[PATCH] D54606: [AMDGPU] Convert insert_vector_elt into set of selects

[Matt Arsenault via Phabricator via llvm-commits]

arsenm added a comment.

Another change I've wanted to look at is changing what AMDGPUPromoteAlloca tries to produce. The dynamic vector indexing is going to be worse if the waterfall loop is going to be required, but it's currently preferred if both are possible.


https://reviews.llvm.org/D54606