Delete unused argument in AArch64MCInstLower constructor: it doesn't

[oota-llvm.git] / lib / Target / NVPTX / NVPTXISelDAGToDAG.cpp

diff --git a/lib/Target/NVPTX/NVPTXISelDAGToDAG.cpp b/lib/Target/NVPTX/NVPTXISelDAGToDAG.cpp
index 406daa009bde8eec168ff8d651be2dfd292790ba..cd0422d78a8c2d62d4684ad6809b5f4f9d5d55a0 100644 (file)
--- a/lib/Target/NVPTX/NVPTXISelDAGToDAG.cpp
+++ b/lib/Target/NVPTX/NVPTXISelDAGToDAG.cpp
@@ -11,7 +11,6 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "nvptx-isel"
 #include "NVPTXISelDAGToDAG.h"
 #include "llvm/IR/GlobalValue.h"
 #include "llvm/IR/Instructions.h"
@@ -23,16 +22,12 @@
 
 using namespace llvm;
 
-static cl::opt<int>
-FMAContractLevel("nvptx-fma-level", cl::ZeroOrMore, cl::Hidden,
-                 cl::desc("NVPTX Specific: FMA contraction (0: don't do it"
-                          " 1: do it  2: do it aggressively"),
-                 cl::init(2));
+#define DEBUG_TYPE "nvptx-isel"
 
 static cl::opt<int> UsePrecDivF32(
     "nvptx-prec-divf32", cl::ZeroOrMore, cl::Hidden,
     cl::desc("NVPTX Specifies: 0 use div.approx, 1 use div.full, 2 use"
-             " IEEE Compliant F32 div.rnd if avaiable."),
+             " IEEE Compliant F32 div.rnd if available."),
     cl::init(2));
 
 static cl::opt<bool>
@@ -57,16 +52,6 @@ NVPTXDAGToDAGISel::NVPTXDAGToDAGISel(NVPTXTargetMachine &tm,
                                      CodeGenOpt::Level OptLevel)
     : SelectionDAGISel(tm, OptLevel),
       Subtarget(tm.getSubtarget<NVPTXSubtarget>()) {
-
-  doFMAF32 = (OptLevel > 0) && Subtarget.hasFMAF32() && (FMAContractLevel >= 1);
-  doFMAF64 = (OptLevel > 0) && Subtarget.hasFMAF64() && (FMAContractLevel >= 1);
-  doFMAF32AGG =
-      (OptLevel > 0) && Subtarget.hasFMAF32() && (FMAContractLevel == 2);
-  doFMAF64AGG =
-      (OptLevel > 0) && Subtarget.hasFMAF64() && (FMAContractLevel == 2);
-
-  allowFMA = (FMAContractLevel >= 1);
-
   doMulWide = (OptLevel > 0);
 }
 
@@ -112,16 +97,21 @@ bool NVPTXDAGToDAGISel::useF32FTZ() const {
   }
 }
 
+bool NVPTXDAGToDAGISel::allowFMA() const {
+  const NVPTXTargetLowering *TL = Subtarget.getTargetLowering();
+  return TL->allowFMA(*MF, OptLevel);
+}
+
 /// Select - Select instructions not customized! Used for
 /// expanded, promoted and normal instructions.
 SDNode *NVPTXDAGToDAGISel::Select(SDNode *N) {
 
   if (N->isMachineOpcode()) {
     N->setNodeId(-1);
-    return NULL; // Already selected.
+    return nullptr; // Already selected.
   }
 
