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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

In recent years, deep neural networks (DNNs), have yielded strong results ona wide range of applications. Graphics Processing Units (GPUs) have been onekey enabling factor leading to the current popularity of DNNs. However, despiteincreasing hardware flexibility and software programming toolchain maturity,high efficiency GPU programming remains difficult: it suffers from highcomplexity, low productivity, and low portability. GPU vendors such as NVIDIAhave spent enormous effort to write special-purpose DNN libraries. However, onother hardware targets, especially mobile GPUs, such vendor libraries are notgenerally available. Thus, the development of portable, open, high-performance,energy-efficient GPU code for DNN operations would enable broader deployment ofDNN-based algorithms. Toward this end, this work presents a framework to enableproductive, high-efficiency GPU programming for DNN computations acrosshardware platforms and programming models. In particular, the frameworkprovides specific support for metaprogramming, autotuning, and DNN-tailoreddata types. Using our framework, we explore implementing DNN operations onthree different hardware targets: NVIDIA, AMD, and Qualcomm GPUs. On NVIDIAGPUs, we show both portability between OpenCL and CUDA as well competitiveperformance compared to the vendor library. On Qualcomm GPUs, we show that ourframework enables productive development of target-specific optimizations, andachieves reasonable absolute performance. Finally, On AMD GPUs, we show initialresults that indicate our framework can yield reasonable performance on a newplatform with minimal effort.

A Metaprogramming and Autotuning Framework for Deploying Deep Learning Applications