Arbitrary Computation on the GPU Using WebGL
summary
Shows how WebGL encodes arbitrary data as textures, runs math in GLSL shaders, and returns results via canvas.getImageData; demos include a 50,000‑particle renderer at 60fps and matrix multiply scaled to 1024×1024 where GPU beats CPU once the serialization overhead is amortized. Calls out two hard limits: readback is very expensive and mobile GPUs show cumulative precision errors, and current WebGL GLSL has no bitwise ops, so hash/Bitcoin‑style workloads and many low‑level algorithms don’t work.
AI-written summary · about
chapters (13)
- 0:00Introduction and speaker intro
- 1:48What is WebGL and how it works
- 4:39Edge detection with WebGL
- 6:49Particles demo and GPU power
- 8:57Limitations of WebGL and alternatives
- 10:42Matrix multiplication on GPU
- 13:12UX tips and GPU acceleration
- 16:06WebGL shader challenges
- 17:51Performance comparison CPU vs GPU
- 19:21Big O notation and algorithm efficiency
- 20:45Potential applications of GPU computing
- 21:49Current blockers for GPU computing
- 23:32Future solutions and summary
related talks
24mArbitrary Computation on the GPU Using WebGL
23mArbitrary Computation on the GPU Using WebGL
22mWebGL For More Than Graphics: Is It Possible?
05mA Brief Intro to GPU Shader Programming in WebGL
34mNext-Generation 3D Graphics on the Web
25mHow to Program Your GPU (and Not Do It Alone)
39mThe Pixel Factory
26mCreative experiments with WebGL shaders in JavaScript