R600/SI: Fix assertion from copying a TargetGlobalAddress

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

diff --git a/lib/Target/R600/AMDGPUTargetTransformInfo.cpp b/lib/Target/R600/AMDGPUTargetTransformInfo.cpp
index e3a546e1d723e1f1efec9beea7b67d78f6227d63..ba4735147a1e13f6eba6cbbe8c4754b5e038ec13 100644 (file)
--- a/lib/Target/R600/AMDGPUTargetTransformInfo.cpp
+++ b/lib/Target/R600/AMDGPUTargetTransformInfo.cpp
@@ -52,7 +52,7 @@ public:
 
   AMDGPUTTI(const AMDGPUTargetMachine *TM)
       : ImmutablePass(ID), TM(TM), ST(TM->getSubtargetImpl()),
-        TLI(TM->getTargetLowering()) {
+        TLI(TM->getSubtargetImpl()->getTargetLowering()) {
     initializeAMDGPUTTIPass(*PassRegistry::getPassRegistry());
   }
 
@@ -79,6 +79,10 @@ public:
 
   PopcntSupportKind getPopcntSupport(unsigned IntTyWidthInBit) const override;
 
+  unsigned getNumberOfRegisters(bool Vector) const override;
+  unsigned getRegisterBitWidth(bool Vector) const override;
+  unsigned getMaximumUnrollFactor() const override;
+
   /// @}
 };
 
@@ -97,6 +101,12 @@ bool AMDGPUTTI::hasBranchDivergence() const { return true; }
 
 void AMDGPUTTI::getUnrollingPreferences(Loop *L,
                                         UnrollingPreferences &UP) const {
+  UP.Threshold = 300; // Twice the default.
+  UP.Count = UINT_MAX;
+  UP.Partial = true;
+
+  // TODO: Do we want runtime unrolling?
+
   for (const BasicBlock *BB : L->getBlocks()) {
     for (const Instruction &I : *BB) {
       const GetElementPtrInst *GEP = dyn_cast<GetElementPtrInst>(&I);
@@ -116,7 +126,7 @@ void AMDGPUTTI::getUnrollingPreferences(Loop *L,
         //
         // Don't use the maximum allowed value here as it will make some
         // programs way too big.
-        UP.Threshold = 500;
+        UP.Threshold = 800;
       }
     }
   }
@@ -127,3 +137,23 @@ AMDGPUTTI::getPopcntSupport(unsigned TyWidth) const {
   assert(isPowerOf2_32(TyWidth) && "Ty width must be power of 2");
   return ST->hasBCNT(TyWidth) ? PSK_FastHardware : PSK_Software;
 }
+
+unsigned AMDGPUTTI::getNumberOfRegisters(bool Vec) const {
+  if (Vec)
+    return 0;
+
+  // Number of VGPRs on SI.
+  if (ST->getGeneration() >= AMDGPUSubtarget::SOUTHERN_ISLANDS)
+    return 256;
+
+  return 4 * 128; // XXX - 4 channels. Should these count as vector instead?
+}
+
+unsigned AMDGPUTTI::getRegisterBitWidth(bool) const {
+  return 32;
+}
+
+unsigned AMDGPUTTI::getMaximumUnrollFactor() const {
+  // Semi-arbitrary large amount.
+  return 64;
+}