Installation#

Requirements#

  • Python 3.10+

  • NGSolve (see https://ngsolve.org)

  • A browser with WebGPU support: Chrome/Edge ≥ 113, Firefox Nightly with dom.webgpu.enabled, or Safari Technology Preview.

Install#

pip install ngsolve_webgpu

For the latest development version:

pip install git+https://github.com/CERBSim/ngsolve_webgpu.git

Building the documentation#

The documentation embeds interactive scenes by setting WEBGPU_EXPORTING=1 during the Sphinx build. A headless Chromium (via Playwright) is launched to provide a real WebGPU device for capturing GPU buffers.

pip install sphinx nbsphinx pydata-sphinx-theme playwright
playwright install chromium
cd docs
WEBGPU_EXPORTING=1 sphinx-build -b html . _build/html

Open _build/html/index.html in a WebGPU-capable browser.

Verifying WebGPU#

Visit https://webgpureport.org to confirm your browser exposes a WebGPU adapter. If no adapter is reported, the embedded canvases will display an error message.