AMDGPU/SI: Fix Depth and Height computation for SI scheduler

[axeldavy]

Summary

Fix Depth and Height to use the block cost, instead of 1.
Fix the Height computation (max instead of min).

llvm/trunk/lib/Target/AMDGPU/SIMachineScheduler.cpp

void SIScheduleBlockCreator::fillStats() {
   for (unsigned i = 0, e = DAGSize; i != e; ++i) {
     int BlockIndice = TopDownIndex2Block[i];
     SIScheduleBlock *Block = CurrentBlocks[BlockIndice];
     if (Block->getPreds().empty())
       Block->Depth = 0;
     else {
       unsigned Depth = 0;
       for (SIScheduleBlock *Pred : Block->getPreds()) {
-        if (Depth < Pred->Depth + 1)
-          Depth = Pred->Depth + 1;
+        if (Depth < Pred->Depth + Pred->getCost())
+          Depth = Pred->Depth + Pred->getCost();
       }
       Block->Depth = Depth;
     }
   }
 
   for (unsigned i = 0, e = DAGSize; i != e; ++i) {
     int BlockIndice = BottomUpIndex2Block[i];
     SIScheduleBlock *Block = CurrentBlocks[BlockIndice];
     if (Block->getSuccs().empty())
       Block->Height = 0;
     else {
       unsigned Height = 0;
       for (const auto &Succ : Block->getSuccs())
-        Height = std::min(Height, Succ.first->Height + 1);
+        Height = std::max(Height, Succ.first->Height + Succ.first->getCost());
       Block->Height = Height;
     }
   }
 }
 
 // SIScheduleBlockScheduler //
 
 SIScheduleBlockScheduler::SIScheduleBlockScheduler(SIScheduleDAGMI *DAG,