Wondering what’s next for npm?Check out our public roadmap! »

    This package has been deprecated

    Author message:

    Package voids license rights

    @q42/gpu-power-estimation

    1.1.0 • Public • Published

    Borrel WebGL Power Estimation

    Catering a 3d web-based experience to the power of a target platform is difficult with such a small amount of information available about the current graphics hardware. This utility aims to provide performance benchmark, clock speed, and memory information about the current hardware by guessing the type of graphics hardware using webgl UNMASKED_RENDERER_WEBGL parameter and looking it up in a cached list of hardware.

    GPU benchmark and spec information scraped from videocardbenchmark.net, techpowerup.com, and notebookcheck.net.

    See your hardware info here!

    License Information

    Code in the /src directory is licensed under the MIT license and free to use. Scraped graphics card data in the /data directory is for demonstration purposes only and is subject to the terms of the respective source websites.

    Matching the Graphics Hardware Name

    The values from UNMASKED_RENDERER_WEBGL are irregular and relatively unpredictable. To find the corresponding hardware in the database list we

    • Check if there is a version number in the current GPU hardware name.
    • Filter the database of GPUs to those that include that version number (or to those that have no version number if none was found).
    • From that list pick the hardware the has the most matching tokens between the names.

    Use

    import { getDetailedInfo, getBasicInfo } from 'gpu-power-estimate';
     
    fetch( './path/to/database.json' )
      .then( res => res.json() )
      .then( database => {
     
        // get the hardware information
        const canvas = document.createElement( 'canvas' );
        const gl = canvas.getContext( 'webgl' );
        const basicInfo = getBasicInfo( gl );
        const detailedInfo = getDetailedInfo( database, gl );
     
        // scale the application
        const capability = detailedInfo ? detailedInfo.performance : 0;
        if ( capability > 6000 ) {
     
          // initialize highest fidelity scene
     
        } else if ( capability > 3000 ) {
     
          // initialize moderately complex scene
     
        } else {
     
          // initialize simplified scene
     
        }
     
      } );
     

    API

    getBasicInfo

    getBasicInfo( context = null : WebGLContext ) : Object | null

    Returns some basic info about the hardware based on the WEBGL_debug_renderer_info extension or null if it's unavailable.

    If the context is not provided then a temporary one will be created.

    {
      // The full graphics hardware
      name,
     
      // A guess as to whether or not the hardware is integrated graphics
      integrated,
     
      // The raw unmasked fields returned from the extension
      unmasked: { vendor, renderer }
     
    }

    getDetailedInfo

    getDetailedInfo(
      database : Object,
      contextOrCard = null : WebGLContext | string
    ) : Object | null

    Returns more detailed hardware information based on the information in the provided database. The database is expected to be an object where the keys are the names of graphics hardware and the values are objects with detail information.

    If a WebGL context or card name to search is not provided then a temporary context will be created.

    The scraped example data provides the following information. Fields are null if unavailable. If no card is found or WEBGL_debug_renderer_info is unavailable then null is returned.

    {
     
      // Name of the GPU
      name,
     
      // Type of graphics card
      // Workstation, Desktop, Mobile, Unknown
      type,
     
      // The 3d and 2d performance results as provided by the
      // PassMark g3d and g2d benchmarks
      performance,
     
      // thermal design power
      tdp,
     
      // total available vram in MB
      memory,
     
      // clock speeds in MHz
      clock,
      memoryClock
     
    }

    Caveats

    For privacy reasons the availability of the UNMASKED_RENDERER_WEBGL parameter may be disabled in which case no estimate can be provided.

    References

    Keywords

    none

    Install

    npm i @q42/gpu-power-estimation

    DownloadsWeekly Downloads

    1

    Version

    1.1.0

    License

    MIT

    Unpacked Size

    4.16 MB

    Total Files

    23

    Last publish

    Collaborators

    • avatar
    • avatar
    • avatar
    • avatar
    • avatar
    • avatar
    • avatar
    • avatar
    • avatar
    • avatar
    • avatar
    • avatar
    • avatar
    • avatar
    • avatar
    • avatar
    • avatar
    • avatar
    • avatar