[llvm-branch-commits] [llvm] [AMDGPU] DS loop wait relaxation -- more test cases and improvements … (PR #171952)

[Matt Arsenault via llvm-branch-commits]

================
@@ -2816,12 +2870,32 @@ void SIInsertWaitcnts::analyzeSingleBBLoopDSLoads(MachineLoop *ML) {
     }
   }
 
-  Info.TotalDSLoads = LoadPosition;
+  // Filter out flushed loads and renumber remaining ones
+  // Also compute the floor wait count - the wait established by same-iteration
+  // use
+  if (LastFlushedPosition > 0) {
+    DenseMap<unsigned, unsigned> NewMap;
----------------
arsenm wrote:

Number of register units is a fixed size 

https://github.com/llvm/llvm-project/pull/171952