-  SDNode *ResNode = NULL;
+  SDNode *ResNode = nullptr;
   switch (N->getOpcode()) {
   case ISD::LOAD:
     ResNode = SelectLoad(N);
@@ -137,7 +127,7 @@ SDNode *NVPTXDAGToDAGISel::Select(SDNode *N) {
   case NVPTXISD::LDGV4:
   case NVPTXISD::LDUV2:
   case NVPTXISD::LDUV4:
-    ResNode = SelectLDGLDUVector(N);
+    ResNode = SelectLDGLDU(N);
     break;
   case NVPTXISD::StoreV2:
   case NVPTXISD::StoreV4:
@@ -163,95 +153,352 @@ SDNode *NVPTXDAGToDAGISel::Select(SDNode *N) {
   case ISD::INTRINSIC_WO_CHAIN:
     ResNode = SelectIntrinsicNoChain(N);
     break;
-  case NVPTXISD::Tex1DFloatI32:
+  case ISD::INTRINSIC_W_CHAIN:
+    ResNode = SelectIntrinsicChain(N);
+    break;
+  case NVPTXISD::Tex1DFloatS32:
   case NVPTXISD::Tex1DFloatFloat:
   case NVPTXISD::Tex1DFloatFloatLevel:
   case NVPTXISD::Tex1DFloatFloatGrad:
-  case NVPTXISD::Tex1DI32I32:
-  case NVPTXISD::Tex1DI32Float:
-  case NVPTXISD::Tex1DI32FloatLevel:
-  case NVPTXISD::Tex1DI32FloatGrad:
-  case NVPTXISD::Tex1DArrayFloatI32:
+  case NVPTXISD::Tex1DS32S32:
+  case NVPTXISD::Tex1DS32Float:
+  case NVPTXISD::Tex1DS32FloatLevel:
+  case NVPTXISD::Tex1DS32FloatGrad:
+  case NVPTXISD::Tex1DU32S32:
+  case NVPTXISD::Tex1DU32Float:
+  case NVPTXISD::Tex1DU32FloatLevel:
+  case NVPTXISD::Tex1DU32FloatGrad:
+  case NVPTXISD::Tex1DArrayFloatS32:
   case NVPTXISD::Tex1DArrayFloatFloat:
   case NVPTXISD::Tex1DArrayFloatFloatLevel:
   case NVPTXISD::Tex1DArrayFloatFloatGrad:
-  case NVPTXISD::Tex1DArrayI32I32:
-  case NVPTXISD::Tex1DArrayI32Float:
-  case NVPTXISD::Tex1DArrayI32FloatLevel:
-  case NVPTXISD::Tex1DArrayI32FloatGrad:
-  case NVPTXISD::Tex2DFloatI32:
+  case NVPTXISD::Tex1DArrayS32S32:
+  case NVPTXISD::Tex1DArrayS32Float:
+  case NVPTXISD::Tex1DArrayS32FloatLevel:
+  case NVPTXISD::Tex1DArrayS32FloatGrad:
+  case NVPTXISD::Tex1DArrayU32S32:
+  case NVPTXISD::Tex1DArrayU32Float:
+  case NVPTXISD::Tex1DArrayU32FloatLevel:
+  case NVPTXISD::Tex1DArrayU32FloatGrad:
+  case NVPTXISD::Tex2DFloatS32:
   case NVPTXISD::Tex2DFloatFloat:
   case NVPTXISD::Tex2DFloatFloatLevel:
   case NVPTXISD::Tex2DFloatFloatGrad:
-  case NVPTXISD::Tex2DI32I32:
-  case NVPTXISD::Tex2DI32Float:
-  case NVPTXISD::Tex2DI32FloatLevel:
-  case NVPTXISD::Tex2DI32FloatGrad:
-  case NVPTXISD::Tex2DArrayFloatI32:
+  case NVPTXISD::Tex2DS32S32:
+  case NVPTXISD::Tex2DS32Float:
+  case NVPTXISD::Tex2DS32FloatLevel:
+  case NVPTXISD::Tex2DS32FloatGrad:
+  case NVPTXISD::Tex2DU32S32:
+  case NVPTXISD::Tex2DU32Float:
+  case NVPTXISD::Tex2DU32FloatLevel:
+  case NVPTXISD::Tex2DU32FloatGrad:
+  case NVPTXISD::Tex2DArrayFloatS32:
   case NVPTXISD::Tex2DArrayFloatFloat:
   case NVPTXISD::Tex2DArrayFloatFloatLevel:
   case NVPTXISD::Tex2DArrayFloatFloatGrad:
-  case NVPTXISD::Tex2DArrayI32I32:
-  case NVPTXISD::Tex2DArrayI32Float:
-  case NVPTXISD::Tex2DArrayI32FloatLevel:
-  case NVPTXISD::Tex2DArrayI32FloatGrad:
-  case NVPTXISD::Tex3DFloatI32:
+  case NVPTXISD::Tex2DArrayS32S32:
+  case NVPTXISD::Tex2DArrayS32Float:
+  case NVPTXISD::Tex2DArrayS32FloatLevel:
+  case NVPTXISD::Tex2DArrayS32FloatGrad:
+  case NVPTXISD::Tex2DArrayU32S32:
+  case NVPTXISD::Tex2DArrayU32Float:
+  case NVPTXISD::Tex2DArrayU32FloatLevel:
+  case NVPTXISD::Tex2DArrayU32FloatGrad:
+  case NVPTXISD::Tex3DFloatS32:
   case NVPTXISD::Tex3DFloatFloat:
   case NVPTXISD::Tex3DFloatFloatLevel:
   case NVPTXISD::Tex3DFloatFloatGrad:
-  case NVPTXISD::Tex3DI32I32:
-  case NVPTXISD::Tex3DI32Float:
-  case NVPTXISD::Tex3DI32FloatLevel:
-  case NVPTXISD::Tex3DI32FloatGrad:
+  case NVPTXISD::Tex3DS32S32:
+  case NVPTXISD::Tex3DS32Float:
+  case NVPTXISD::Tex3DS32FloatLevel:
+  case NVPTXISD::Tex3DS32FloatGrad:
+  case NVPTXISD::Tex3DU32S32:
+  case NVPTXISD::Tex3DU32Float:
+  case NVPTXISD::Tex3DU32FloatLevel:
+  case NVPTXISD::Tex3DU32FloatGrad:
+  case NVPTXISD::TexCubeFloatFloat:
+  case NVPTXISD::TexCubeFloatFloatLevel:
+  case NVPTXISD::TexCubeS32Float:
+  case NVPTXISD::TexCubeS32FloatLevel:
+  case NVPTXISD::TexCubeU32Float:
+  case NVPTXISD::TexCubeU32FloatLevel:
+  case NVPTXISD::TexCubeArrayFloatFloat:
+  case NVPTXISD::TexCubeArrayFloatFloatLevel:
+  case NVPTXISD::TexCubeArrayS32Float:
+  case NVPTXISD::TexCubeArrayS32FloatLevel:
+  case NVPTXISD::TexCubeArrayU32Float:
+  case NVPTXISD::TexCubeArrayU32FloatLevel:
+  case NVPTXISD::Tld4R2DFloatFloat:
+  case NVPTXISD::Tld4G2DFloatFloat:
+  case NVPTXISD::Tld4B2DFloatFloat:
+  case NVPTXISD::Tld4A2DFloatFloat:
+  case NVPTXISD::Tld4R2DS64Float:
+  case NVPTXISD::Tld4G2DS64Float:
+  case NVPTXISD::Tld4B2DS64Float:
+  case NVPTXISD::Tld4A2DS64Float:
+  case NVPTXISD::Tld4R2DU64Float:
+  case NVPTXISD::Tld4G2DU64Float:
+  case NVPTXISD::Tld4B2DU64Float:
+  case NVPTXISD::Tld4A2DU64Float:
+  case NVPTXISD::TexUnified1DFloatS32:
+  case NVPTXISD::TexUnified1DFloatFloat:
+  case NVPTXISD::TexUnified1DFloatFloatLevel:
+  case NVPTXISD::TexUnified1DFloatFloatGrad:
+  case NVPTXISD::TexUnified1DS32S32:
+  case NVPTXISD::TexUnified1DS32Float:
+  case NVPTXISD::TexUnified1DS32FloatLevel:
+  case NVPTXISD::TexUnified1DS32FloatGrad:
+  case NVPTXISD::TexUnified1DU32S32:
+  case NVPTXISD::TexUnified1DU32Float:
+  case NVPTXISD::TexUnified1DU32FloatLevel:
+  case NVPTXISD::TexUnified1DU32FloatGrad:
+  case NVPTXISD::TexUnified1DArrayFloatS32:
+  case NVPTXISD::TexUnified1DArrayFloatFloat:
+  case NVPTXISD::TexUnified1DArrayFloatFloatLevel:
+  case NVPTXISD::TexUnified1DArrayFloatFloatGrad:
+  case NVPTXISD::TexUnified1DArrayS32S32:
+  case NVPTXISD::TexUnified1DArrayS32Float:
+  case NVPTXISD::TexUnified1DArrayS32FloatLevel:
+  case NVPTXISD::TexUnified1DArrayS32FloatGrad:
+  case NVPTXISD::TexUnified1DArrayU32S32:
+  case NVPTXISD::TexUnified1DArrayU32Float:
+  case NVPTXISD::TexUnified1DArrayU32FloatLevel:
+  case NVPTXISD::TexUnified1DArrayU32FloatGrad:
+  case NVPTXISD::TexUnified2DFloatS32:
+  case NVPTXISD::TexUnified2DFloatFloat:
+  case NVPTXISD::TexUnified2DFloatFloatLevel:
+  case NVPTXISD::TexUnified2DFloatFloatGrad:
+  case NVPTXISD::TexUnified2DS32S32:
+  case NVPTXISD::TexUnified2DS32Float:
+  case NVPTXISD::TexUnified2DS32FloatLevel:
+  case NVPTXISD::TexUnified2DS32FloatGrad:
+  case NVPTXISD::TexUnified2DU32S32:
+  case NVPTXISD::TexUnified2DU32Float:
+  case NVPTXISD::TexUnified2DU32FloatLevel:
+  case NVPTXISD::TexUnified2DU32FloatGrad:
+  case NVPTXISD::TexUnified2DArrayFloatS32:
+  case NVPTXISD::TexUnified2DArrayFloatFloat:
+  case NVPTXISD::TexUnified2DArrayFloatFloatLevel:
+  case NVPTXISD::TexUnified2DArrayFloatFloatGrad:
+  case NVPTXISD::TexUnified2DArrayS32S32:
+  case NVPTXISD::TexUnified2DArrayS32Float:
+  case NVPTXISD::TexUnified2DArrayS32FloatLevel:
+  case NVPTXISD::TexUnified2DArrayS32FloatGrad:
+  case NVPTXISD::TexUnified2DArrayU32S32:
+  case NVPTXISD::TexUnified2DArrayU32Float:
+  case NVPTXISD::TexUnified2DArrayU32FloatLevel:
+  case NVPTXISD::TexUnified2DArrayU32FloatGrad:
+  case NVPTXISD::TexUnified3DFloatS32:
+  case NVPTXISD::TexUnified3DFloatFloat:
+  case NVPTXISD::TexUnified3DFloatFloatLevel:
+  case NVPTXISD::TexUnified3DFloatFloatGrad:
+  case NVPTXISD::TexUnified3DS32S32:
+  case NVPTXISD::TexUnified3DS32Float:
+  case NVPTXISD::TexUnified3DS32FloatLevel:
+  case NVPTXISD::TexUnified3DS32FloatGrad:
+  case NVPTXISD::TexUnified3DU32S32:
+  case NVPTXISD::TexUnified3DU32Float:
+  case NVPTXISD::TexUnified3DU32FloatLevel:
+  case NVPTXISD::TexUnified3DU32FloatGrad:
+  case NVPTXISD::TexUnifiedCubeFloatFloat:
+  case NVPTXISD::TexUnifiedCubeFloatFloatLevel:
+  case NVPTXISD::TexUnifiedCubeS32Float:
+  case NVPTXISD::TexUnifiedCubeS32FloatLevel:
+  case NVPTXISD::TexUnifiedCubeU32Float:
+  case NVPTXISD::TexUnifiedCubeU32FloatLevel:
+  case NVPTXISD::TexUnifiedCubeArrayFloatFloat:
+  case NVPTXISD::TexUnifiedCubeArrayFloatFloatLevel:
+  case NVPTXISD::TexUnifiedCubeArrayS32Float:
+  case NVPTXISD::TexUnifiedCubeArrayS32FloatLevel:
+  case NVPTXISD::TexUnifiedCubeArrayU32Float:
+  case NVPTXISD::TexUnifiedCubeArrayU32FloatLevel:
+  case NVPTXISD::Tld4UnifiedR2DFloatFloat:
+  case NVPTXISD::Tld4UnifiedG2DFloatFloat:
+  case NVPTXISD::Tld4UnifiedB2DFloatFloat:
+  case NVPTXISD::Tld4UnifiedA2DFloatFloat:
+  case NVPTXISD::Tld4UnifiedR2DS64Float:
+  case NVPTXISD::Tld4UnifiedG2DS64Float:
+  case NVPTXISD::Tld4UnifiedB2DS64Float:
+  case NVPTXISD::Tld4UnifiedA2DS64Float:
+  case NVPTXISD::Tld4UnifiedR2DU64Float:
+  case NVPTXISD::Tld4UnifiedG2DU64Float:
+  case NVPTXISD::Tld4UnifiedB2DU64Float:
+  case NVPTXISD::Tld4UnifiedA2DU64Float:
     ResNode = SelectTextureIntrinsic(N);
     break;
+  case NVPTXISD::Suld1DI8Clamp:
+  case NVPTXISD::Suld1DI16Clamp:
+  case NVPTXISD::Suld1DI32Clamp:
+  case NVPTXISD::Suld1DI64Clamp:
+  case NVPTXISD::Suld1DV2I8Clamp:
+  case NVPTXISD::Suld1DV2I16Clamp:
+  case NVPTXISD::Suld1DV2I32Clamp:
+  case NVPTXISD::Suld1DV2I64Clamp:
+  case NVPTXISD::Suld1DV4I8Clamp:
+  case NVPTXISD::Suld1DV4I16Clamp:
+  case NVPTXISD::Suld1DV4I32Clamp:
+  case NVPTXISD::Suld1DArrayI8Clamp:
+  case NVPTXISD::Suld1DArrayI16Clamp:
+  case NVPTXISD::Suld1DArrayI32Clamp:
+  case NVPTXISD::Suld1DArrayI64Clamp:
+  case NVPTXISD::Suld1DArrayV2I8Clamp:
+  case NVPTXISD::Suld1DArrayV2I16Clamp:
+  case NVPTXISD::Suld1DArrayV2I32Clamp:
+  case NVPTXISD::Suld1DArrayV2I64Clamp:
+  case NVPTXISD::Suld1DArrayV4I8Clamp:
+  case NVPTXISD::Suld1DArrayV4I16Clamp:
+  case NVPTXISD::Suld1DArrayV4I32Clamp:
+  case NVPTXISD::Suld2DI8Clamp:
+  case NVPTXISD::Suld2DI16Clamp:
+  case NVPTXISD::Suld2DI32Clamp:
+  case NVPTXISD::Suld2DI64Clamp:
+  case NVPTXISD::Suld2DV2I8Clamp:
+  case NVPTXISD::Suld2DV2I16Clamp:
+  case NVPTXISD::Suld2DV2I32Clamp:
+  case NVPTXISD::Suld2DV2I64Clamp:
+  case NVPTXISD::Suld2DV4I8Clamp:
+  case NVPTXISD::Suld2DV4I16Clamp:
+  case NVPTXISD::Suld2DV4I32Clamp:
+  case NVPTXISD::Suld2DArrayI8Clamp:
+  case NVPTXISD::Suld2DArrayI16Clamp:
+  case NVPTXISD::Suld2DArrayI32Clamp:
+  case NVPTXISD::Suld2DArrayI64Clamp:
+  case NVPTXISD::Suld2DArrayV2I8Clamp:
+  case NVPTXISD::Suld2DArrayV2I16Clamp:
+  case NVPTXISD::Suld2DArrayV2I32Clamp:
+  case NVPTXISD::Suld2DArrayV2I64Clamp:
+  case NVPTXISD::Suld2DArrayV4I8Clamp:
+  case NVPTXISD::Suld2DArrayV4I16Clamp:
+  case NVPTXISD::Suld2DArrayV4I32Clamp:
+  case NVPTXISD::Suld3DI8Clamp:
+  case NVPTXISD::Suld3DI16Clamp:
+  case NVPTXISD::Suld3DI32Clamp:
+  case NVPTXISD::Suld3DI64Clamp:
+  case NVPTXISD::Suld3DV2I8Clamp:
+  case NVPTXISD::Suld3DV2I16Clamp:
+  case NVPTXISD::Suld3DV2I32Clamp:
+  case NVPTXISD::Suld3DV2I64Clamp:
+  case NVPTXISD::Suld3DV4I8Clamp:
+  case NVPTXISD::Suld3DV4I16Clamp:
+  case NVPTXISD::Suld3DV4I32Clamp:
   case NVPTXISD::Suld1DI8Trap:
   case NVPTXISD::Suld1DI16Trap:
   case NVPTXISD::Suld1DI32Trap:
+  case NVPTXISD::Suld1DI64Trap:
   case NVPTXISD::Suld1DV2I8Trap:
   case NVPTXISD::Suld1DV2I16Trap:
   case NVPTXISD::Suld1DV2I32Trap:
+  case NVPTXISD::Suld1DV2I64Trap:
   case NVPTXISD::Suld1DV4I8Trap:
   case NVPTXISD::Suld1DV4I16Trap:
   case NVPTXISD::Suld1DV4I32Trap:
   case NVPTXISD::Suld1DArrayI8Trap:
   case NVPTXISD::Suld1DArrayI16Trap:
   case NVPTXISD::Suld1DArrayI32Trap:
+  case NVPTXISD::Suld1DArrayI64Trap:
   case NVPTXISD::Suld1DArrayV2I8Trap:
   case NVPTXISD::Suld1DArrayV2I16Trap:
   case NVPTXISD::Suld1DArrayV2I32Trap:
+  case NVPTXISD::Suld1DArrayV2I64Trap:
   case NVPTXISD::Suld1DArrayV4I8Trap:
   case NVPTXISD::Suld1DArrayV4I16Trap:
   case NVPTXISD::Suld1DArrayV4I32Trap:
   case NVPTXISD::Suld2DI8Trap:
   case NVPTXISD::Suld2DI16Trap:
   case NVPTXISD::Suld2DI32Trap:
+  case NVPTXISD::Suld2DI64Trap:
   case NVPTXISD::Suld2DV2I8Trap:
   case NVPTXISD::Suld2DV2I16Trap:
   case NVPTXISD::Suld2DV2I32Trap:
+  case NVPTXISD::Suld2DV2I64Trap:
   case NVPTXISD::Suld2DV4I8Trap:
   case NVPTXISD::Suld2DV4I16Trap:
   case NVPTXISD::Suld2DV4I32Trap:
   case NVPTXISD::Suld2DArrayI8Trap:
   case NVPTXISD::Suld2DArrayI16Trap:
   case NVPTXISD::Suld2DArrayI32Trap:
+  case NVPTXISD::Suld2DArrayI64Trap:
   case NVPTXISD::Suld2DArrayV2I8Trap:
   case NVPTXISD::Suld2DArrayV2I16Trap:
   case NVPTXISD::Suld2DArrayV2I32Trap:
+  case NVPTXISD::Suld2DArrayV2I64Trap:
   case NVPTXISD::Suld2DArrayV4I8Trap:
   case NVPTXISD::Suld2DArrayV4I16Trap:
   case NVPTXISD::Suld2DArrayV4I32Trap:
   case NVPTXISD::Suld3DI8Trap:
   case NVPTXISD::Suld3DI16Trap:
   case NVPTXISD::Suld3DI32Trap:
+  case NVPTXISD::Suld3DI64Trap:
   case NVPTXISD::Suld3DV2I8Trap:
   case NVPTXISD::Suld3DV2I16Trap:
   case NVPTXISD::Suld3DV2I32Trap:
+  case NVPTXISD::Suld3DV2I64Trap:
   case NVPTXISD::Suld3DV4I8Trap:
   case NVPTXISD::Suld3DV4I16Trap:
   case NVPTXISD::Suld3DV4I32Trap:
+  case NVPTXISD::Suld1DI8Zero:
+  case NVPTXISD::Suld1DI16Zero:
+  case NVPTXISD::Suld1DI32Zero:
+  case NVPTXISD::Suld1DI64Zero:
+  case NVPTXISD::Suld1DV2I8Zero:
+  case NVPTXISD::Suld1DV2I16Zero:
+  case NVPTXISD::Suld1DV2I32Zero:
+  case NVPTXISD::Suld1DV2I64Zero:
+  case NVPTXISD::Suld1DV4I8Zero:
+  case NVPTXISD::Suld1DV4I16Zero:
+  case NVPTXISD::Suld1DV4I32Zero:
+  case NVPTXISD::Suld1DArrayI8Zero:
+  case NVPTXISD::Suld1DArrayI16Zero:
+  case NVPTXISD::Suld1DArrayI32Zero:
+  case NVPTXISD::Suld1DArrayI64Zero:
+  case NVPTXISD::Suld1DArrayV2I8Zero:
+  case NVPTXISD::Suld1DArrayV2I16Zero:
+  case NVPTXISD::Suld1DArrayV2I32Zero:
+  case NVPTXISD::Suld1DArrayV2I64Zero:
+  case NVPTXISD::Suld1DArrayV4I8Zero:
+  case NVPTXISD::Suld1DArrayV4I16Zero:
+  case NVPTXISD::Suld1DArrayV4I32Zero:
+  case NVPTXISD::Suld2DI8Zero:
+  case NVPTXISD::Suld2DI16Zero:
+  case NVPTXISD::Suld2DI32Zero:
+  case NVPTXISD::Suld2DI64Zero:
+  case NVPTXISD::Suld2DV2I8Zero:
+  case NVPTXISD::Suld2DV2I16Zero:
+  case NVPTXISD::Suld2DV2I32Zero:
+  case NVPTXISD::Suld2DV2I64Zero:
+  case NVPTXISD::Suld2DV4I8Zero:
+  case NVPTXISD::Suld2DV4I16Zero:
+  case NVPTXISD::Suld2DV4I32Zero:
+  case NVPTXISD::Suld2DArrayI8Zero:
+  case NVPTXISD::Suld2DArrayI16Zero:
+  case NVPTXISD::Suld2DArrayI32Zero:
+  case NVPTXISD::Suld2DArrayI64Zero:
+  case NVPTXISD::Suld2DArrayV2I8Zero:
+  case NVPTXISD::Suld2DArrayV2I16Zero:
+  case NVPTXISD::Suld2DArrayV2I32Zero:
+  case NVPTXISD::Suld2DArrayV2I64Zero:
+  case NVPTXISD::Suld2DArrayV4I8Zero:
+  case NVPTXISD::Suld2DArrayV4I16Zero:
+  case NVPTXISD::Suld2DArrayV4I32Zero:
+  case NVPTXISD::Suld3DI8Zero:
+  case NVPTXISD::Suld3DI16Zero:
+  case NVPTXISD::Suld3DI32Zero:
+  case NVPTXISD::Suld3DI64Zero:
+  case NVPTXISD::Suld3DV2I8Zero:
+  case NVPTXISD::Suld3DV2I16Zero:
+  case NVPTXISD::Suld3DV2I32Zero:
+  case NVPTXISD::Suld3DV2I64Zero:
+  case NVPTXISD::Suld3DV4I8Zero:
+  case NVPTXISD::Suld3DV4I16Zero:
+  case NVPTXISD::Suld3DV4I32Zero:
     ResNode = SelectSurfaceIntrinsic(N);
     break;
+  case ISD::AND:
+  case ISD::SRA:
+  case ISD::SRL:
+    // Try to select BFE
+    ResNode = SelectBFE(N);
+    break;
   case ISD::ADDRSPACECAST:
     ResNode = SelectAddrSpaceCast(N);
     break;
@@ -263,6 +510,21 @@ SDNode *NVPTXDAGToDAGISel::Select(SDNode *N) {
   return SelectCode(N);
 }
 
