The main advantage of WebGL is that cross-device, cross-platform and zero-install. It’s the new revolution for 3D entertainment on the Internet. It’s the future.

Javascript with CUDA: Elevating JavaScript Performance Through GPU Power
http://mozillalabs.com/jetpack/2010/01/25/elevating-javascript-performance-through-gpu-power/
It’s possible to utilize CUDA from within Jetpack now. (See the Jetpack-to-CUDA project)

Jetpack.toGPU() is that hint. Under the hood of Jetpack.toGPU() there is translation of function(nums, numsSize) {…} to CUDA Kernel and a method of sending C-code to the GPU. Jetpack.toGPU() should be non-blocking. It means “some job” will be done while resNumbers is calculating. Of course, a developer could write their own CUDA Kernel in C to be sent to the GPU. This technique is like in similar to the PyCUDA project.
WebGL on Mobile: Nokia N900 impressively demos WebGL 3D graphics
http://www.engadget.com/2009/12/03/nokia-n900-impressively-demos-webgl-3d-graphics/

Start to Learn WebGL Now? WebGL Lesson
http://learningwebgl.com/blog/?p=28