An Extensible System for Multilevel Automatic Data Partition and Mapping

Abstract

Abstract-Automatic data distribution is a key feature to obtain efficient implementations from abstract and portable parallel codes. We present a highly efficient and extensible runtime library that integrates techniques for automatic data partition and mapping. It uses a novel approach to define an abstract interface and a plug-in system to encapsulate different types of regular and irregular techniques, helping to generate codes which are independent of the exact mapping functions selected. Currently, it supports hierarchical tiling of arrays with dense and stride domains, that allows the implementation of both data and task parallelism using a SPMD model. It automatically computes appropriate domain partitions for a selected virtual topology, mapping them to available processors with static or dynamic load-balancing techniques. Our library also allows the construction of reusable communication patterns that efficiently exploit MPI communication capabilities. The use of our library greatly reduces the complexity of data distribution and communication, hiding the details of the underlying architecture. The library can be used as an abstract layer for building generic tiling operations as well. Our experimental results show that the use of this library allows to achieve similar performance as carefully-implemented manual versions for several, well-known parallel kernels and benchmarks in distributed and multicore systems, and substantially reduces programming effort.

MARL-Ped+Hitmap: Towards Improving Agent-Based Simulations with Distributed Arrays

“…Hitmap also provides mechanisms for the construction of reusable communication patterns at runtime that adapt to the data partition, creating a low number of aggregated communications. This leads, for example, to a performance efficiency comparable to UPC, with a reduced programming complexity and development effort [7]. Hitmap is used as a runtime system for the Trasgo parallel programming framework [8], that offers an approach similar to PGAS languages.…”

“…Hitmap [7] is a library for the partition, mapping, and management of hierarchically distributed data structures at runtime. It was originally designed for dense arrays, and has been also extended to support sparse data structures, such as sparse matrices or graphs, using the same methodology and interface [6].…”

“…This requirement limits the performance of the simulations for a restricted number of processors that is lesser than the number of agents. On the other hand, Hitmap [7] is a library designed to ease the task of programming parallel applications by using distributed arrays. It includes abstractions for the automatic partitioning and mapping of arrays with arbitrary granularity, as well as the automatic construction of flexible communication patterns adapted to the partition.…”

Exploiting distributed and shared memory hierarchies with Hitmap

“…These models do not promote the different parallelism and optimization techniques needed when processes and threads are hierarchically deployed in a hybrid environment. Comparing with UPC [3], it has been shown that Hitmap reduces the programming complexity and the memory footprint, while obtaining similar efficiency [1]. Another example of PGAS is Chapel, that proposes an interface to add mapping modules to the system [4].…”

“…Our approach is built upon the Hitmap library [1]. Hitmap reduces code development effort performing highly-efficient data distributions and aggregated communications, expressed in an abstract form.…”