+SDNode *NVPTXDAGToDAGISel::SelectIntrinsicChain(SDNode *N) {
+  unsigned IID = cast<ConstantSDNode>(N->getOperand(1))->getZExtValue();
+  switch (IID) {
+  default:
+    return NULL;
+  case Intrinsic::nvvm_ldg_global_f:
+  case Intrinsic::nvvm_ldg_global_i:
+  case Intrinsic::nvvm_ldg_global_p:
+  case Intrinsic::nvvm_ldu_global_f:
+  case Intrinsic::nvvm_ldu_global_i:
+  case Intrinsic::nvvm_ldu_global_p:
+    return SelectLDGLDU(N);
+  }
+}
+
 static unsigned int getCodeAddrSpace(MemSDNode *N,
                                      const NVPTXSubtarget &Subtarget) {
   const Value *Src = N->getMemOperand()->getValue();
@@ -288,7 +550,7 @@ SDNode *NVPTXDAGToDAGISel::SelectIntrinsicNoChain(SDNode *N) {
   unsigned IID = cast<ConstantSDNode>(N->getOperand(0))->getZExtValue();
   switch (IID) {
   default:
-    return NULL;
+    return nullptr;
   case Intrinsic::nvvm_texsurf_handle_internal:
     return SelectTexSurfHandle(N);
   }
@@ -366,14 +628,14 @@ SDNode *NVPTXDAGToDAGISel::SelectLoad(SDNode *N) {
   SDLoc dl(N);
   LoadSDNode *LD = cast<LoadSDNode>(N);
   EVT LoadedVT = LD->getMemoryVT();
-  SDNode *NVPTXLD = NULL;
+  SDNode *NVPTXLD = nullptr;
 
