Paper Info
Title
Enhanced Performance Analysis of Multi-core Applications with an Integrated Tool-chain - Using Scalasca and Vampir to Optimise the Metal Forming Simulation FE Software INDEED.
Abstract
Programming and optimising large parallel applications for multi-core systems is an ambitious and time consuming challenge. Therefore, a number of software tools have been developed in the past to assist the programmer in optimising their codes. Scalasca and Vampir are two of these performance-analysis tools that are already well established and recognised in the community. While Scalasca locates, quantifies, and presents performance flaws (defined as event patterns) in a compact and simple to use tree representation, the Vampir framework visualises all details of the function call sequences and communication patterns in a great variety of timeline or statistic displays. Scalasca locates and presents the problems found and their severity but gives limited practical hints on the case history to find a solution. Looking at the huge amount of detailed data in Vampir can make it hard to find the interesting spots in spatial and temporal resolution for identification of problems and selection for further investigation. This observation led to the combination of these tools resulting in a more productive tool-chain. In this paper we present our approach to connect Scalasca and Vampir through a common bus system. A practical real-world example is provided and the beneficial impact of the integration on the work-flow is shown by optimising the adaptive mesh refinement and load balancing phases of the metal forming simulation FE software INDEED.
Year
DOI
Venue
2009
10.3233/978-1-60750-530-3-701
PARALLEL COMPUTING: FROM MULTICORES AND GPU'S TO PETASCALE
Keywords
Field
DocType
MPI,OpenMP,performance optimisation,tools integration,Scalasca,Vampir
Computer science,Parallel computing,Software,Multi-core processor,Metal forming
Conference
Volume
ISSN
Citations 
19
0927-5452
0
PageRank 
References 
Authors
0.34
0
5
Name
Order
Citations
PageRank
Thomas William100.34
Hartmut Mix216914.34
B. Mohr341.67
René Menzel400.34
Felix Voigtländer541.06