Don't use a potentially expensive shift if all we want is one set bit.

[oota-llvm.git] / lib / Target / R600 / AMDGPUTargetMachine.cpp

diff --git a/lib/Target/R600/AMDGPUTargetMachine.cpp b/lib/Target/R600/AMDGPUTargetMachine.cpp
index 90f72de8ea7bb898b93e1fb168057a8d821aae16..7a14e50f86dc59e8de3721ca4cba4696238ea778 100644 (file)
--- a/lib/Target/R600/AMDGPUTargetMachine.cpp
+++ b/lib/Target/R600/AMDGPUTargetMachine.cpp
@@ -148,7 +148,11 @@ bool AMDGPUPassConfig::addPostRegAlloc() {
 }
 
 bool AMDGPUPassConfig::addPreSched2() {
+  const AMDGPUSubtarget &ST = TM->getSubtarget<AMDGPUSubtarget>();
 
+  if (ST.getGeneration() <= AMDGPUSubtarget::NORTHERN_ISLANDS) {
+    addPass(createR600EmitClauseMarkers(*TM));
+  }
   addPass(&IfConverterID);
   return false;
 }
@@ -158,7 +162,6 @@ bool AMDGPUPassConfig::addPreEmitPass() {
   if (ST.getGeneration() <= AMDGPUSubtarget::NORTHERN_ISLANDS) {
     addPass(createAMDGPUCFGPreparationPass(*TM));
     addPass(createAMDGPUCFGStructurizerPass(*TM));
-    addPass(createR600EmitClauseMarkers(*TM));
     addPass(createR600ExpandSpecialInstrsPass(*TM));
     addPass(&FinalizeMachineBundlesID);
     addPass(createR600Packetizer(*TM));