[Mlir-commits] [mlir] [mlir][gpu] Introduce the `gpu.conditional_execution` op (PR #78013)

[Guray Ozen]

grypp wrote:

Thank you for the clarification. I understand that if one first outlines the `gpu.launch`, the pass can determine whether it is the host or device based on the context. Nice approach!

I am wondering if you have plan to support cases like below? 
```
#if defined(__CUDA_ARCH__) && __CUDA_ARCH__ >= 900
// sm_90+
#elif defined(__CUDA_ARCH__) && __CUDA_ARCH__ >= 800
// sm_80
#else 
// host 
#endif 
```

https://github.com/llvm/llvm-project/pull/78013