Executing Dynamic Data-Graph Computations Deterministically Using Chromatic Scheduling

Chromatic Scheduling of Dynamic Data-Graph Computations

Data-graph computations are a parallel-programming model popularized by programming systems such as Pregel, GraphLab, PowerGraph, and GraphChi. A fundamental issue in parallelizing data-graph computations is the avoidance of races between computation occuring on overlapping regions of the graph. Common solutions such as locking protocols and bulk-synchronous execution often sacrifice performanc...

Maintaining dynamic graph properties deterministically

In this paper we present deterministic fully dynamic algorithms for maintaining several properties on undirected graphs subject to edge insertions and deletions, in polylogarithmic time per operation. Combining techniques from [6, 10], we can maintain a minimum spanning forest of a graph with k different edge weights in O(k log n) amortized time per update; maintain an 1+ -approximation of the ...

Parallel algorithms for scheduling data-graph computations

A data-graph computation — popularized by such programming systems as Pregel, GraphLab, Galois, Ligra, PowerGraph, and GraphChi — is an algorithm that iteratively performs local updates on the vertices of a graph. During each round of a data-graph computation, a user-supplied update function atomically modifies the data associated with a vertex as a function of the vertex’s prior data and that ...

Dynamic scheduling of parallel computations

Structures of parallel programs are usually modeled by task graphs in the scheduling literature. Such graphs are sometimes obtained while compiling the parallel programs. In many other cases, however, they can be determined only at run time. In this paper, we consider the scheduling of parallel computations whose task graphs are generated at run time. We analyze the case where the task graph ha...

Dynamic Scheduling of Substructure Computations