llvm-project

6cbd91ea - [AArch64][GlobalISel] Combine G_UNMERGE(G_DUPLANE16) -> G_DUPLANE16 (#142731)

[AArch64][GlobalISel] Combine G_UNMERGE(G_DUPLANE16) -> G_DUPLANE16 (#142731) We will generate G_UNMERGE(G_DUPLANE16) due to the legalization of shuffle vector splats with mismatching vector sizes. The G_DUPLANE intrinsics can handle different vector sizes (128bit and 64bit output, for example), and we can combine away the unmerge.