Vulkan-Docs VS WASI

Pipelines (or in general terms PSOs) are the most problematic aspect of Vulkan / DX12 - much more than synchronization! Large parts of the gamedev industry seems to recognize all the performance issues with pipelines and therefore companies are experimenting with newer models like the VK_EXT_shader_object extension ("Vulkan without Pipelines": https://www.khronos.org/blog/you-can-use-vulkan-without-pipe...).

I've written a detailed comment about this before here (https://news.ycombinator.com/item?id=37843946#37845431) but for a much more comprehensive explanation by an engineer from Nintendo read the initial proposal for the VK_EXT_shader_object extension: https://github.com/KhronosGroup/Vulkan-Docs/blob/main/propos...).

There's also Casey Muratori's mail to the Vulkan advisory on 2015 that basically predicts this whole clusterfuck would happen: https://github.com/cmuratori/misc/blob/main/vulkan_dynamic_s...

Apple -> MoltenVK is an emulation layer and doesn't give you as much control as using Metal directly.

Nintendo and Sony prefer their own APIs, NVN and LibGNM, and AIUI Vulkan is a second-class API on those platforms which does not offer as much power, it is widely understood studios use NVN and LibGNM to get access to the real hardware on those platforms.

Windows/AMD/NVidia/Intel -> HW manufacturers tend to prototype and release new features with D3D first and then 'backport' them to Vulkan after a while. DirectX 12 for example had mesh shaders for over 2 years before Vulkan got a vendor neutral extension for them[0]

Android and Linux are the only platform where Vulkan is a first-class citizen.

You could maybe argue Nvidia treats Vulkan as a first-class citizen because they tend to have vendor-specific Vulkan extensions for the latest features available before anyone else. But otherwise, no, Vulkan is not a first-class API anywhere except Linux and Android.

Graphics API wars are alive and well.

[0] https://github.com/KhronosGroup/Vulkan-Docs/issues/1423

WASI Co-chair here. Nothing in WASI is "somehow blocked by Google", or indeed blocked by anyone at all. Graphics support in WASI hasn't been developed simply because nobody has put energy into developing graphics support in WASI.

At the end of 2023 we counted around 40 contributors who have been working on WASI specifications and implementations: https://github.com/WebAssembly/meetings/blob/main/wasi/2023/... . That is a great growth for our project from a few years ago when that issue was filed, but as you can see from what people are working on, its all much more foundational pieces than a graphics interface. Also, if you look at who is employing those contributors, its largely vendors who are interested in WASI in the context of serverless. That doesn't mean WASI is limited to only serverless, but that has been the focus from contributors so far.

By rolling out WASI on top of the WASM Component Model we have built a sound foundation for creating WASI proposals that support more problem domains, such as embedded systems (@mc_woods and his colleagues are helping with this), or graphics if someone is interested in putting in the work. Our guide to how to create proposals is found here: https://github.com/WebAssembly/WASI/blob/main/Contributing.m... .

> As I understand it, it's not even really possible today to make WebAssembly do anything meaningful in the browser without trampolining back out to JavaScript anyway, which seems like a remarkable missed opportunity.

That's the underlying messy API it's built on. There are specs to make the API more standardized like https://github.com/WebAssembly/WASI

But overall, yeah, it feels like a shiny new toy everyone is excited about and wants to use. Some toys can be fun to play with, but it doesn't mean we have to rewrite production systems in it. Sometimes, or most of the time, toys don't become useful tools.

Snapshot Preview 1 is the standard all tools are building to right now. The specification is available here: https://github.com/WebAssembly/WASI/blob/main/legacy/preview...

It's pretty unreadable though!

Preview 2 looks like it will be a big change, and is just being finalised at the moment. I'd expect that when preview 2 is available there will be an improvement in the quality of documentation. I'm not sure how long it will take after release for tools to start switching to it. I'd expect Preview 1 will still be the main target at least for the rest of this year.

> Like WTF does this mean? The repo tells me nothing

Directly above the sentence you quoted:

"Interposition in the context of WASI interfaces is the ability for a Webassembly instance to implement a given WASI interface, and for a consumer WebAssembly instance to be able to use this implementation transparently. This can be used to adapt or attenuate the functionality of a WASI API without changing the code using it."

> and I've still yet to see a clear write-up about what WASI is.

In the same document: [0]

> WTF is wit?

The first link in that document ("Starting in Preview2, WASI APIs are defined using the Wit IDL.") is [1].

> I click on "legacy" and I see preview0 and preview1, which are basically unreadable proto-specs.

The README for the legacy directory [2] clearly explains what they are.

> Where's a single well-written WASI spec?

"Development of each API happens in its own repo, which you can access from the proposals list." [3]

> Whatever WASI is doing, I don't like it.

Clearly not - you've gone out of your way to ignore all of the documentation that answers your questions.

> And neither does AssemblyScript team apparently

The AssemblyScript team have a bone to pick with WASI based on their misunderstanding of what WASI is for (it is not intended for use on the web) and WASI's disinterest in supporting UTF-16 strings. You can see for yourself in [4].

[0]: https://github.com/WebAssembly/WASI/tree/main#wasi-high-leve...

The Bytecode Alliance initiated a sub-project called the WebAssembly System Interface (WASI). WASI is an API that allows WebAssembly access to system features such as files, filesystems, Berkeley sockets, clocks, and random numbers. WASI acts as a system-level interface for WebAssembly, so incorporating a runtime into a host environment and building a platform is easier.

We are excited to contribute back to Wasmtime and the component model, as well as to new projects and proposals emerging in this space (such as new Wasm proposals, like WASI Preview 2, wasi-keyvalue, wasi-sql or wasi-cloud).

I've been reading the following repositories.

https://github.com/WebAssembly/WASI/blob/main/Proposals.md