Improving MPI applications with a new MPI_Info and the use of the memoization

Authors:
Alejandro Calderón;Jesús Carretero;Felix García-Carballeira;Javier Fernandez;Daniel Higuero;Borja Bergua
Affiliations:
Carlos III University of Madrid, Madrid, Spain;Carlos III University of Madrid, Madrid, Spain;Carlos III University of Madrid, Madrid, Spain;Carlos III University of Madrid, Madrid, Spain;Carlos III University of Madrid, Madrid, Spain;Carlos III University of Madrid, Madrid, Spain
Venue:
Proceedings of the 20th European MPI Users' Group Meeting
Year:
2013

Citing 10
Cited 0

Prolog programming for artificial intelligence

Prolog programming for artificial intelligence
Fast out-of-order processor simulation using memoization

Proceedings of the eighth international conference on Architectural support for programming languages and operating systems
Writing efficient programs

Writing efficient programs
Caching Function Results: Faster Arithmetic by Avoiding Unnecessary Computation

Caching Function Results: Faster Arithmetic by Avoiding Unnecessary Computation
Partial memoization of concurrency and communication

Proceedings of the 14th ACM SIGPLAN international conference on Functional programming
Lock-free parallel dynamic programming

Journal of Parallel and Distributed Computing
Stable deterministic multithreading through schedule memoization

OSDI'10 Proceedings of the 9th USENIX conference on Operating systems design and implementation
Open MPI: a flexible high performance MPI

PPAM'05 Proceedings of the 6th international conference on Parallel Processing and Applied Mathematics
Solving big data challenges for enterprise application performance management

Proceedings of the VLDB Endowment
A Speed-up Technique for an Auto-Memoization Processor by Reusing Partial Results of Instruction Regions

ICNC '12 Proceedings of the 2012 Third International Conference on Networking and Computing

Quantified Score

Hi-index	0.00

Visualization

Abstract

The MPI forum is actively working for a better MPI standard. The results are the new version 3 of the MPI standard, and the efforts for the incoming MPI 3.1/4.0. The technological changes provide many opportunities for improvements and new ideas. This paper introduces two main contributions in this direction: (1) how to improve the MPI_Info object implementation, and (2) a new way of using the former improved MPI_Info object as a storage solution. The MPI_Info object [1] is described as an unordered set of key-value pairs (both key and value are strings, and keys are unique). And it is implemented as a linked list in the two major MPI implementations available. We propose a new MPI_Info object implementation that (1) is based on the use of hash tables (what improves the overall performance), and (2) abstracts the underlying hash table infrastructure (what facilitates the usage of the most appropriated solution). Our proposal opens the possibility of extending the use of the MPI_Info object as a shared storage solution among MPI processes. To demonstrate the capabilities of our proposal, we explore the utilization of the memoization technique on MPI applications in order to improve the execution performance.