Loading...
Please wait, while we are loading the content...
Similar Documents
Performance scalability prediction on multicomputers
| Content Provider | Semantic Scholar |
|---|---|
| Author | Reed, Daniel A. Mendes, Celso L. |
| Copyright Year | 1997 |
| Abstract | Despite the performance potential of multicomputers, several factors have limited their widespread adoption. Of these, performance variability is among the most significant. Execution of some programs may yield only a small fraction of peak system performance, whereas others approach the system's theoretical performance peak. Moreover, the observed performance may change substantially as application program parameters vary. Data parallel languages, which facilitate the programming of multicomputers, increase the semantic distance between the program's source code and its observable performance, thus aggravating the performance problem. In this thesis, we propose a new methodology to predict the performance scalability of data parallel applications on multicomputers. Our technique represents the execution time of a program as a symbolic expression that is a function of the number of processors (P), problem size (N), and other system-dependent parameters. This methodology is based on information collected at compile time. By extending an existing data parallel compiler (Fortran D95), we derive, during compilation, a symbolic model that represents the cost of each high-level program section and, inductively, of the complete program. These symbolic expressions may be simplified externally with current symbolic tools. Predicting performance of the program for a given pair $(P,\ N)$ requires simply the evaluation of its corresponding cost expression. We validate our implementation by predicting scalability of a variety of loop nests, with distinct computation and communication patterns. To demonstrate the applicability of our technique, we present a series of concrete performance problems where it was successfully employed: prediction of total execution time, identification and tracking of bottlenecks, cross-system prediction, and evaluation of code transformations. These examples show that the technique would be useful both to users, in optimizing and tuning their programs, and to advanced compilers, which would have a means to evaluate the expected performance of a synthesized code. According to the results of our study, by integrating compilation, performance analysis and symbolic manipulation tools, it is possible to correctly predict, in an automated fashion, the major performance variations of a data parallel program written in a high-level language. |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | http://vibes.cs.uiuc.edu/Publications/Theses/MendesPhD.ps.gz |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
