Skip to content

Overview

Martin Hartl edited this page Nov 7, 2017 · 3 revisions

Reference Overview

The AllScale API provides two important concepts for achieving large-scale parallel performance:

  • data structures that are distributed-memory-enabled
  • parallel operators that offer efficient means of performing common operations in parallel

Besides that, a number of frequently-used utilities such as containers are offered to increase the productiveness of application developers.

Utilities

Name Description
Bag Unordered container (also known as a multiset)
Vector An n-dimensional array of values

Data Structures

Name Description
Grid Regular n-dimensional array, size defined dynamically at creation
AdaptiveGrid Refinable n-dimensional array
Mesh Multi-layered representation of a graph structure with multiple node and edge types

Parallel Operators

Name Description
Pfor Iterates over a given range of elements and applies a given function in parallel
Stencil Iteratively applies a given update function to an n-dimensional array of elements in parallel
VCycle Iteratively applies a given update function on a mesh in a multi-grid fashion
Clone this wiki locally