Bachelor thesis project for TU Wien by Benedikt Peter.
Note: WebGPU currently requires Chrome Canary with the enable-unsafe-webgpu flag enabled.
- Clone or download the program to your pc.
- Use the command npm install to install the dependencies.
- Compile with one of the the commands: npm run dev, npm run prod, or npm run watch.
- Run the server with an IDE and go to http://127.0.0.1:5500 in Chrome Canary.
To use the benchmarking feature the browser must support timestamp queries. For Chrome this is allowed when launching with the "--disable-dawn-features=disallow_unsafe_apis" flag. After loading the site in a browser that supports timestamp queries, a benchmark can be run by entering the duration in seconds into the "Duration" field and then pressing the "Start Benchmarking" button. After the benchmark is finished you will receive a csv file in which every row contains the start and end times for both the compute pass and the render pass for a frame.