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

Edge computing and IoT applications are severely constrained by limitedhardware resource. This makes memory consuming DNN frameworks not applicable toedge computing. Simple algorithms such as direct convolution are finding theirway in embedded machine learning. As one of the most widely used platforms forDNN acceleration, GPUs face the bottleneck of on-chip bandwidth. This workintroduces a GPU DNN execution architecture that targets on relieving theon-chip bandwidth bottleneck by reducing data movement through opportunisticcomputing. We first investigate data access patterns in the hardware viewrather than the software view. Then we propose two opportunistic computingtechniques to predictably perform computation when data is available with thehelp of assistant warps. By moving computation to data, our techniques are ableto significantly reduce data movement and relieve the DNN execution bottleneck.Our evaluation results show that the proposed technique can improve DNNapplication performance as much as 55%.

Alleviating Bottlenecks for DNN Execution on GPUs via Opportunistic Computing