[llvm-branch-commits] [llvm-branch] r205808 - Merging r205738:

Tom Stellard thomas.stellard at amd.com
Tue Apr 8 14:47:17 PDT 2014


Author: tstellar
Date: Tue Apr  8 16:47:17 2014
New Revision: 205808

URL: http://llvm.org/viewvc/llvm-project?rev=205808&view=rev
Log:
Merging r205738:

------------------------------------------------------------------------
r205738 | atrick | 2014-04-07 17:29:22 -0400 (Mon, 07 Apr 2014) | 16 lines

Put a limit on ScheduleDAGSDNodes::ClusterNeighboringLoads to avoid blowing up compile time.

Fixes PR16365 - Extremely slow compilation in -O1 and -O2.

The SD scheduler has a quadratic implementation of load clustering
which absolutely blows up compile time for large blocks with constant
pool loads. The MI scheduler has a better implementation of load
clustering. However, we have not done the work yet to completely
eliminate the SD scheduler. Some benchmarks still seem to benefit from
early load clustering, although maybe by chance.

As an intermediate term fix, I just put a nice limit on the number of
DAG users to search before finding a match. With this limit there are no
binary differences in the LLVM test suite, and the PR16365 test case
does not suffer any compile time impact from this routine.

------------------------------------------------------------------------

Modified:
    llvm/branches/release_34/lib/CodeGen/SelectionDAG/ScheduleDAGSDNodes.cpp

Modified: llvm/branches/release_34/lib/CodeGen/SelectionDAG/ScheduleDAGSDNodes.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/branches/release_34/lib/CodeGen/SelectionDAG/ScheduleDAGSDNodes.cpp?rev=205808&r1=205807&r2=205808&view=diff
==============================================================================
--- llvm/branches/release_34/lib/CodeGen/SelectionDAG/ScheduleDAGSDNodes.cpp (original)
+++ llvm/branches/release_34/lib/CodeGen/SelectionDAG/ScheduleDAGSDNodes.cpp Tue Apr  8 16:47:17 2014
@@ -219,8 +219,11 @@ void ScheduleDAGSDNodes::ClusterNeighbor
   DenseMap<long long, SDNode*> O2SMap;  // Map from offset to SDNode.
   bool Cluster = false;
   SDNode *Base = Node;
+  // This algorithm requires a reasonably low use count before finding a match
+  // to avoid uselessly blowing up compile time in large blocks.
+  unsigned UseCount = 0;
   for (SDNode::use_iterator I = Chain->use_begin(), E = Chain->use_end();
-       I != E; ++I) {
+       I != E && UseCount < 100; ++I, ++UseCount) {
     SDNode *User = *I;
     if (User == Node || !Visited.insert(User))
       continue;
@@ -237,6 +240,8 @@ void ScheduleDAGSDNodes::ClusterNeighbor
     if (Offset2 < Offset1)
       Base = User;
     Cluster = true;
+    // Reset UseCount to allow more matches.
+    UseCount = 0;
   }
 
   if (!Cluster)





More information about the llvm-branch-commits mailing list