wonnx VS learn-wgpu

The two I know of are IREE and Kompute[1]. I'm not sure how much momentum the latter has, I don't see it referenced much. There's also a growing body of work that uses Vulkan indirectly through WebGPU. This is currently lagging in performance due to lack of subgroups and cooperative matrix mult, but I see that gap closing. There I think wonnx[2] has the most momentum, but I am aware of other efforts.

[1]: https://kompute.cc/

[2]: https://github.com/webonnx/wonnx

To a first approximation, Kompute[1] is that. It doesn't seem to be catching on, I'm seeing more buzz around WebGPU solutions, including wonnx[2] and more hand-rolled approaches, and IREE[3], the latter of which has a Vulkan back-end.

[1]: https://kompute.cc/

[2]: https://github.com/webonnx/wonnx

[3]: https://github.com/openxla/iree

There's also a third-party WebGPU implementation: https://github.com/webonnx/wonnx

By experimental I meant e.g. using WGPU to run compute shaders like wonnx, which is working fine but only on a very restricted set of devices and browsers.

https://news.ycombinator.com/item?id=35696031 ... TIL about wonnx: https://github.com/webonnx/wonnx#in-the-browser-using-webgpu...

microsoft/onnxruntime: https://github.com/microsoft/onnxruntime

Apache/arrow has language-portable Tensors for cpp: https://arrow.apache.org/docs/cpp/api/tensor.html and rust: https://docs.rs/arrow/latest/arrow/tensor/struct.Tensor.html and Python: https://arrow.apache.org/docs/python/api/tables.html#tensors https://arrow.apache.org/docs/python/generated/pyarrow.Tenso...

Fwiw it looks like the llama.cpp Tensor is from ggml, for which there are CUDA and OpenCL implementations (but not yet ROCm, or a WebGPU shim for use with emscripten transpilation to WASM): https://github.com/ggerganov/llama.cpp/blob/master/ggml.h

Are the recommendable ways to cast e.g. arrow Tensors to pytorch/tensorflow?

FWIU, Rust has a better compilation to WASM; and that's probably faster than already-compiled-to-JS/ES TensorFlow + WebGPU.

What's a fair benchmark?

wonnx has done some fantastic work in this regard, so that's where we plan to start once we get there. In terms of general discussion of alternate backends, see this issue.

> GPU in other ways, such as training ML models and then using them via an inference engine all powered by your local GPU?

Have a look at wonnix https://github.com/webonnx/wonnx

A WebGPU-accelerated ONNX inference run-time written 100% in Rust, ready for native and the web

Looking forward to your WebGPU ML runtime! Also, why not contribute back to WONNX? (https://github.com/webonnx/wonnx)

You can indeed perform inference using WebGPU (see e.g. [1] for GPU-accelerated inference of ONNX models on WebGPU; I am one of the authors).

The point made above is that WebGPU can only be used for GPU's and not really for other types of 'neural accelerators' (like e.g. the ANE on Apple devices).

[1] https://github.com/webonnx/wonnx

Nice.

Speaking of Snake game, if you want to go even deeper, you can try to use the wgpu crate to combine Rust and WebGPU to write everything from scratch. Here is the tutorial:

https://sotrh.github.io/learn-wgpu/#what-is-wgpu

I once wrote a code editor with wgpu, from font rendering to char/line state management (very rough) for music live coding:

https://github.com/glicol/glicol-wgpu

It runs in browsers, even including Safari!

https://sotrh.github.io/learn-wgpu/ This is probably the best resource out there for learning wgpu specifically. If you're unfamiliar with graphics, the learnopengl one is good. If you've got experience though, jumping right into that one is a shout or looking at some vulkan ones as they're pretty similar in terms of architecture.

- https://sotrh.github.io/learn-wgpu

Most of The WGPU I learnt is from https://sotrh.github.io/learn-wgpu/ but it doesn't really talk about designing n stuff, I thought of checking out the source code for Bevy or even games like veloren. But well, their codebases are pretty big to get started in the first place.

So I was wondering if opting for option 1 would be better to begin with. OpenGL has a much bigger community and wgpu only has its documentation which I hear is not quite up there yet. There is this excellent tutorial for wgpu that I read through, but it seems like wgpu can be a lot more complicated than starting with OpenGL.

I started with wgpu tutorial (https://sotrh.github.io/learn-wgpu/) since I like the idea of portability and it's a Rust-first library, but it seems I'm missing some foundations of how CG works in general: the code is given, a little of explanation like it assumes I already know something, maybe I'm wrong, but I wish there was a longer explicit version.

If you haven't seen it: https://sotrh.github.io/learn-wgpu/ is a good introduction that will explain most of what you asked, then can refer to rend3d or bevys renderer to see how a render graph works.