Skandium: Multi-core Programming with Algorithmic Skeletons

Abstract

This paper argues that algorithmic skeletons are a suitable programming model for multi-core architectures. The high-level abstractions offered by algorithmic skeletons provide a simple way for non-parallel programmers to address parallel programming.Previous algorithmic skeleton frameworks and libraries have addressed distributed computing environments such as Clusters and Grids. This paper proposes a parallel skeleton library, Skandium; and concludes, after an experimental evaluation, that algorithmic skeletons are an effective methodology to program multi-core architectures.

Fastflow: High‐Level and Efficient Streaming on Multicore

“…2. FastFlow high-level patterns appear in various algorithmic skeleton frameworks, including Skandium [22], Muesli [10] and Muskel [3]. The parallel design patterns presented in [23] also include equivalents of the FastFlow high-level patterns.…”

Parallel Haskell implementations of the N‐body problem

“…Microsoft's new Dryad system [24] takes a similar approach, extending its Daytona system [13] for iterative MapReduce patterns. Many systems for skeleton-based parallelism have been developed, and the most notable of these systems are eSkel [3], Müsli [10], Skandium [29] and P 3 L [1]. A good, up-to-date survey on skeletons is given in [23].…”

Structured Data Access Annotations for Massively Parallel Computations

“…In contrast with many skeleton frameworks (including SkeTo [16], Muesli [9] and SkePU) which consider only maps over "collection" input data, we assume the one used in P3L [7] and Skandium [15]: the responsibility for specifying how the subtask items are build out of the input data (set) is left to the application programmer, as is the specification of the re-construction of the result from the collection of partial results. In P3L, the programmer is asked to use the formal parameters of the map as actual parameters of the worker skeleton using "star variables"-a kind of ∀i variable-to establish correspondences between the task and the subtask data items.…”