   // do not support pre/post inc/dec
   if (LD->isIndexed())
-    return NULL;
+    return nullptr;
 
   if (!LoadedVT.isSimple())
-    return NULL;
+    return nullptr;
 
   // Address Space Setting
   unsigned int codeAddrSpace = getCodeAddrSpace(LD, Subtarget);
@@ -396,7 +658,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLoad(SDNode *N) {
     else if (num == 4)
       vecType = NVPTX::PTXLdStInstCode::V4;
     else
-      return NULL;
+      return nullptr;
   }
 
   // Type Setting: fromType + fromTypeWidth
@@ -445,7 +707,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLoad(SDNode *N) {
       Opcode = NVPTX::LD_f64_avar;
       break;
     default:
-      return NULL;
+      return nullptr;
     }
     SDValue Ops[] = { getI32Imm(isVolatile), getI32Imm(codeAddrSpace),
                       getI32Imm(vecType), getI32Imm(fromType),
@@ -474,7 +736,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLoad(SDNode *N) {
       Opcode = NVPTX::LD_f64_asi;
       break;
     default:
-      return NULL;
+      return nullptr;
     }
     SDValue Ops[] = { getI32Imm(isVolatile), getI32Imm(codeAddrSpace),
                       getI32Imm(vecType), getI32Imm(fromType),
@@ -504,7 +766,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLoad(SDNode *N) {
         Opcode = NVPTX::LD_f64_ari_64;
         break;
       default:
-        return NULL;
+        return nullptr;
       }
     } else {
       switch (TargetVT) {
@@ -527,7 +789,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLoad(SDNode *N) {
         Opcode = NVPTX::LD_f64_ari;
         break;
       default:
-        return NULL;
+        return nullptr;
       }
     }
     SDValue Ops[] = { getI32Imm(isVolatile), getI32Imm(codeAddrSpace),
@@ -556,7 +818,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLoad(SDNode *N) {
         Opcode = NVPTX::LD_f64_areg_64;
         break;
       default:
-        return NULL;
+        return nullptr;
       }
     } else {
       switch (TargetVT) {
@@ -579,7 +841,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLoad(SDNode *N) {
         Opcode = NVPTX::LD_f64_areg;
         break;
       default:
-        return NULL;
+        return nullptr;
       }
     }
     SDValue Ops[] = { getI32Imm(isVolatile), getI32Imm(codeAddrSpace),
@@ -588,7 +850,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLoad(SDNode *N) {
     NVPTXLD = CurDAG->getMachineNode(Opcode, dl, TargetVT, MVT::Other, Ops);
   }
 
-  if (NVPTXLD != NULL) {
+  if (NVPTXLD) {
     MachineSDNode::mmo_iterator MemRefs0 = MF->allocateMemRefsArray(1);
     MemRefs0[0] = cast<MemSDNode>(N)->getMemOperand();
     cast<MachineSDNode>(NVPTXLD)->setMemRefs(MemRefs0, MemRefs0 + 1);
@@ -609,7 +871,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLoadVector(SDNode *N) {
   EVT LoadedVT = MemSD->getMemoryVT();
 
   if (!LoadedVT.isSimple())
-    return NULL;
+    return nullptr;
 
   // Address Space Setting
   unsigned int CodeAddrSpace = getCodeAddrSpace(MemSD, Subtarget);
@@ -655,7 +917,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLoadVector(SDNode *N) {
     VecType = NVPTX::PTXLdStInstCode::V4;
     break;
   default:
-    return NULL;
+    return nullptr;
   }
 
   EVT EltVT = N->getValueType(0);
@@ -663,11 +925,11 @@ SDNode *NVPTXDAGToDAGISel::SelectLoadVector(SDNode *N) {
   if (SelectDirectAddr(Op1, Addr)) {
     switch (N->getOpcode()) {
     default:
-      return NULL;
+      return nullptr;
     case NVPTXISD::LoadV2:
       switch (EltVT.getSimpleVT().SimpleTy) {
       default:
-        return NULL;
+        return nullptr;
       case MVT::i8:
         Opcode = NVPTX::LDV_i8_v2_avar;
         break;
@@ -691,7 +953,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLoadVector(SDNode *N) {
     case NVPTXISD::LoadV4:
       switch (EltVT.getSimpleVT().SimpleTy) {
       default:
-        return NULL;
+        return nullptr;
       case MVT::i8:
         Opcode = NVPTX::LDV_i8_v4_avar;
         break;
@@ -717,11 +979,11 @@ SDNode *NVPTXDAGToDAGISel::SelectLoadVector(SDNode *N) {
                  : SelectADDRsi(Op1.getNode(), Op1, Base, Offset)) {
     switch (N->getOpcode()) {
     default:
-      return NULL;
+      return nullptr;
     case NVPTXISD::LoadV2:
       switch (EltVT.getSimpleVT().SimpleTy) {
       default:
-        return NULL;
+        return nullptr;
       case MVT::i8:
         Opcode = NVPTX::LDV_i8_v2_asi;
         break;
@@ -745,7 +1007,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLoadVector(SDNode *N) {
     case NVPTXISD::LoadV4:
       switch (EltVT.getSimpleVT().SimpleTy) {
       default:
-        return NULL;
+        return nullptr;
       case MVT::i8:
         Opcode = NVPTX::LDV_i8_v4_asi;
         break;
@@ -772,11 +1034,11 @@ SDNode *NVPTXDAGToDAGISel::SelectLoadVector(SDNode *N) {
     if (Subtarget.is64Bit()) {
       switch (N->getOpcode()) {
       default:
-        return NULL;
+        return nullptr;
       case NVPTXISD::LoadV2:
         switch (EltVT.getSimpleVT().SimpleTy) {
         default:
-          return NULL;
+          return nullptr;
         case MVT::i8:
           Opcode = NVPTX::LDV_i8_v2_ari_64;
           break;
@@ -800,7 +1062,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLoadVector(SDNode *N) {
       case NVPTXISD::LoadV4:
         switch (EltVT.getSimpleVT().SimpleTy) {
         default:
-          return NULL;
+          return nullptr;
         case MVT::i8:
           Opcode = NVPTX::LDV_i8_v4_ari_64;
           break;
@@ -819,11 +1081,11 @@ SDNode *NVPTXDAGToDAGISel::SelectLoadVector(SDNode *N) {
     } else {
       switch (N->getOpcode()) {
       default:
-        return NULL;
+        return nullptr;
       case NVPTXISD::LoadV2:
         switch (EltVT.getSimpleVT().SimpleTy) {
         default:
-          return NULL;
+          return nullptr;
         case MVT::i8:
           Opcode = NVPTX::LDV_i8_v2_ari;
           break;
@@ -847,7 +1109,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLoadVector(SDNode *N) {
       case NVPTXISD::LoadV4:
         switch (EltVT.getSimpleVT().SimpleTy) {
         default:
-          return NULL;
+          return nullptr;
         case MVT::i8:
           Opcode = NVPTX::LDV_i8_v4_ari;
           break;
@@ -874,11 +1136,11 @@ SDNode *NVPTXDAGToDAGISel::SelectLoadVector(SDNode *N) {
     if (Subtarget.is64Bit()) {
       switch (N->getOpcode()) {
       default:
-        return NULL;
+        return nullptr;
       case NVPTXISD::LoadV2:
         switch (EltVT.getSimpleVT().SimpleTy) {
         default:
-          return NULL;
+          return nullptr;
         case MVT::i8:
           Opcode = NVPTX::LDV_i8_v2_areg_64;
           break;
@@ -902,7 +1164,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLoadVector(SDNode *N) {
       case NVPTXISD::LoadV4:
         switch (EltVT.getSimpleVT().SimpleTy) {
         default:
-          return NULL;
+          return nullptr;
         case MVT::i8:
           Opcode = NVPTX::LDV_i8_v4_areg_64;
           break;
@@ -921,11 +1183,11 @@ SDNode *NVPTXDAGToDAGISel::SelectLoadVector(SDNode *N) {
     } else {
       switch (N->getOpcode()) {
       default:
-        return NULL;
+        return nullptr;
       case NVPTXISD::LoadV2:
         switch (EltVT.getSimpleVT().SimpleTy) {
         default:
-          return NULL;
+          return nullptr;
         case MVT::i8:
           Opcode = NVPTX::LDV_i8_v2_areg;
           break;
@@ -949,7 +1211,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLoadVector(SDNode *N) {
       case NVPTXISD::LoadV4:
         switch (EltVT.getSimpleVT().SimpleTy) {
         default:
-          return NULL;
+          return nullptr;
         case MVT::i8:
           Opcode = NVPTX::LDV_i8_v4_areg;
           break;
@@ -980,26 +1242,105 @@ SDNode *NVPTXDAGToDAGISel::SelectLoadVector(SDNode *N) {
   return LD;
 }
 
-SDNode *NVPTXDAGToDAGISel::SelectLDGLDUVector(SDNode *N) {
+SDNode *NVPTXDAGToDAGISel::SelectLDGLDU(SDNode *N) {
 
