WebGPU charting library. MIT licensed.
Built for dashboards.
WebGPU charting library for large datasets, real-time streaming, multi-chart dashboards, and 3D series. Zero runtime dependencies.
GPU-side sampling and binning for million-point scatter, plus heatmaps, candlesticks, and multi-series timeseries. MIT licensed.
One million points. Binned on the device.
Streaming mountain fills.
Dozens of translucent area series appending every frame with GPU rendering.
Negative + positive stacks
Live stacked bars with positive and negative values on WebGPU.
Ready to create your own?
Full example index
Interactive ChartGPU demos: streaming dashboards, candlesticks, scatter density, heatmaps, 3D point clouds, stacked area, and more.
FAQ
Questions about ChartGPU
Short answers for engineers evaluating WebGPU charting libraries for streaming, multi-chart dashboards, and high-rate series.
What is ChartGPU?
ChartGPU is an open-source TypeScript charting library built on WebGPU. It is built for streaming multi-series dashboards, million-point scatter density, heatmaps, candlesticks, and multi-chart boards—by rendering and sampling on the GPU instead of Canvas 2D or SVG.
How do I install ChartGPU?
Run npm install @chartgpu/chartgpu, then create a chart with ChartGPU.create(container, options).
Requires a browser with WebGPU support (Chrome or Edge 113+, Safari 18+).
Package: @chartgpu/chartgpu on npm.
Is ChartGPU free and open source?
Yes. ChartGPU is MIT licensed and fully open source. Source lives on GitHub (ChartGPU/ChartGPU).
What browsers support ChartGPU?
ChartGPU requires WebGPU: Chrome or Edge 113 and later, or Safari 18 and later. Firefox support depends on platform and flags. Without WebGPU, live GPU demos will not render.
How is ChartGPU different from Chart.js, ECharts, or D3?
ChartGPU is WebGPU-native. It uploads series to GPU buffers, can decimate large line series with compute shaders, and bins scatter density on the device—paths suited to large series and multi-chart boards where Canvas 2D and SVG paths often struggle. See the live demos and performance guide. Chart.js, many ECharts configs, and D3 typically rely on Canvas 2D, SVG, or CPU sampling paths better suited to smaller series.
What chart types does ChartGPU support?
Line, area (including stacked mountain and step), bar, scatter, scatter density, candlestick, OHLC, pie/donut, uniform heatmap/spectrogram, band/range, error bars, impulse/stem, plus 3D point cloud and 3D surface when coordinateSystem is cartesian3d.
Can ChartGPU handle streaming and real-time data?
Yes. ChartGPU is built for streaming: appendData with optional maxPoints FIFO, auto-scroll, multi-series live dashboards, and GPU-side paths so you can keep appending without full re-uploads every frame for eligible line series.