GPU profiling for WebGPU — measure compute pass timing, memory usage, and pipeline performance in the browser.
- Compute pass timing — precise GPU timestamps for WebGPU compute passes
- Memory tracking — buffer allocation and usage monitoring
- Pipeline analysis — compare shader performance across configurations
- Browser-native — runs entirely in WebGPU, no native dependencies
# Serve the profiler
python -m http.server 8000
# Open in a WebGPU-capable browserPerformance tooling for the SuperInstance conservation-spectral-webgpu pipeline. Profiles the GPU-accelerated spectral computations that run in the browser.
MIT