4 ms·

Lots of languages _already_ compile to SPIR-V - GLSL and HLSL for a start. Why add _another_ syntax?

shmerl 7y ago
I suppose current options aren't optimal. Or at least that could be a valid motivation to create one. If the motivation is to replace SPIR-V itself, then it's not valid.
- mwkaufma 7y ago
  Except it wasn't designed to be an "optimal target for SPIR-V" -- it was designed by Apple according to Metal's semantics, to promote Metal over Vulkan, and had to be rejiggered. It's worse, by your standard, than existing, battle-proven GLSL/HLSL toolchains.
  - kvark 7y ago
    This is not correct. The WGSL is renamed from Tint, which is a Google's proposal for a shading language, in agreement with all parties.
- kevingadd 7y ago
  They're on record that their motive is disputes with Khronos, it's in some past meeting notes.
  - shmerl 7y ago
    Then this should be rejected.
    - gfxgirl 7y ago
      reject it and there will be no WebGPU for 1.5 billion iOS devices just like there is no WebGL2 for 1.5 billion iOS devices.
      - shmerl 7y ago
        Why should Apple dictate their approach to everyone else then? If they are so resolved to remain outsiders technology wise because they are crazy bent on NIH, let them be outsiders. I'd say everyone else should take a hard stance on this and show Apple to the door, instead of dancing under their tune.
        xenadu02 7y ago
        It's called a "standards process" for a reason. Even if you personally have a different opinion the point of a standard is to get everyone implementing it on-board with that standard. Releasing a standard that a major member opposes and won't implement is a pointless exercise.
        shmerl 7y ago
        So if there won't be any consensus, then what? I see no reason to oblige Apple in this case then.
        vetinari 7y ago
        When Apple adopts some technology, they don't care whether others can do the same, or what problems that will cause to others. They might use the same medicine.
- imtringued 7y ago
  The motivation is to salvage SPIR-V in a way that makes Apple happy.
kvark 7y ago
Apple and Microsoft investigated the possibility of specifying WebHLSL tightly. The main concern was that it's a ton of work in terms of semantics. At the same time, SPIR-V semantics is fairly well specified, so we as a group figured that the familiar syntax of HLSL is not as important as the proper GPU behavior and our time specifying it. We've yet to see if it was the right decision - we might as well get stuck with discussing the syntax :)
imtringued 7y ago
From a technical standpoint a subset of SPIR-V that is safe to use in web browsers is exactly what is needed. However, there are IP problems that prevent direct usage of SPIR-V. [0] In short W3C needs full ownership of the spec so that it can be completely independent from Khronos. A textual representation of SPIR-V basically sidesteps these problems while still maintaining the ability to have a trivial SPIRV -> WSGL and WSGL -> SPIRV compiler. [0] https://news.ycombinator.com/item?id=22539633 https://news.ycombinator.com/item?id=22539633
- shmerl 7y ago
  They refused to explain what IP problems there are, and given it's Apple, I doubt they are valid.

I suppose current options aren't optimal. Or at least that could be a valid motivation to create one. If the motivation is to replace SPIR-V itself, then it's not valid.

mwkaufma 7y ago
Except it wasn't designed to be an "optimal target for SPIR-V" -- it was designed by Apple according to Metal's semantics, to promote Metal over Vulkan, and had to be rejiggered. It's worse, by your standard, than existing, battle-proven GLSL/HLSL toolchains.
- kvark 7y ago
  This is not correct. The WGSL is renamed from Tint, which is a Google's proposal for a shading language, in agreement with all parties.
kevingadd 7y ago
They're on record that their motive is disputes with Khronos, it's in some past meeting notes.
- shmerl 7y ago
  Then this should be rejected.
  - gfxgirl 7y ago
    reject it and there will be no WebGPU for 1.5 billion iOS devices just like there is no WebGL2 for 1.5 billion iOS devices.
    - shmerl 7y ago
      Why should Apple dictate their approach to everyone else then? If they are so resolved to remain outsiders technology wise because they are crazy bent on NIH, let them be outsiders. I'd say everyone else should take a hard stance on this and show Apple to the door, instead of dancing under their tune.
      - xenadu02 7y ago
        It's called a "standards process" for a reason. Even if you personally have a different opinion the point of a standard is to get everyone implementing it on-board with that standard. Releasing a standard that a major member opposes and won't implement is a pointless exercise.
        shmerl 7y ago
        So if there won't be any consensus, then what? I see no reason to oblige Apple in this case then.
        vetinari 7y ago
        When Apple adopts some technology, they don't care whether others can do the same, or what problems that will cause to others. They might use the same medicine.
imtringued 7y ago
The motivation is to salvage SPIR-V in a way that makes Apple happy.

kvark 7y ago

Apple and Microsoft investigated the possibility of specifying WebHLSL tightly. The main concern was that it's a ton of work in terms of semantics. At the same time, SPIR-V semantics is fairly well specified, so we as a group figured that the familiar syntax of HLSL is not as important as the proper GPU behavior and our time specifying it. We've yet to see if it was the right decision - we might as well get stuck with discussing the syntax :)

imtringued 7y ago

From a technical standpoint a subset of SPIR-V that is safe to use in web browsers is exactly what is needed. However, there are IP problems that prevent direct usage of SPIR-V. [0] In short W3C needs full ownership of the spec so that it can be completely independent from Khronos. A textual representation of SPIR-V basically sidesteps these problems while still maintaining the ability to have a trivial SPIRV -> WSGL and WSGL -> SPIRV compiler. [0] https://news.ycombinator.com/item?id=22539633 https://news.ycombinator.com/item?id=22539633

shmerl 7y ago
They refused to explain what IP problems there are, and given it's Apple, I doubt they are valid.