   SDValue Chain = N->getOperand(0);
-  SDValue Op1 = N->getOperand(1);
+  SDValue Op1;
+  MemSDNode *Mem;
+  bool IsLDG = true;
+
+  // If this is an LDG intrinsic, the address is the third operand. Its its an
+  // LDG/LDU SD node (from custom vector handling), then its the second operand
+  if (N->getOpcode() == ISD::INTRINSIC_W_CHAIN) {
+    Op1 = N->getOperand(2);
+    Mem = cast<MemIntrinsicSDNode>(N);
+    unsigned IID = cast<ConstantSDNode>(N->getOperand(1))->getZExtValue();
+    switch (IID) {
+    default:
+      return NULL;
+    case Intrinsic::nvvm_ldg_global_f:
+    case Intrinsic::nvvm_ldg_global_i:
+    case Intrinsic::nvvm_ldg_global_p:
+      IsLDG = true;
+      break;
+    case Intrinsic::nvvm_ldu_global_f:
+    case Intrinsic::nvvm_ldu_global_i:
+    case Intrinsic::nvvm_ldu_global_p:
+      IsLDG = false;
+      break;
+    }
+  } else {
+    Op1 = N->getOperand(1);
+    Mem = cast<MemSDNode>(N);
+  }
+
   unsigned Opcode;
   SDLoc DL(N);
   SDNode *LD;
-  MemSDNode *Mem = cast<MemSDNode>(N);
   SDValue Base, Offset, Addr;
 
-  EVT EltVT = Mem->getMemoryVT().getVectorElementType();
+  EVT EltVT = Mem->getMemoryVT();
+  if (EltVT.isVector()) {
+    EltVT = EltVT.getVectorElementType();
+  }
 
   if (SelectDirectAddr(Op1, Addr)) {
     switch (N->getOpcode()) {
     default:
-      return NULL;
+      return nullptr;
+    case ISD::INTRINSIC_W_CHAIN:
+      if (IsLDG) {
+        switch (EltVT.getSimpleVT().SimpleTy) {
+        default:
+          return nullptr;
+        case MVT::i8:
+          Opcode = NVPTX::INT_PTX_LDG_GLOBAL_i8avar;
+          break;
+        case MVT::i16:
+          Opcode = NVPTX::INT_PTX_LDG_GLOBAL_i16avar;
+          break;
+        case MVT::i32:
+          Opcode = NVPTX::INT_PTX_LDG_GLOBAL_i32avar;
+          break;
+        case MVT::i64:
+          Opcode = NVPTX::INT_PTX_LDG_GLOBAL_i64avar;
+          break;
+        case MVT::f32:
+          Opcode = NVPTX::INT_PTX_LDG_GLOBAL_f32avar;
+          break;
+        case MVT::f64:
+          Opcode = NVPTX::INT_PTX_LDG_GLOBAL_f64avar;
+          break;
+        }
+      } else {
+        switch (EltVT.getSimpleVT().SimpleTy) {
+        default:
+          return nullptr;
+        case MVT::i8:
+          Opcode = NVPTX::INT_PTX_LDU_GLOBAL_i8avar;
+          break;
+        case MVT::i16:
+          Opcode = NVPTX::INT_PTX_LDU_GLOBAL_i16avar;
+          break;
+        case MVT::i32:
+          Opcode = NVPTX::INT_PTX_LDU_GLOBAL_i32avar;
+          break;
+        case MVT::i64:
+          Opcode = NVPTX::INT_PTX_LDU_GLOBAL_i64avar;
+          break;
+        case MVT::f32:
+          Opcode = NVPTX::INT_PTX_LDU_GLOBAL_f32avar;
+          break;
+        case MVT::f64:
+          Opcode = NVPTX::INT_PTX_LDU_GLOBAL_f64avar;
+          break;
+        }
+      }
+      break;
     case NVPTXISD::LDGV2:
       switch (EltVT.getSimpleVT().SimpleTy) {
       default:
-        return NULL;
+        return nullptr;
       case MVT::i8:
         Opcode = NVPTX::INT_PTX_LDG_G_v2i8_ELE_avar;
         break;
@@ -1023,7 +1364,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLDGLDUVector(SDNode *N) {
     case NVPTXISD::LDUV2:
       switch (EltVT.getSimpleVT().SimpleTy) {
       default:
-        return NULL;
+        return nullptr;
       case MVT::i8:
         Opcode = NVPTX::INT_PTX_LDU_G_v2i8_ELE_avar;
         break;
@@ -1047,7 +1388,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLDGLDUVector(SDNode *N) {
     case NVPTXISD::LDGV4:
       switch (EltVT.getSimpleVT().SimpleTy) {
       default:
-        return NULL;
+        return nullptr;
       case MVT::i8:
         Opcode = NVPTX::INT_PTX_LDG_G_v4i8_ELE_avar;
         break;
@@ -1065,7 +1406,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLDGLDUVector(SDNode *N) {
     case NVPTXISD::LDUV4:
       switch (EltVT.getSimpleVT().SimpleTy) {
       default:
-        return NULL;
+        return nullptr;
       case MVT::i8:
         Opcode = NVPTX::INT_PTX_LDU_G_v4i8_ELE_avar;
         break;
@@ -1083,19 +1424,67 @@ SDNode *NVPTXDAGToDAGISel::SelectLDGLDUVector(SDNode *N) {
     }
 
     SDValue Ops[] = { Addr, Chain };
-    LD = CurDAG->getMachineNode(Opcode, DL, N->getVTList(),
-                                ArrayRef<SDValue>(Ops, 2));
+    LD = CurDAG->getMachineNode(Opcode, DL, N->getVTList(), Ops);
   } else if (Subtarget.is64Bit()
                  ? SelectADDRri64(Op1.getNode(), Op1, Base, Offset)
                  : SelectADDRri(Op1.getNode(), Op1, Base, Offset)) {
     if (Subtarget.is64Bit()) {
       switch (N->getOpcode()) {
       default:
-        return NULL;
+        return nullptr;
+      case ISD::INTRINSIC_W_CHAIN:
+        if (IsLDG) {
+          switch (EltVT.getSimpleVT().SimpleTy) {
+          default:
+            return nullptr;
+          case MVT::i8:
+            Opcode = NVPTX::INT_PTX_LDG_GLOBAL_i8ari64;
+            break;
+          case MVT::i16:
+            Opcode = NVPTX::INT_PTX_LDG_GLOBAL_i16ari64;
+            break;
+          case MVT::i32:
+            Opcode = NVPTX::INT_PTX_LDG_GLOBAL_i32ari64;
+            break;
+          case MVT::i64:
+            Opcode = NVPTX::INT_PTX_LDG_GLOBAL_i64ari64;
+            break;
+          case MVT::f32:
+            Opcode = NVPTX::INT_PTX_LDG_GLOBAL_f32ari64;
+            break;
+          case MVT::f64:
+            Opcode = NVPTX::INT_PTX_LDG_GLOBAL_f64ari64;
+            break;
+          }
+        } else {
+          switch (EltVT.getSimpleVT().SimpleTy) {
+          default:
+            return nullptr;
+          case MVT::i8:
+            Opcode = NVPTX::INT_PTX_LDU_GLOBAL_i8ari64;
+            break;
+          case MVT::i16:
+            Opcode = NVPTX::INT_PTX_LDU_GLOBAL_i16ari64;
+            break;
+          case MVT::i32:
+            Opcode = NVPTX::INT_PTX_LDU_GLOBAL_i32ari64;
+            break;
+          case MVT::i64:
+            Opcode = NVPTX::INT_PTX_LDU_GLOBAL_i64ari64;
+            break;
+          case MVT::f32:
+            Opcode = NVPTX::INT_PTX_LDU_GLOBAL_f32ari64;
+            break;
+          case MVT::f64:
+            Opcode = NVPTX::INT_PTX_LDU_GLOBAL_f64ari64;
+            break;
+          }
+        }
+        break;
       case NVPTXISD::LDGV2:
         switch (EltVT.getSimpleVT().SimpleTy) {
         default:
-          return NULL;
+          return nullptr;
         case MVT::i8:
           Opcode = NVPTX::INT_PTX_LDG_G_v2i8_ELE_ari64;
           break;
@@ -1119,7 +1508,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLDGLDUVector(SDNode *N) {
       case NVPTXISD::LDUV2:
         switch (EltVT.getSimpleVT().SimpleTy) {
         default:
-          return NULL;
+          return nullptr;
         case MVT::i8:
           Opcode = NVPTX::INT_PTX_LDU_G_v2i8_ELE_ari64;
           break;
@@ -1143,7 +1532,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLDGLDUVector(SDNode *N) {
       case NVPTXISD::LDGV4:
         switch (EltVT.getSimpleVT().SimpleTy) {
         default:
-          return NULL;
+          return nullptr;
         case MVT::i8:
           Opcode = NVPTX::INT_PTX_LDG_G_v4i8_ELE_ari64;
           break;
@@ -1161,7 +1550,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLDGLDUVector(SDNode *N) {
       case NVPTXISD::LDUV4:
         switch (EltVT.getSimpleVT().SimpleTy) {
         default:
-          return NULL;
+          return nullptr;
         case MVT::i8:
           Opcode = NVPTX::INT_PTX_LDU_G_v4i8_ELE_ari64;
           break;
@@ -1180,11 +1569,60 @@ SDNode *NVPTXDAGToDAGISel::SelectLDGLDUVector(SDNode *N) {
     } else {
       switch (N->getOpcode()) {
       default:
-        return NULL;
+        return nullptr;
+      case ISD::INTRINSIC_W_CHAIN:
+        if (IsLDG) {
+          switch (EltVT.getSimpleVT().SimpleTy) {
+          default:
+            return nullptr;
+          case MVT::i8:
+            Opcode = NVPTX::INT_PTX_LDG_GLOBAL_i8ari;
+            break;
+          case MVT::i16:
+            Opcode = NVPTX::INT_PTX_LDG_GLOBAL_i16ari;
+            break;
+          case MVT::i32:
+            Opcode = NVPTX::INT_PTX_LDG_GLOBAL_i32ari;
+            break;
+          case MVT::i64:
+            Opcode = NVPTX::INT_PTX_LDG_GLOBAL_i64ari;
+            break;
+          case MVT::f32:
+            Opcode = NVPTX::INT_PTX_LDG_GLOBAL_f32ari;
+            break;
+          case MVT::f64:
+            Opcode = NVPTX::INT_PTX_LDG_GLOBAL_f64ari;
+            break;
+          }
+        } else {
+          switch (EltVT.getSimpleVT().SimpleTy) {
+          default:
+            return nullptr;
+          case MVT::i8:
+            Opcode = NVPTX::INT_PTX_LDU_GLOBAL_i8ari;
+            break;
+          case MVT::i16:
+            Opcode = NVPTX::INT_PTX_LDU_GLOBAL_i16ari;
+            break;
+          case MVT::i32:
+            Opcode = NVPTX::INT_PTX_LDU_GLOBAL_i32ari;
+            break;
+          case MVT::i64:
+            Opcode = NVPTX::INT_PTX_LDU_GLOBAL_i64ari;
+            break;
+          case MVT::f32:
+            Opcode = NVPTX::INT_PTX_LDU_GLOBAL_f32ari;
+            break;
+          case MVT::f64:
+            Opcode = NVPTX::INT_PTX_LDU_GLOBAL_f64ari;
+            break;
+          }
+        }
+        break;
       case NVPTXISD::LDGV2:
         switch (EltVT.getSimpleVT().SimpleTy) {
         default:
-          return NULL;
+          return nullptr;
         case MVT::i8:
           Opcode = NVPTX::INT_PTX_LDG_G_v2i8_ELE_ari32;
           break;
@@ -1208,7 +1646,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLDGLDUVector(SDNode *N) {
       case NVPTXISD::LDUV2:
         switch (EltVT.getSimpleVT().SimpleTy) {
         default:
-          return NULL;
+          return nullptr;
         case MVT::i8:
           Opcode = NVPTX::INT_PTX_LDU_G_v2i8_ELE_ari32;
           break;
@@ -1232,7 +1670,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLDGLDUVector(SDNode *N) {
       case NVPTXISD::LDGV4:
         switch (EltVT.getSimpleVT().SimpleTy) {
         default:
-          return NULL;
+          return nullptr;
         case MVT::i8:
           Opcode = NVPTX::INT_PTX_LDG_G_v4i8_ELE_ari32;
           break;
@@ -1250,7 +1688,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLDGLDUVector(SDNode *N) {
       case NVPTXISD::LDUV4:
         switch (EltVT.getSimpleVT().SimpleTy) {
         default:
-          return NULL;
+          return nullptr;
         case MVT::i8:
           Opcode = NVPTX::INT_PTX_LDU_G_v4i8_ELE_ari32;
           break;
@@ -1270,17 +1708,65 @@ SDNode *NVPTXDAGToDAGISel::SelectLDGLDUVector(SDNode *N) {
 
