Amdahl's Law as an Instrumentation Tool for Building Eecient Parallel Code and Corresponding Author
نویسندگان
چکیده
Using Amdahl's law as an instrumentation guide we illustrate a technique for developing e cient code on Massively Parallel Processor (MPP) performance-class networks to solve nontrivial, high performance scienti c problems. We also show the importance of collective communication within the Message Passing Interface (MPI) paradigm for some applications. Given the popularity of Beowulf-like clusters of workstations our work also indicates the necessity of a scalable high performance network for obtaining e cient performance in parallel code. Using our approach we were able to obtain an e ective speedup (comparison with the best sequential time) of 170 when using 256 of the Cray T3E 900 processing elements (PEs) to solve a carbon, molecular-dynamic problem. We also examine the approach on a very di erent application: a Lanczos eigenvalue solver. 3
منابع مشابه
Using AMDAHL's Law as a Metric to Drive Code Parallelization: Two Case Studies
Using Amdahl's Law as a metric we illustrate a technique for developing eÆcient code on Massively Parallel Processor (MPP) performance-class networks to solve nontrivial, high performance scienti c problems. We also show the importance of collective communication within the Message Passing Interface (MPI) paradigm for some applications. Given the popularity of Beowulf-like clusters of workstati...
متن کاملAmdahl's and Gustafson-Barsis laws revisited
The paper presents a simple derivation of the Gustafson-Barsis law from the Amdahl's law. In the computer literature these two laws describing the speedup limits of parallel applications are derived separately. It is shown, that treating the time of the execution of the sequential part of the application as a constant, in few lines the Gustafson-Barsis law can be obtained from the Amdahl's law ...
متن کاملScalable Problems and Memory-bounded Speedup Scalable Problems and Memory-bounded Speedup
In this paper three models of parallel speedup are studied. They are xed-size speedup, xed-time speedup and memory-bounded speedup. The latter two consider the relationship between speedup and problem scalability. Two sets of speedup formulations are derived for these three models. One set considers uneven workload allocation and communication overhead, and gives more accurate estimation. Anoth...
متن کاملScalable Problems and Memory-Bounded Speedup
In this paper three models of parallel speedup are studied. They are fixed-size speedup, fixed-time speedup and memory-bounded speedup. The latter two consider the relationship between speedup and problem scalability. Two sets of speedup formulations are derived for these three models. One set considers uneven workload allocation and communication overhead, and gives more accurate estimation. A...
متن کاملParallel Numerical Algorithms: an Introduction
The rapidly evolving field of parallel computing has seen four eras, substantially overlapping, yet progressive. The first, which we could label the Stone Age, was characterized by having speed-up as its primary objective (" Solve my problem faster... "). The tyranny of Amdahl's Law, combined with architectural convergence to systems that added memory in fixed proportion to processors, led to t...
متن کامل