ash VS wgpu

If you want to automate the creation of certain code that will be nested under the src directory, and committed into source control, there are a variety of ways to do that. One option is to create a sub-crate in your project that does the code-gen and you can run it as needed. The ash project is an example of this.

There is also [Vulkano](https://github.com/vulkano-rs/vulkano). It has a safe high level api and lower level layers, all the way down to [ash](https://github.com/ash-rs/ash) which is more or less raw vulkan. It's more explicit and verbose than [wgpu](https://github.com/gfx-rs/wgpu) though, so maybe try wgpu first and see how you like it.

Yes. I use Ash bindings for Vulkan and raw GLSL for shaders. I tried to use spirv-std for shaders—it was really great to have the benefits of Rust (structs, enums, strict type assertions, etc.)—until I needed to use buffer references. Unfortunaly, it is not yet implemented.

This is great, of all the vulkan go bindings I have been waiting for one that uses vk.xml to generate the bindings, because it's the only viable way to keep the bindings up-to-date. (fyi the popular [ash-rs](https://github.com/ash-rs/ash) is created the same-way).

Hi, I am having an issue on pop-os wayland where when I run a vulkan application I get an error symbol lookup error: /lib/x86_64-linux-gnu/libnvidia-vulkan-producer.so: undefined symbol: wlEglCreateSurfaceExport I have tested this with https://github.com/ash-rs/ash running the triangle example.

Nice and simple. Its quite portable too. But simplicity and ease of use come with some limitations. Ash is much more complex but can extract every bit of power from your card if needed. Wgpu-rs github comes with many examples and you can find a really nice tutorial here

low level vulkan bindings

Glium and ash provide low level access to different common graphics api's. I'm sure there's a good directx-11/12 bindings as well but I'm unfamiliar with what people use.

fn read_uncompressed_buffer( reader: &mut R, length: usize, // estimated via a seek or other mechanism file_is_little_endian: bool, ) -> Result> { let bytes = length * std::mem::size_of::(); // it is undefined behavior to call read_exact on un-initialized, https://doc.rust-lang.org/std/io/trait.Read.html#tymethod.read // see also https://github.com/MaikKlein/ash/issues/354#issue-781730580 let mut buffer = vec![0u64, length]; unsafe { // transmute u64 to bytes. let slice = std::slice::from_raw_parts_mut( buffer.as_mut_ptr() as *mut u8, length * std::mem::size_of::(), ); reader.read_exact(slice)?; } if is_native_little_endian() != file_is_little_endian { swap(&mut buffer, file_is_little_endian) } }

// it is undefined behavior to call read_exact on un-initialized, https://doc.rust-lang.org/std/io/trait.Read.html#tymethod.read // see also https://github.com/MaikKlein/ash/issues/354#issue-781730580 let mut buffer = vec![0u64, length]; unsafe { // transmute u64 to bytes. let slice = std::slice::from_raw_parts_mut( buffer.as_mut_ptr() as *mut u8, length * std::mem::size_of::(), ); reader.read_exact(slice)?; } if is_native_little_endian() != file_is_little_endian { swap(&mut buffer, file_is_little_endian) }

People have already mentioned Metal, but if you want cross platform, https://github.com/gfx-rs/wgpu has a vulkan-like API and cross compiles to all the various GPU frameworks. I believe it uses https://github.com/KhronosGroup/MoltenVK to run on Macs. You can also see the metal shader transpilation results for debugging.

This is in reference to [1].

[1] https://github.com/gfx-rs/wgpu/issues/6434

> Did they actually write that somewhere?

wgpu.rs for comments about "general purpose", "modern" is my own interjection, and [1] about the difference in priorities between the WGPU maintainers and (funnily enough) you.

I do believe a cross-platform and modern interface is difficult to square with high performance. I don't think either side is wrong to choose the priorities it has, but someone will be underserved. They probably need far more manpower/talent than they have to improve performance proportionately.

[1] https://github.com/gfx-rs/wgpu/discussions/5525

Sorry, I'll try to be clearer. QRhi docs[1] say "The Qt Rendering Hardware Interface is an abstraction for hardware accelerated graphics APIs, such as, OpenGL, OpenGL ES, Direct3D, Metal, and Vulkan." And PySide6 includes a (python) wrapper for QRhi[2]. Meanwhile, pygfx builds on wgpu-py[3] which builds on wgpu[4] which is a "is a cross-platform, safe, pure-rust graphics API. It runs natively on Vulkan, Metal, D3D12, and OpenGL".

So, from the standpoint of someone using PySide6, QRhi and pygfx seem to be alternative paths to the exact same underlying graphics APIs.

Thus my question: How do they compare? How should I make an informed comparison between them?

[1] https://doc.qt.io/qt-6/qrhi.html

[2] https://doc.qt.io/qtforpython-6/PySide6/QtWidgets/QRhiWidget...

[3] https://github.com/pygfx/wgpu-py/

[4] https://github.com/gfx-rs/wgpu

Everything will be using compute shaders for the foreseeable future. [WGPU](https://wgpu.rs/) abstracts that to work with WebGPU on browsers, DirectX/Vulkan on Windows, Metal on Mac, and Vulkan on Linux and Android. There may be opportunities to explore vendor-specific options like CUDA in the far future to leverage a further increase in performance, but compute shaders are portable and nearly as good.

Instead of building their GPU support atop CUDA/NVIDIA [0], I’m wondering why they didn’t instead go with WebGPU [1] via something like wgpu [2]. Using wgpu, they could offer cross-platform compatibility across several graphics API’s, covering a wide range of hardware including NVIDIA GeForce and Quadro, AMD Radeon, Intel Iris and Arc, ARM Mali, and Apple’s integrated GPU’s.

They note the following [0]:

> The GPU module is under active development. APIs and semantics might change between Codon releases.

The thing is, based on the current syntax and semantics I see, it’ll almost certainly need to change to support non-NVIDIA devices, so I think it might be a better idea to just go with WebGPU compute pipelines sooner rather than later.

Just my two pennies…

[0]: https://docs.exaloop.io/codon/advanced/gpu

[1]: https://www.w3.org/TR/webgpu

[2]: https://wgpu.rs

wgpu supports WebGPU: https://github.com/gfx-rs/wgpu :

> While WebGPU does not support any shading language other than WGSL, we will automatically convert your non-WGSL shaders if you're running on WebGPU.

https://github.com/gfx-rs/wgpu?tab=readme-ov-file#tracking-t...