Watch David Airlie Talk About Vulkan & RADV From LCA2017

[Hi-Angel]

The hypocrite. First he sides with ATI on destroying a proper open source display driver, then he acts like the big savior a few years down the road, when ATI is unpopular once again. Unbelievable. Powerplay all the way.

Isn't that great that he does something nice in the present.

[smitty3268]

Thank you for your answer! Still cant say I am convinced by this reasoning. Amd already have driver built around llvm ir. Same llvm backend or close one us used in rocm project I think. And it is natural to expect them to develop spirv llvm frontend as it is part of opengl 2.1.

Yeah, I think directly translating spir-v to llvm ir is probably the superior solution (only by a small amount though), but the key point is that the spir-v => NIR translation already exists and is working well. Perhaps AMD will eventually release something that goes straight to llvm ir, but then they're supposed to release a full vulkan driver too, and who knows when that will happen. It's a lot of work for a driver developed by 2 people in their spare time (and now 1 guy from Valve as well) when the NIR translation already works and isn't going to cost you very much.

[difron]

Yeah, I think directly translating spir-v to llvm ir is probably the superior solution (only by a small amount though), but the key point is that the spir-v => NIR translation already exists and is working well. Perhaps AMD will eventually release something that goes straight to llvm ir, but then they're supposed to release a full vulkan driver too, and who knows when that will happen. It's a lot of work for a driver developed by 2 people in their spare time (and now 1 guy from Valve as well) when the NIR translation already works and isn't going to cost you very much.

So you answer is that development of NIR->LLVM IR translator is simpler than SPIRV->LLVM IR translator? But both are translations between IR, why it would be simpler?