English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

We present Atos, a task-parallel GPU dynamic scheduling framework that isespecially suited to dynamic irregular applications. Compared to the dominantBulk Synchronous Parallel (BSP) frameworks, Atos exposes additional concurrencyby supporting task-parallel formulations of applications with relaxeddependencies, achieving higher GPU utilization, which is particularlysignificant for problems with concurrency bottlenecks. Atos also offersimplicit task-parallel load balancing in addition to data-parallel loadbalancing, providing users the flexibility to balance between them to achieveoptimal performance. Finally, Atos allows users to adapt to different use casesby controlling the kernel strategy and task-parallel granularity. Wedemonstrate that each of these controls is important in practice. We evaluateand analyze the performance of Atos vs. BSP on three applications:breadth-first search, PageRank, and graph coloring. Atos implementationsachieve geomean speedups of 3.44x, 2.1x, and 2.77x and peak speedups of 12.8x,3.2x, and 9.08x across three case studies, compared to a state-of-the-art BSPGPU implementation. Beyond simply quantifying the speedup, we extensivelyanalyze the reasons behind each speedup. This deeper understanding allows us toderive general guidelines for how to select the optimal Atos configuration fordifferent applications. Finally, our analysis provides insights for futuredynamic scheduling framework designs.

Atos: A Task-Parallel GPU Dynamic Scheduling Framework for Dynamic Irregular Computations