PublicationDetail

Chair for Computer Aided Medical Procedures & Augmented Reality
Lehrstuhl für Informatikanwendungen in der Medizin & Augmented Reality

G. Klinker, I-Yu Chen
PerfVisS: A Performance Visualizer for High Performance Fortran Programs on Workstation Farms
Technical report 94/4, Digital Equipment Corporation, Cambridge Research Lab (DEC-CRL), One Kendall Square, Cambridge, MA, June 1994. (bib)

Writing efficient code for parallel processors is still a rather complex and little understood task. Tools to gather and analyze performance information during program execution are essential for understanding the reasons for inefficient executions. Current performance visualization systems provide only a limited set of capabilities, hardwired into huge, monolithic programs. Programmers need a very flexible environment in which they can mix and match different performance visualization tools. Research in scientific visualization has developed several environments to visualize, explore and analyze large quantities of data. Reusing these capabilities to visualize MIMD performance data significantly helps in the development of a performance profiler. However, scientific visualization methods have to be altered appropriately to address the non-geometric nature of performance data. This paper presents a profiler, PerfVisS?, which builds upon existing Telecollaborative Data Exploration (TDE) technology. PerfVisS? provides several linked views of the performance data, both for gaining a general overview of the program performance and for detailed data and code inspection. It also allows users to adapt the system to their own viewing preferences, using the AVS visual programming interface. Furthermore, it provides tools for telecollaborative performance tuning, allowing programmers to share their views with colleagues at remote sites. PerfVisS? runs on Alpha AXP\raisebox{.8ex}{\tiny TM} workstation farms but could be used as well on other similar MIMD architectures such as symmetric multiprocessors (SMPs).
This material is presented to ensure timely dissemination of scholarly and technical work. Copyright and all rights therein are retained by authors or by other copyright holders. All persons copying this information are expected to adhere to the terms and constraints invoked by each authors copyright. In most cases, these works may not be reposted without the explicit permission of the copyright holder.



Edit | Attach | Refresh | Diffs | More | Revision r1.13 - 30 Jan 2019 - 15:16 - LeslieCasas

Lehrstuhl für Computer Aided Medical Procedures & Augmented Reality    rss.gif