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

Performance optimization is an increasingly challenging but often repetitivetask. While each platform has its quirks, the underlying code transformationsrely on data movement and computational characteristics that recur acrossapplications. This paper proposes to leverage those similarities byconstructing an embedding space for subprograms. The continuous space capturesboth static and dynamic properties of loop nests via symbolic code analysis andperformance profiling, respectively. Performance embeddings enable directknowledge transfer of performance tuning between applications, which can resultfrom autotuning or tailored improvements. We demonstrate this transfer tuningapproach on case studies in deep neural networks, dense and sparse linearalgebra compositions, and numerical weather prediction stencils. Transfertuning reduces the search complexity by up to four orders of magnitude andoutperforms the MKL library in sparse-dense matrix multiplication. The resultsexhibit clear correspondences between program characteristics andoptimizations, outperforming prior specialized state-of-the-art approaches andgeneralizing beyond their capabilities.

Performance Embeddings: A Similarity-based Approach to Automatic Performance Optimization