     SDValue Ops[] = { Base, Offset, Chain };
 
-    LD = CurDAG->getMachineNode(Opcode, DL, N->getVTList(),
-                                ArrayRef<SDValue>(Ops, 3));
+    LD = CurDAG->getMachineNode(Opcode, DL, N->getVTList(), Ops);
   } else {
     if (Subtarget.is64Bit()) {
       switch (N->getOpcode()) {
       default:
-        return NULL;
+        return nullptr;
+      case ISD::INTRINSIC_W_CHAIN:
+        if (IsLDG) {
+          switch (EltVT.getSimpleVT().SimpleTy) {
+          default:
+            return nullptr;
+          case MVT::i8:
+            Opcode = NVPTX::INT_PTX_LDG_GLOBAL_i8areg64;
+            break;
+          case MVT::i16:
+            Opcode = NVPTX::INT_PTX_LDG_GLOBAL_i16areg64;
+            break;
+          case MVT::i32:
+            Opcode = NVPTX::INT_PTX_LDG_GLOBAL_i32areg64;
+            break;
+          case MVT::i64:
+            Opcode = NVPTX::INT_PTX_LDG_GLOBAL_i64areg64;
+            break;
+          case MVT::f32:
+            Opcode = NVPTX::INT_PTX_LDG_GLOBAL_f32areg64;
+            break;
+          case MVT::f64:
+            Opcode = NVPTX::INT_PTX_LDG_GLOBAL_f64areg64;
+            break;
+          }
+        } else {
+          switch (EltVT.getSimpleVT().SimpleTy) {
+          default:
+            return nullptr;
+          case MVT::i8:
+            Opcode = NVPTX::INT_PTX_LDU_GLOBAL_i8areg64;
+            break;
+          case MVT::i16:
+            Opcode = NVPTX::INT_PTX_LDU_GLOBAL_i16areg64;
+            break;
+          case MVT::i32:
+            Opcode = NVPTX::INT_PTX_LDU_GLOBAL_i32areg64;
+            break;
+          case MVT::i64:
+            Opcode = NVPTX::INT_PTX_LDU_GLOBAL_i64areg64;
+            break;
+          case MVT::f32:
+            Opcode = NVPTX::INT_PTX_LDU_GLOBAL_f32areg64;
+            break;
+          case MVT::f64:
+            Opcode = NVPTX::INT_PTX_LDU_GLOBAL_f64areg64;
+            break;
+          }
+        }
+        break;
       case NVPTXISD::LDGV2:
         switch (EltVT.getSimpleVT().SimpleTy) {
         default:
-          return NULL;
+          return nullptr;
         case MVT::i8:
           Opcode = NVPTX::INT_PTX_LDG_G_v2i8_ELE_areg64;
           break;
@@ -1304,7 +1790,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLDGLDUVector(SDNode *N) {
       case NVPTXISD::LDUV2:
         switch (EltVT.getSimpleVT().SimpleTy) {
         default:
-          return NULL;
+          return nullptr;
         case MVT::i8:
           Opcode = NVPTX::INT_PTX_LDU_G_v2i8_ELE_areg64;
           break;
@@ -1328,7 +1814,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLDGLDUVector(SDNode *N) {
       case NVPTXISD::LDGV4:
         switch (EltVT.getSimpleVT().SimpleTy) {
         default:
-          return NULL;
+          return nullptr;
         case MVT::i8:
           Opcode = NVPTX::INT_PTX_LDG_G_v4i8_ELE_areg64;
           break;
@@ -1346,7 +1832,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLDGLDUVector(SDNode *N) {
       case NVPTXISD::LDUV4:
         switch (EltVT.getSimpleVT().SimpleTy) {
         default:
-          return NULL;
+          return nullptr;
         case MVT::i8:
           Opcode = NVPTX::INT_PTX_LDU_G_v4i8_ELE_areg64;
           break;
@@ -1365,11 +1851,60 @@ SDNode *NVPTXDAGToDAGISel::SelectLDGLDUVector(SDNode *N) {
     } else {
       switch (N->getOpcode()) {
       default:
-        return NULL;
+        return nullptr;
+      case ISD::INTRINSIC_W_CHAIN:
+        if (IsLDG) {
+          switch (EltVT.getSimpleVT().SimpleTy) {
+          default:
+            return nullptr;
+          case MVT::i8:
+            Opcode = NVPTX::INT_PTX_LDG_GLOBAL_i8areg;
+            break;
+          case MVT::i16:
+            Opcode = NVPTX::INT_PTX_LDG_GLOBAL_i16areg;
+            break;
+          case MVT::i32:
+            Opcode = NVPTX::INT_PTX_LDG_GLOBAL_i32areg;
+            break;
+          case MVT::i64:
+            Opcode = NVPTX::INT_PTX_LDG_GLOBAL_i64areg;
+            break;
+          case MVT::f32:
+            Opcode = NVPTX::INT_PTX_LDG_GLOBAL_f32areg;
+            break;
+          case MVT::f64:
+            Opcode = NVPTX::INT_PTX_LDG_GLOBAL_f64areg;
+            break;
+          }
+        } else {
+          switch (EltVT.getSimpleVT().SimpleTy) {
+          default:
+            return nullptr;
+          case MVT::i8:
+            Opcode = NVPTX::INT_PTX_LDU_GLOBAL_i8areg;
+            break;
+          case MVT::i16:
+            Opcode = NVPTX::INT_PTX_LDU_GLOBAL_i16areg;
+            break;
+          case MVT::i32:
+            Opcode = NVPTX::INT_PTX_LDU_GLOBAL_i32areg;
+            break;
+          case MVT::i64:
+            Opcode = NVPTX::INT_PTX_LDU_GLOBAL_i64areg;
+            break;
+          case MVT::f32:
+            Opcode = NVPTX::INT_PTX_LDU_GLOBAL_f32areg;
+            break;
+          case MVT::f64:
+            Opcode = NVPTX::INT_PTX_LDU_GLOBAL_f64areg;
+            break;
+          }
+        }
+        break;
       case NVPTXISD::LDGV2:
         switch (EltVT.getSimpleVT().SimpleTy) {
         default:
-          return NULL;
+          return nullptr;
         case MVT::i8:
           Opcode = NVPTX::INT_PTX_LDG_G_v2i8_ELE_areg32;
           break;
@@ -1393,7 +1928,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLDGLDUVector(SDNode *N) {
       case NVPTXISD::LDUV2:
         switch (EltVT.getSimpleVT().SimpleTy) {
         default:
-          return NULL;
+          return nullptr;
         case MVT::i8:
           Opcode = NVPTX::INT_PTX_LDU_G_v2i8_ELE_areg32;
           break;
@@ -1417,7 +1952,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLDGLDUVector(SDNode *N) {
       case NVPTXISD::LDGV4:
         switch (EltVT.getSimpleVT().SimpleTy) {
         default:
-          return NULL;
+          return nullptr;
         case MVT::i8:
           Opcode = NVPTX::INT_PTX_LDG_G_v4i8_ELE_areg32;
           break;
@@ -1435,7 +1970,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLDGLDUVector(SDNode *N) {
       case NVPTXISD::LDUV4:
         switch (EltVT.getSimpleVT().SimpleTy) {
         default:
-          return NULL;
+          return nullptr;
         case MVT::i8:
           Opcode = NVPTX::INT_PTX_LDU_G_v4i8_ELE_areg32;
           break;
@@ -1454,12 +1989,11 @@ SDNode *NVPTXDAGToDAGISel::SelectLDGLDUVector(SDNode *N) {
     }
 
