application with SVM-Fortran. It desc...
This article outlines the parallelization of an irregular grid application based on shared virtual m...
Modern supercomputers with multi-core nodes enhanced by accelerators, as well as hybrid programming ...
SVM-Fortran is a language designed to program highly parallel systems with a global address space. A...
Most performance debugging and tuning of parallel programs is based on the "measure-modify"...