kompute VS nickel

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

Great to see the positive momentum of this framework! Best wishes and upvotes from the Vulkan Kompute team :)

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

I look in an example and see similar instructions, stating that the build should be quite simple. But again, it doesn't work. It generates a bunch of folders with Visual Studio stuff, but no executables, no libraries, or anything like that.

I can't figure out how, and there are no tutorials. According to https://kompute.cc/overview/build-system.html I should simply run "cmake -Bbuild". But this doesn't output what I need, and when I look in the Makefile I get the sense that this is more an example Makefile... but then that contradicts the above tutorial.

Kompute is in my opinion good example to take inspiration for abstractions.

Currently, there's a few Vulkan compute frameworks floating around (like Kompute). I would work with those. Kompute simplifies a lot of the biolerplate and seems like you could benefit from using it.

Try Kompute, a project from the Linux foundation. It is quite simple to use, and does not require deep knowledge of graphics API. It’s a bit painful to setup, but it kinda works well (and I have a project going on on it)

This so much.

This makes alot of sense considering the Linux Foundation is also in charge of Kompute which is likely to be the basis of vendor agnostic GPGPU, and thus the basis of vendor agnostic GPU-based machine learning.

So, its key features are:

1. domain-specific: designed for conveniently creating and composing derivations. This reason alone already justifies a new language, or an embedded domain-specific language (such as the Guile/Scheme for guix), or a mix of both (Starlark, the build language of Bazel embedded in a restricted Python-variant).

2. purely functional: this ties well into the philosophical backing of Nix the package manager, which aims to be purely functional, also known as hermeticity in other build systems (Bazel).

3. lazily evaluated: similar to other build systems (including Bazel), so that you can build only what you need on demand.

4. dynamically typed: this one is controversial. Being dynamically typed—in other words, not developing a type system—gets Nix out of the door first. But users often complain about the lack of proper types and modularity. There are experiments to address this, such as Nickel (https://github.com/tweag/nickel).

It is understandable that a one-pager may not have space for the whys.

Nickel：Nickel is a straightforward configuration language aimed at automatically generating static configuration files. Essentially, it's akin to JSON with the addition of functions and types.

Also look at nickel which is an evolution of nix. It's my favorite in this space.

nickel-lang.org

https://github.com/tweag/nickel

Or this https://github.com/tweag/nickel

Nickel is a programming language. While HCL is just a configuration format, so not really comparable.

Here's a comparison with similar tools: https://github.com/tweag/nickel#comparison

As for 'providence', I suppose you meant provenance :) it's been delayed because this was less critical for 1.0 to decide on or to implement (as it: it doesn't break backward compatibility in any way to add this feature in the short term), but this is very much on the roadmap: Issue #235. That's a must-have in a language with merging like Nickel.

Have you considered using a different language for templating? this could be a BIG selling point. Some good ones are cue-lang (though I haven't seen support for rust), kcl or nickel-lang.