     SDValue Ops[] = { Op1, Chain };
-    LD = CurDAG->getMachineNode(Opcode, DL, N->getVTList(),
-                                ArrayRef<SDValue>(Ops, 2));
+    LD = CurDAG->getMachineNode(Opcode, DL, N->getVTList(), Ops);
   }
 
   MachineSDNode::mmo_iterator MemRefs0 = MF->allocateMemRefsArray(1);
-  MemRefs0[0] = cast<MemSDNode>(N)->getMemOperand();
+  MemRefs0[0] = Mem->getMemOperand();
   cast<MachineSDNode>(LD)->setMemRefs(MemRefs0, MemRefs0 + 1);
 
   return LD;
@@ -1469,14 +2003,14 @@ SDNode *NVPTXDAGToDAGISel::SelectStore(SDNode *N) {
   SDLoc dl(N);
   StoreSDNode *ST = cast<StoreSDNode>(N);
   EVT StoreVT = ST->getMemoryVT();
-  SDNode *NVPTXST = NULL;
+  SDNode *NVPTXST = nullptr;
 
   // do not support pre/post inc/dec
   if (ST->isIndexed())
-    return NULL;
+    return nullptr;
 
   if (!StoreVT.isSimple())
-    return NULL;
+    return nullptr;
 
   // Address Space Setting
   unsigned int codeAddrSpace = getCodeAddrSpace(ST, Subtarget);
@@ -1499,7 +2033,7 @@ SDNode *NVPTXDAGToDAGISel::SelectStore(SDNode *N) {
     else if (num == 4)
       vecType = NVPTX::PTXLdStInstCode::V4;
     else
-      return NULL;
+      return nullptr;
   }
 
   // Type Setting: toType + toTypeWidth
@@ -1543,7 +2077,7 @@ SDNode *NVPTXDAGToDAGISel::SelectStore(SDNode *N) {
       Opcode = NVPTX::ST_f64_avar;
       break;
     default:
-      return NULL;
+      return nullptr;
     }
     SDValue Ops[] = { N1, getI32Imm(isVolatile), getI32Imm(codeAddrSpace),
                       getI32Imm(vecType), getI32Imm(toType),
@@ -1572,7 +2106,7 @@ SDNode *NVPTXDAGToDAGISel::SelectStore(SDNode *N) {
       Opcode = NVPTX::ST_f64_asi;
       break;
     default:
-      return NULL;
+      return nullptr;
     }
     SDValue Ops[] = { N1, getI32Imm(isVolatile), getI32Imm(codeAddrSpace),
                       getI32Imm(vecType), getI32Imm(toType),
@@ -1602,7 +2136,7 @@ SDNode *NVPTXDAGToDAGISel::SelectStore(SDNode *N) {
         Opcode = NVPTX::ST_f64_ari_64;
         break;
       default:
-        return NULL;
+        return nullptr;
       }
     } else {
       switch (SourceVT) {
@@ -1625,7 +2159,7 @@ SDNode *NVPTXDAGToDAGISel::SelectStore(SDNode *N) {
         Opcode = NVPTX::ST_f64_ari;
         break;
       default:
-        return NULL;
+        return nullptr;
       }
     }
     SDValue Ops[] = { N1, getI32Imm(isVolatile), getI32Imm(codeAddrSpace),
@@ -1654,7 +2188,7 @@ SDNode *NVPTXDAGToDAGISel::SelectStore(SDNode *N) {
         Opcode = NVPTX::ST_f64_areg_64;
         break;
       default:
-        return NULL;
+        return nullptr;
       }
     } else {
       switch (SourceVT) {
@@ -1677,7 +2211,7 @@ SDNode *NVPTXDAGToDAGISel::SelectStore(SDNode *N) {
         Opcode = NVPTX::ST_f64_areg;
         break;
       default:
-        return NULL;
+        return nullptr;
       }
     }
     SDValue Ops[] = { N1, getI32Imm(isVolatile), getI32Imm(codeAddrSpace),
@@ -1686,7 +2220,7 @@ SDNode *NVPTXDAGToDAGISel::SelectStore(SDNode *N) {
     NVPTXST = CurDAG->getMachineNode(Opcode, dl, MVT::Other, Ops);
   }
 
-  if (NVPTXST != NULL) {
+  if (NVPTXST) {
     MachineSDNode::mmo_iterator MemRefs0 = MF->allocateMemRefsArray(1);
     MemRefs0[0] = cast<MemSDNode>(N)->getMemOperand();
     cast<MachineSDNode>(NVPTXST)->setMemRefs(MemRefs0, MemRefs0 + 1);
@@ -1753,7 +2287,7 @@ SDNode *NVPTXDAGToDAGISel::SelectStoreVector(SDNode *N) {
     N2 = N->getOperand(5);
     break;
   default:
-    return NULL;
+    return nullptr;
   }
 
   StOps.push_back(getI32Imm(IsVolatile));
@@ -1765,11 +2299,11 @@ SDNode *NVPTXDAGToDAGISel::SelectStoreVector(SDNode *N) {
   if (SelectDirectAddr(N2, Addr)) {
     switch (N->getOpcode()) {
     default:
-      return NULL;
+      return nullptr;
     case NVPTXISD::StoreV2:
       switch (EltVT.getSimpleVT().SimpleTy) {
       default:
-        return NULL;
+        return nullptr;
       case MVT::i8:
         Opcode = NVPTX::STV_i8_v2_avar;
         break;
@@ -1793,7 +2327,7 @@ SDNode *NVPTXDAGToDAGISel::SelectStoreVector(SDNode *N) {
     case NVPTXISD::StoreV4:
       switch (EltVT.getSimpleVT().SimpleTy) {
       default:
-        return NULL;
+        return nullptr;
       case MVT::i8:
         Opcode = NVPTX::STV_i8_v4_avar;
         break;
@@ -1815,11 +2349,11 @@ SDNode *NVPTXDAGToDAGISel::SelectStoreVector(SDNode *N) {
                  : SelectADDRsi(N2.getNode(), N2, Base, Offset)) {
     switch (N->getOpcode()) {
     default:
-      return NULL;
+      return nullptr;
     case NVPTXISD::StoreV2:
       switch (EltVT.getSimpleVT().SimpleTy) {
       default:
-        return NULL;
+        return nullptr;
       case MVT::i8:
         Opcode = NVPTX::STV_i8_v2_asi;
         break;
@@ -1843,7 +2377,7 @@ SDNode *NVPTXDAGToDAGISel::SelectStoreVector(SDNode *N) {
     case NVPTXISD::StoreV4:
       switch (EltVT.getSimpleVT().SimpleTy) {
       default:
-        return NULL;
+        return nullptr;
       case MVT::i8:
         Opcode = NVPTX::STV_i8_v4_asi;
         break;
@@ -1867,11 +2401,11 @@ SDNode *NVPTXDAGToDAGISel::SelectStoreVector(SDNode *N) {
     if (Subtarget.is64Bit()) {
       switch (N->getOpcode()) {
       default:
-        return NULL;
+        return nullptr;
       case NVPTXISD::StoreV2:
         switch (EltVT.getSimpleVT().SimpleTy) {
         default:
-          return NULL;
+          return nullptr;
         case MVT::i8:
           Opcode = NVPTX::STV_i8_v2_ari_64;
           break;
@@ -1895,7 +2429,7 @@ SDNode *NVPTXDAGToDAGISel::SelectStoreVector(SDNode *N) {
       case NVPTXISD::StoreV4:
         switch (EltVT.getSimpleVT().SimpleTy) {
         default:
-          return NULL;
+          return nullptr;
         case MVT::i8:
           Opcode = NVPTX::STV_i8_v4_ari_64;
           break;
@@ -1914,11 +2448,11 @@ SDNode *NVPTXDAGToDAGISel::SelectStoreVector(SDNode *N) {
     } else {
       switch (N->getOpcode()) {
       default:
-        return NULL;
+        return nullptr;
       case NVPTXISD::StoreV2:
         switch (EltVT.getSimpleVT().SimpleTy) {
         default:
-          return NULL;
+          return nullptr;
         case MVT::i8:
           Opcode = NVPTX::STV_i8_v2_ari;
           break;
@@ -1942,7 +2476,7 @@ SDNode *NVPTXDAGToDAGISel::SelectStoreVector(SDNode *N) {
       case NVPTXISD::StoreV4:
         switch (EltVT.getSimpleVT().SimpleTy) {
         default:
-          return NULL;
+          return nullptr;
         case MVT::i8:
           Opcode = NVPTX::STV_i8_v4_ari;
           break;
@@ -1965,11 +2499,11 @@ SDNode *NVPTXDAGToDAGISel::SelectStoreVector(SDNode *N) {
     if (Subtarget.is64Bit()) {
       switch (N->getOpcode()) {
       default:
-        return NULL;
+        return nullptr;
       case NVPTXISD::StoreV2:
         switch (EltVT.getSimpleVT().SimpleTy) {
         default:
-          return NULL;
+          return nullptr;
         case MVT::i8:
           Opcode = NVPTX::STV_i8_v2_areg_64;
           break;
@@ -1993,7 +2527,7 @@ SDNode *NVPTXDAGToDAGISel::SelectStoreVector(SDNode *N) {
       case NVPTXISD::StoreV4:
         switch (EltVT.getSimpleVT().SimpleTy) {
         default:
-          return NULL;
+          return nullptr;
         case MVT::i8:
           Opcode = NVPTX::STV_i8_v4_areg_64;
           break;
@@ -2012,11 +2546,11 @@ SDNode *NVPTXDAGToDAGISel::SelectStoreVector(SDNode *N) {
     } else {
       switch (N->getOpcode()) {
       default:
-        return NULL;
+        return nullptr;
       case NVPTXISD::StoreV2:
         switch (EltVT.getSimpleVT().SimpleTy) {
         default:
-          return NULL;
+          return nullptr;
         case MVT::i8:
           Opcode = NVPTX::STV_i8_v2_areg;
           break;
@@ -2040,7 +2574,7 @@ SDNode *NVPTXDAGToDAGISel::SelectStoreVector(SDNode *N) {
       case NVPTXISD::StoreV4:
         switch (EltVT.getSimpleVT().SimpleTy) {
         default:
-          return NULL;
+          return nullptr;
         case MVT::i8:
           Opcode = NVPTX::STV_i8_v4_areg;
           break;
@@ -2081,7 +2615,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLoadParam(SDNode *Node) {
   unsigned VecSize;
   switch (Node->getOpcode()) {
   default:
-    return NULL;
+    return nullptr;
   case NVPTXISD::LoadParam:
     VecSize = 1;
     break;
@@ -2100,11 +2634,11 @@ SDNode *NVPTXDAGToDAGISel::SelectLoadParam(SDNode *Node) {
 
