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

Generic matrix multiplication (GEMM) and one-dimensionalconvolution/cross-correlation (CONV) kernels often constitute the bulk of thecompute- and memory-intensive processing within image/audio recognition andmatching systems. We propose a novel method to scale the energy and processingthroughput of GEMM and CONV kernels for such error-tolerant multimediaapplications by adjusting the precision of computation. Our technique employslinear projections to the input matrix or signal data during the top-level GEMMand CONV blocking and reordering. The GEMM and CONV kernel processing then usesthe projected inputs and the results are accumulated to form the final outputs.Throughput and energy scaling takes place by changing the number of projectionscomputed by each kernel, which in turn produces approximate results, i.e.changes the precision of the performed computation. Results derived from avoltage- and frequency-scaled ARM Cortex A15 processor running face recognitionand music matching algorithms demonstrate that the proposed approach allows for280%~440% increase of processing throughput and 75%~80% decrease of energyconsumption against optimized GEMM and CONV kernels without any impact in theobtained recognition or matching accuracy. Even higher gains can be obtained ifone is willing to tolerate some reduction in the accuracy of the recognitionand matching applications.

Precision-Energy-Throughput Scaling Of Generic Matrix Multiplication and Convolution Kernels Via Linear Projections