figshare
Browse
poster.pdf (527.75 kB)

peachpyio-web-app-for-performance-tuning.pdf

Download (527.75 kB)
poster
posted on 2017-02-28, 06:18 authored by Marat DukhanMarat Dukhan
As the progress in transistor scaling slows down, low-level performance tuning becomes an important tool to accelerate compute-intensive codes. However, the community of experts who could optimize the codes on low-level is tiny, and their ability to contribute optimizations to open-source software is often restricted by lack of access to relevant hardware. We present PeachPy.io, and IDE for low-level performance tuning that works inside a Web browser. On PeachPy.io developers can modify assembly kernels, run them on different hardware platforms, analyze performance bottlenecks through hardware performance counters, and compare against alternatives without any time-consuming client-side configuration. PeachPy.io crowd-sources performance tuning, and lets anyone interested try their skills in low-level optimization without spending time on local software configuration or investing financial resources to acquire hardware.

Poster presented at SIAM CSE17 PP108 Minisymposterium: Software Productivity and Sustainability for CSE and Data Science

Funding

U.S. National Science Foundation (NSF) Award Number 1339745

History

Usage metrics

    Licence

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC