[llvm] r218182 - Using a deque to manage the stack of nodes is faster here.

[Lenny Maiorani]

Author: lenny
Date: Sat Sep 20 08:29:20 2014
New Revision: 218182

URL: http://llvm.org/viewvc/llvm-project?rev=218182&view=rev
Log:
Using a deque to manage the stack of nodes is faster here.

  Vector is slow due to many reallocations as the size regularly changes in
  unpredictable ways. See the investigation provided on the mailing list for
  more information:

http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20120116/135228.html

Modified:
    llvm/trunk/lib/Transforms/Scalar/EarlyCSE.cpp

Modified: llvm/trunk/lib/Transforms/Scalar/EarlyCSE.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Scalar/EarlyCSE.cpp?rev=218182&r1=218181&r2=218182&view=diff
==============================================================================
--- llvm/trunk/lib/Transforms/Scalar/EarlyCSE.cpp (original)
+++ llvm/trunk/lib/Transforms/Scalar/EarlyCSE.cpp Sat Sep 20 08:29:20 2014
@@ -26,7 +26,7 @@
 #include "llvm/Support/RecyclingAllocator.h"
 #include "llvm/Target/TargetLibraryInfo.h"
 #include "llvm/Transforms/Utils/Local.h"
-#include <vector>
+#include <deque>
 using namespace llvm;
 
 #define DEBUG_TYPE "early-cse"
@@ -560,7 +560,11 @@ bool EarlyCSE::runOnFunction(Function &F
   if (skipOptnoneFunction(F))
     return false;
 
-  std::vector<StackNode *> nodesToProcess;
+  // Note, deque is being used here because there is significant performance gains
+  // over vector when the container becomes very large due to the specific access
+  // patterns. For more information see the mailing list discussion on this:
+  // http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20120116/135228.html
+  std::deque<StackNode *> nodesToProcess;
 
   DataLayoutPass *DLP = getAnalysisIfAvailable<DataLayoutPass>();
   DL = DLP ? &DLP->getDataLayout() : nullptr;