Skip to main content

vis.gl · a GPU compute foundation

GPU compute, from batches to live views.

Build command graphs over chunked data. Keep vectors on the GPU, compose reusable operations, and turn a stream of batches into a living analytical view.

Get started Browse examples
  • WebGPU + WebGL2
  • Zero-copy batches
  • Incremental graphs
Live GPU streamappend · reuse · remove
Starting a live batch view…Requesting a WebGPU device and preparing the graph.
GPUDataFrameIncremental command graphZero-copy batches

The foundation

One graph for every batch.

Treat a single resource or a vector of chunks the same way. The graph tracks ownership, dependencies, and invalidation so new data can reuse the work already done.

Explore the system

See all examples
WebGPU
GPU compute

GPU Data Analysis

Stream batches through reductions, histograms, grids, and grouped aggregates.

Mobile qualityAdvancedStreamingComputeExperimental
WebGPU
Data visualization

Million Row Crossfilter

Brush, filter, and explore a linked GPU-resident dataset without readbacks.

Mobile qualityAdvancedDataInteractiveExperimental
WebGPU
GPU rendering

GPU Frustum Culling

Cull and draw thousands of objects with a command graph on the GPU.

Mobile qualityAdvancedRenderingComputeExperimental
WebGPU
Raster analytics

Raster Lab

Compose raster operations over chunked data and keep intermediate results reusable.

Mobile qualityIntermediateRasterData
WebGPU
GPU data

Arrow Filtering

Filter Arrow columns in place and pass the selected vector to the next stage.

Mobile qualityIntermediateArrowData
WebGPU
Simulation

Fluid Foundry

A real-time fluid simulation built from reusable GPU compute primitives.

Mobile qualityAdvancedSimulationComputeExperimental

Ready to compose?

Start with a vector. End with a view.

Read the docs