Spike: WebGL Performance for Very Large Models (MSFT)

[mavimaec]

trafficstars

NOTES

• [ ] How do we fail gracefully when the requested set of displayable objects exceeds current memory limits?
  • [ ] How do we avoid (or recover gracefully from) out-of-memory errors?
  • [ ] How do we track our memory consumption?
    • [ ] Is there a browser API to determine the current memory consumption?
    • [ ] Is there a library which helps us manage memory?
• [ ] Can we create an HTTP request which asks for a collection of buffer subsections?
  • [ ] Can an HTTP response be written to a span inside an array buffer?
• [ ] Should we add VertexOffset attribute to g3d?

Test Files

The goal is to load the following files successfully in the WebGL viewer:

• kahua_navis_dedup.vim.gz (vertex buffer currently exceeds ArrayBuffer allocation limits)
• TODO: add more test files