In this first post, I'd like to start by writing a few background information about WebGPU, as well as a few comparisons with the predecessor webGL.
History of 3D graphics on the web
First steps with VRML and Shockwave 3D
One of the first technologies on the internet to display 3D content was VRML (Virtual Reality Modeling Language) and Adobe Shockwave 3D. These were developed in the late 1990s and were initially the pioneers in this field. This made it possible to describe 3D scenes and interact with them. However, implementation was not easy. The first games were also created with Shockwave 3D and Java applets, but the need for plugins and extensions meant that this technology was not able to establish itself.
First breakthrough with WebGL
In 2011, the specification of WebGL 1.0 was published. WebGL is based on openGL ES 2.0, where ES stands for embedded systems. openGL ES is a library optimized for low-performance hardware, often embedded systems or mobile devices, based on the openGL framework. However, some cutbacks were made to ensure portability and performance. WebGL 2.0 was then released in 2017 and included improved shader usage and is still one of the standards of 3D web technology today. At that time, some very well-known 3D libraries had already been established in Javascript, which make it easy to program games or develop 3D applications. These are widely used today and are often the first port of call when you want to get started.
WebGPU
Since WebGL is still too limited in some places, e.g. it only has vertex and fragment shaders (more on this in later stages), calculations cannot be performed efficiently here. Raytracing and the use of machine learning also had to be improved. With WebGPU, the first implementations by Mozilla (wgpu) and Google (Dawn) were implemented in 2023, which support Vulkan, DirectX12, Metal and OpenGLES as a foundation. In addition, WebGPU is also easy to develop with webassembly; the implementation of wgpu is in Rust and Dawn was implemented in C++. Accordingly, header files are available.
The compute shader has been added as a further shader, which can perform GPU-accelerated calculations in a holistic pipeline. This is extremely useful for particle systems, for example. There are also many other changes that offer more flexibility and possibilities.
Since WebGPU is still under development, it cannot yet be used in all browsers. Currently, only Firefox and Chrome are available, and both have their own implementation, which in some places still differ from the not quite defined standard. Even „WebGPU“ is currently only the working title, although I personally do not expect that this will change.
You can find the current draft of the features and functions at https://www.w3.org/TR/webgpu/. You can also find more examples here.
In the next chapter, we will now start developing with WebGPU. We will start with the basics and check whether the browser is WebGPU compatible.
