[AMDGPU] Fix DPP sequence in atomic optimizer.

This commit fixes the DPP sequence in the atomic optimizer (which was
previously missing the row_shr:3 step), and works around a read_register
exec bug by using a ballot instead.

Differential Revision: https://reviews.llvm.org/D57737

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@353703 91177308-0d34-0410-b5e6-96231b3b80d8
7 files changed