Accurately measuring MPI broadcasts in a computational grid
- LLNL
An MPI library's implementation of broadcast communication can significantly affect the performance of applications built with that library. In order to choose between similar implementations or to evaluate available libraries, accurate measurements of broadcast performance are required. As we demonstrate, existing methods for measuring broadcast performance are either inaccurate or inadequate. Fortunately, we have designed an accurate method for measuring broadcast performance, even in a challenging grid environment. Measuring broadcast performance is not easy. Simply sending one broadcast after another allows them to proceed through the network concurrently, thus resulting in inaccurate per broadcast timings. Existing methods either fail to eliminate this pipelining effect or eliminate it by introducing overheads that are as difficult to measure as the performance of the broadcast itself. This problem becomes even more challenging in grid environments. Latencies a long different links can vary significantly. Thus, an algorithm's performance is difficult to predict from it's communication pattern. Even when accurate pre-diction is possible, the pattern is often unknown. Our method introduces a measurable overhead to eliminate the pipelining effect, regardless of variations in link latencies. choose between different available implementations. Also, accurate and complete measurements could guide use of a given implementation to improve application performance. These choices will become even more important as grid-enabled MPI libraries [6, 7] become more common since bad choices are likely to cost significantly more in grid environments. In short, the distributed processing community needs accurate, succinct and complete measurements of collective communications performance. Since successive collective communications can often proceed concurrently, accurately measuring them is difficult. Some benchmarks use knowledge of the communication algorithm to predict the timing of events and, thus, eliminate concurrency between the collective communications that they measure. However, accurate event timing predictions are often impossible since network delays and local processing overheads are stochastic. Further, reasonable predictions are not possible if source code of the implementation is unavailable to the benchmark. We focus on measuring the performance of broadcast communication.
- Research Organization:
- Lawrence Livermore National Lab., CA (US)
- Sponsoring Organization:
- USDOE Office of Defense Programs (DP) (US)
- DOE Contract Number:
- W-7405-ENG-48
- OSTI ID:
- 12133
- Report Number(s):
- UCRL-JC-133177--Rev-1; DP0101031; DP0101031
- Country of Publication:
- United States
- Language:
- English
Similar Records
Test suite for evaluating performance of MPI implementations that support MPI_THREAD_MULTIPLE.
Test suite for evaluating performance of multithreaded MPI communication.