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 Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

GPU-embedded systems have gained popularity across various domains due totheir efficient power consumption. However, in order to meet the demands ofreal-time or time-consuming applications running on these systems, it iscrucial for them to be tuned to exhibit high performance. This paper addressesthe issue by developing and comparing two tuning methodologies on GPU-embeddedsystems, and also provides performance insights for developers and researchersseeking to optimize applications running on these architectures. We focus onparallel prefix operations, such as FFT, scan primitives, and tridiagonalsystem solvers, which are performance-critical components in many applications.The study introduces an analytical model-driven tuning methodology and aMachine Learning (ML)-based tuning methodology. We evaluate the performance ofthe two tuning methodologies for different parallel prefix implementations ofthe BPLG library in an NVIDIA Jetson system, and compare their performance tothe ones achieved through an exhaustive search. The findings shed light on thebest strategies for handling the open challenge of performance portability formajor computational patterns among server and embedded devices, providingpractical guidance for offline and online tuning. We also address the existinggap in performance studies for parallel computational patterns in GPU-embeddedsystems by comparing the BPLG performance against other state-of-the-artlibraries, including CUSPARSE, CUB, and CUFFT.

Performance Tuning for GPU-Embedded Systems: Machine-Learning-based and Analytical Model-driven Tuning Methodologies