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

Emerging deep learning workloads urgently need fast general matrixmultiplication (GEMM). To meet such demand, one of the critical features ofmachine-learning-specific accelerators such as NVIDIA Tensor Cores, AMD MatrixCores, and Google TPUs is the support of mixed-precision enabled GEMM. For DNNmodels, lower-precision FP data formats and computation offer acceptablecorrectness but significant performance, area, and memory footprintimprovement. While promising, the mixed-precision computation on errorresilience remains unexplored. To this end, we develop a fault injectionframework that systematically injects fault into the mixed-precisioncomputation results. We investigate how the faults affect the accuracy ofmachine learning applications. Based on the error resilience characteristics,we offer lightweight error detection and correction solutions thatsignificantly improve the overall model accuracy if the models experiencehardware faults. The solutions can be efficiently integrated into theaccelerator's pipelines.

MPGemmFI: A Fault Injection Technique for Mixed Precision GEMM in ML Applications