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

Research in automatic parallelization of loop-centric programs started withstatic analysis, then broadened its arsenal to include dynamicinspection-execution and speculative execution, the best results involvinghybrid static-dynamic schemes. Beyond the detection of parallelism in asequential program, scalable parallelization on many-core processors involveshard and interesting parallelism adaptation and mapping challenges. Thesechallenges include tailoring data locality to the memory hierarchy, structuringindependent tasks hierarchically to exploit multiple levels of parallelism,tuning the synchronization grain, balancing the execution load, decoupling theexecution into thread-level pipelines, and leveraging heterogeneous hardwarewith specialized accelerators. The polyhedral framework allows to model,construct and apply very complex loop nest transformations addressing most ofthe parallelism adaptation and mapping challenges. But apart fromhardware-specific, back-end oriented transformations (if-conversion, tracescheduling, value prediction), loop nest optimization has essentially ignoreddynamic and speculative techniques. Research in polyhedral compilation recentlyreached a significant milestone towards the support of dynamic, data-dependentcontrol flow. This opens a large avenue for blending dynamic analyses andspeculative techniques with advanced loop nest optimizations. Selectingreal-world examples from SPEC benchmarks and numerical kernels, we make a casefor the design of synergistic static, dynamic and speculative looptransformation techniques. We also sketch the embedding of dynamic information,including speculative assumptions, in the heart of affine transformation searchspaces.

The Potential of Synergistic Static, Dynamic and Speculative Loop Nest Optimizations for Automatic Parallelization