Skip to content

Performance

Where studies run, how to keep a large study quick, and what stays responsive while one runs.

Reference · 1 min read

Wrengler's studies run on your own machine, GPU first, and the panel is built to keep Rhino responsive on large models.

Where studies run#

On Apple Silicon, studies run on the GPU. NVIDIA GPU support is in progress. Everywhere else, studies run on the CPU. If the GPU is missing, out of date, or has a problem during a run, the study moves to the CPU on its own and the answer is the same. Diagnostics shows how to check which one a study used.

Keep a study quick#

  • Start with a coarse grid. A finer grid gives a smoother heatmap, but more sensors take longer. Refine once the pattern looks right.
  • Keep the context focused. On a large urban model, put only the geometry that actually shades your target in the context group.
  • Expect radiation to take longer than sunlight. Each sensor looks at the sun and the whole sky, so start with a wider grid spacing than you would for sunlight.
  • Expect a long period to take longer. A year of hourly sun positions is more work than a single day.

While a study runs#

  • The viewport stays interactive. Heatmaps draw at lower detail while you pan, zoom or rotate, and at full detail when the view settles.
  • Large documents don't freeze Rhino. Opening the panel on a large model keeps Rhino responsive.
  • Many selected objects. If you change the selection while Wrengler is still reading the last one, it moves on to the newest.

The same answer on any hardware#

Every faster way of running a study is checked against one reference, and that reference is checked against physics you can derive by hand. A faster path is used only because it gives the same answer.