[Mlir-commits] [llvm] [mlir] [OMPIRBuilder] Hoist alloca's to entry blocks of compiler-emitted GPU reduction functions (PR #181359)

[Jeff Sandoval]

jeffreysandoval wrote:

Thanks for the detailed explanation!  I agree that placing alloca's in the entry block whenever possible is best, but for cases where that isn't possible then it's helpful to know more of the context.

https://github.com/llvm/llvm-project/pull/181359