APerformancePortabilityFrameworkforPython Nader Al Awar Steven Zhu nader.alawar@utexas.edu stevenzhu@utexas.edu TheUniversity of Texas at Austin TheUniversity of Texas at Austin Austin, Texas, USA Austin, Texas, USA George Biros Milos ...
Filetype PDF | Posted on 03 Feb 2023 | 2 years ago
The words contained in this file might help you see if this file matches what you are looking for:
...Aperformanceportabilityframeworkforpython nader al awar steven zhu alawar utexas edu stevenzhu theuniversity of texas at austin usa george biros milos gligoric gbiros acm org abstract introduction kokkosis a programming model for writing performance portable traditionally parallel high code scientific applica applications all major computing platforms tions is written in low level architecture specific it provides abstractions data management and common par hpc frameworkssuchasopenmp cuda allel operations allowing developers to write per others these frameworks require that the user be aware formance with minimal knowledge details order efficient kokkos implemented as heavily templated c library example optimal layout two dimensional array differs however not ideal rapid prototyping quick across different hardware devices row on cpu openmp gorithmic exploration an increasing number use enable cached memory accesses vs column gpu python machine learning ana coalesced additionally each l...