My personal hell of translating DXIL to SPIR-V – part 3(themaister.net)
themaister.net
My personal hell of translating DXIL to SPIR-V – part 3
https://themaister.net/blog/2021/11/07/my-personal-hell-of-translating-dxil-to-spir-v-part-3/
7 comments
I cannot tell what the author intends. Is it hoped MS will make D3Dx code more portable to Vulkan? Or that Vulkan will be made more portable from D3Dx? Is it meant to help the next poor sod obliged to do the same chore? Or an extended, and ultimately futile, gripe?
The author is a contributor to vkd3d-proton. vkd3d is an implementation of Direct3D 12 on top of Vulkan, which is used in Wine and Proton to let you run D3D12 apps and games on Linux.
The intention of the article seems to be just to document his experience solving this problem, but it might be useful to anyone who finds themselves having to translate shaders from one language to another, which is disappointingly common in low-level cross-platform graphics programming, since every OS vendor likes to have their own shader language.
The intention of the article seems to be just to document his experience solving this problem, but it might be useful to anyone who finds themselves having to translate shaders from one language to another, which is disappointingly common in low-level cross-platform graphics programming, since every OS vendor likes to have their own shader language.
Thank you. Context isn't everything, but it is a lot.
I am astonished it is even possible for DXIL to be automatically translated to SPIR-V, at runtime, or at all.
I am astonished it is even possible for DXIL to be automatically translated to SPIR-V, at runtime, or at all.
From the first part:
> In this blog series I’d like to go through the DXIL format, explain the problems I’ve had to solve and hopefully serve as an introduction to basic compiler theory, explained by yours truly who has no idea what they’re talking about.
However, simply putting this out in public, especially if it hits HN, is useful since the number of people who can really understand and help is very tiny, they're clustered, and more likely to read HN over lots of other things.
> In this blog series I’d like to go through the DXIL format, explain the problems I’ve had to solve and hopefully serve as an introduction to basic compiler theory, explained by yours truly who has no idea what they’re talking about.
However, simply putting this out in public, especially if it hits HN, is useful since the number of people who can really understand and help is very tiny, they're clustered, and more likely to read HN over lots of other things.
This is cool and nerdy and all, but how this can possibly appeal broadly enough to hit HN front page is beyond my understanding - as is 99% of the article. Sorry, had to say it, feel free to downvote.
I'd rather have an in depth article about SPIR-V on the front page even if I'm not a gfx engineer than another generic, mainstream article that's unrelated to computer stuff. We have already too many of those and we're slowly turning into Ars Technica.
Are you implying you have a theory?