Comparison of Static and Dynamic Resource Allocation Strategies for Matrix Multiplication

Olivier Beaumont 1, 2 Lionel Eyraud-Dubois 1, 2 Abdou Guermouche 1, 3 Thomas Lambert 1, 2
2 Realopt - Reformulations based algorithms for Combinatorial Optimization
LaBRI - Laboratoire Bordelais de Recherche en Informatique, IMB - Institut de Mathématiques de Bordeaux, Inria Bordeaux - Sud-Ouest
3 HiePACS - High-End Parallel Algorithms for Challenging Numerical Simulations
LaBRI - Laboratoire Bordelais de Recherche en Informatique, Inria Bordeaux - Sud-Ouest
Abstract : The tremendous increase in the size and heterogeneity of supercomputers makes it very difficult to predict the performance of a scheduling algorithm. In this context, relying on purely static scheduling and resource allocation strategies, that make scheduling and allocation decisions based on the dependency graph and the platform description, is expected to lead to large and unpredictable makespans whenever the behavior of the platform does not match the predictions. For this reason, the common practice in most runtime libraries is to rely on purely dynamic scheduling strategies, that make short-sighted scheduling decisions at runtime based on the estimations of the duration of the different tasks on the different available resources and on the state of the machine. In this paper, we consider the special case of Matrix Multiplication, for which a number of static allocation algorithms to minimize the amount of communications have been proposed. Through a set of extensive simulations, we analyze the behavior of static, dynamic, and hybrid strategies, and we assess the possible benefits of introducing more static knowledge and allocation decisions in runtime libraries.
Complete list of metadatas

Cited literature [27 references]  Display  Hide  Download

https://hal.inria.fr/hal-01163936
Contributor : Olivier Beaumont <>
Submitted on : Thursday, October 15, 2015 - 7:28:30 PM
Last modification on : Tuesday, May 7, 2019 - 11:42:03 AM
Long-term archiving on : Wednesday, April 26, 2017 - 11:58:24 PM

File

PID3840069 (1).pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01163936, version 2

Citation

Olivier Beaumont, Lionel Eyraud-Dubois, Abdou Guermouche, Thomas Lambert. Comparison of Static and Dynamic Resource Allocation Strategies for Matrix Multiplication. 26th IEEE International Symposium on Computer Architecture and High Performance Computing (SBAC-PAD), 2015, Oct 2015, Florianopolis, Brazil. ⟨hal-01163936v2⟩

Share

Metrics

Record views

394

Files downloads

1060