5 ms·

Looks cool. Any inherent limitations or performance issues with graphs that are very large? (20M+ elements)

• samatman 4y ago
  Why uh. Would you expect that to work? How much of your life were you planning to dedicate to scrolling from one side of the image to another?
  • guhidalg 4y ago
    I think he expects to be able to zoom out.
    • rgrieselhuber 4y ago
      Yup for zoomed out views. It’s handy when visualizing large websites.
• mkl95 4y ago
  Unless you use instanced rendering via some GPU API it's going to take some galaxy brain code to make that work.
  • zero_shift 4y ago
    Forget about rendering, just doing the layout would be extremely challenging.
    • stragies 4y ago
      Some of the graphs (near the bottom) of the GraphViz example page are quite large: https://graphviz.org/gallery/ https://graphviz.org/gallery/ And there's a link to a gallery with even bigger ones, so #NumberOfNodes alone does not seem to be the main limiting metric. Maybe combined with the choice of rendering layout.
  • lmeyerov 4y ago
    We do webgl rendering of big graphs as part of Graphistry, and before the 20M nodes break the system, the ~10X more edges will ;+) But more seriously, a careful webgl implementation gets you maybe halfway there, and most I've seen break well before then. It gets tricky with browser js memory limits and raw perf. We are looking into hitting more 100M level with interactive speeds... But takes creativity & opts I wouldn't expect for a diagramming tool :) and, if your kind of thing, we are hiring :)
• alixanderwang 4y ago
  we're going to have to fallback to stable diffusion for that one