ash VS SPIRV-LLVM-Translator

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) }

Of course, CUDA is field tested framework to do job done. But it is interesting to look at other perspective technologies. Are there many differences between NVIDIA PTX and SPIR-V? Isn't possible to compile (Rust, C++, CUDA, etc.) to LLVM IR, then use tool like LLVM to SPIR-V translator and have GPU code that runs on any vendor card?