   switch (VecSize) {
   default:
-    return NULL;
+    return nullptr;
   case 1:
     switch (MemVT.getSimpleVT().SimpleTy) {
     default:
-      return NULL;
+      return nullptr;
     case MVT::i1:
       Opc = NVPTX::LoadParamMemI8;
       break;
@@ -2131,7 +2665,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLoadParam(SDNode *Node) {
   case 2:
     switch (MemVT.getSimpleVT().SimpleTy) {
     default:
-      return NULL;
+      return nullptr;
     case MVT::i1:
       Opc = NVPTX::LoadParamMemV2I8;
       break;
@@ -2158,7 +2692,7 @@ SDNode *NVPTXDAGToDAGISel::SelectLoadParam(SDNode *Node) {
   case 4:
     switch (MemVT.getSimpleVT().SimpleTy) {
     default:
-      return NULL;
+      return nullptr;
     case MVT::i1:
       Opc = NVPTX::LoadParamMemV4I8;
       break;
@@ -2211,7 +2745,7 @@ SDNode *NVPTXDAGToDAGISel::SelectStoreRetval(SDNode *N) {
   unsigned NumElts = 1;
   switch (N->getOpcode()) {
   default:
-    return NULL;
+    return nullptr;
   case NVPTXISD::StoreRetval:
     NumElts = 1;
     break;
@@ -2236,11 +2770,11 @@ SDNode *NVPTXDAGToDAGISel::SelectStoreRetval(SDNode *N) {
   unsigned Opcode = 0;
   switch (NumElts) {
   default:
-    return NULL;
+    return nullptr;
   case 1:
     switch (Mem->getMemoryVT().getSimpleVT().SimpleTy) {
     default:
-      return NULL;
+      return nullptr;
     case MVT::i1:
       Opcode = NVPTX::StoreRetvalI8;
       break;
@@ -2267,7 +2801,7 @@ SDNode *NVPTXDAGToDAGISel::SelectStoreRetval(SDNode *N) {
   case 2:
     switch (Mem->getMemoryVT().getSimpleVT().SimpleTy) {
     default:
-      return NULL;
+      return nullptr;
     case MVT::i1:
       Opcode = NVPTX::StoreRetvalV2I8;
       break;
@@ -2294,7 +2828,7 @@ SDNode *NVPTXDAGToDAGISel::SelectStoreRetval(SDNode *N) {
   case 4:
     switch (Mem->getMemoryVT().getSimpleVT().SimpleTy) {
     default:
-      return NULL;
+      return nullptr;
     case MVT::i1:
       Opcode = NVPTX::StoreRetvalV4I8;
       break;
@@ -2337,7 +2871,7 @@ SDNode *NVPTXDAGToDAGISel::SelectStoreParam(SDNode *N) {
   unsigned NumElts = 1;
   switch (N->getOpcode()) {
   default:
-    return NULL;
+    return nullptr;
   case NVPTXISD::StoreParamU32:
   case NVPTXISD::StoreParamS32:
   case NVPTXISD::StoreParam:
@@ -2368,11 +2902,11 @@ SDNode *NVPTXDAGToDAGISel::SelectStoreParam(SDNode *N) {
   default:
     switch (NumElts) {
     default:
-      return NULL;
+      return nullptr;
     case 1:
       switch (Mem->getMemoryVT().getSimpleVT().SimpleTy) {
       default:
-        return NULL;
+        return nullptr;
       case MVT::i1:
         Opcode = NVPTX::StoreParamI8;
         break;
@@ -2399,7 +2933,7 @@ SDNode *NVPTXDAGToDAGISel::SelectStoreParam(SDNode *N) {
     case 2:
       switch (Mem->getMemoryVT().getSimpleVT().SimpleTy) {
       default:
-        return NULL;
+        return nullptr;
       case MVT::i1:
         Opcode = NVPTX::StoreParamV2I8;
         break;
@@ -2426,7 +2960,7 @@ SDNode *NVPTXDAGToDAGISel::SelectStoreParam(SDNode *N) {
     case 4:
       switch (Mem->getMemoryVT().getSimpleVT().SimpleTy) {
       default:
-        return NULL;
+        return nullptr;
       case MVT::i1:
         Opcode = NVPTX::StoreParamV4I8;
         break;
@@ -2481,16 +3015,14 @@ SDNode *NVPTXDAGToDAGISel::SelectStoreParam(SDNode *N) {
 
 SDNode *NVPTXDAGToDAGISel::SelectTextureIntrinsic(SDNode *N) {
   SDValue Chain = N->getOperand(0);
-  SDValue TexRef = N->getOperand(1);
-  SDValue SampRef = N->getOperand(2);
-  SDNode *Ret = NULL;
+  SDNode *Ret = nullptr;
   unsigned Opc = 0;
   SmallVector<SDValue, 8> Ops;
 
   switch (N->getOpcode()) {
-  default: return NULL;
-  case NVPTXISD::Tex1DFloatI32:
-    Opc = NVPTX::TEX_1D_F32_I32;
+  default: return nullptr;
+  case NVPTXISD::Tex1DFloatS32:
+    Opc = NVPTX::TEX_1D_F32_S32;
     break;
   case NVPTXISD::Tex1DFloatFloat:
     Opc = NVPTX::TEX_1D_F32_F32;
@@ -2501,20 +3033,32 @@ SDNode *NVPTXDAGToDAGISel::SelectTextureIntrinsic(SDNode *N) {
   case NVPTXISD::Tex1DFloatFloatGrad:
     Opc = NVPTX::TEX_1D_F32_F32_GRAD;
     break;
-  case NVPTXISD::Tex1DI32I32:
-    Opc = NVPTX::TEX_1D_I32_I32;
+  case NVPTXISD::Tex1DS32S32:
+    Opc = NVPTX::TEX_1D_S32_S32;
+    break;
+  case NVPTXISD::Tex1DS32Float:
+    Opc = NVPTX::TEX_1D_S32_F32;
+    break;
+  case NVPTXISD::Tex1DS32FloatLevel:
+    Opc = NVPTX::TEX_1D_S32_F32_LEVEL;
+    break;
+  case NVPTXISD::Tex1DS32FloatGrad:
+    Opc = NVPTX::TEX_1D_S32_F32_GRAD;
     break;
-  case NVPTXISD::Tex1DI32Float:
-    Opc = NVPTX::TEX_1D_I32_F32;
+  case NVPTXISD::Tex1DU32S32:
+    Opc = NVPTX::TEX_1D_U32_S32;
     break;
-  case NVPTXISD::Tex1DI32FloatLevel:
-    Opc = NVPTX::TEX_1D_I32_F32_LEVEL;
+  case NVPTXISD::Tex1DU32Float:
+    Opc = NVPTX::TEX_1D_U32_F32;
     break;
-  case NVPTXISD::Tex1DI32FloatGrad:
-    Opc = NVPTX::TEX_1D_I32_F32_GRAD;
+  case NVPTXISD::Tex1DU32FloatLevel:
+    Opc = NVPTX::TEX_1D_U32_F32_LEVEL;
     break;
-  case NVPTXISD::Tex1DArrayFloatI32:
-    Opc = NVPTX::TEX_1D_ARRAY_F32_I32;
+  case NVPTXISD::Tex1DU32FloatGrad:
+    Opc = NVPTX::TEX_1D_U32_F32_GRAD;
+    break;
+  case NVPTXISD::Tex1DArrayFloatS32:
+    Opc = NVPTX::TEX_1D_ARRAY_F32_S32;
     break;
   case NVPTXISD::Tex1DArrayFloatFloat:
     Opc = NVPTX::TEX_1D_ARRAY_F32_F32;
@@ -2525,20 +3069,32 @@ SDNode *NVPTXDAGToDAGISel::SelectTextureIntrinsic(SDNode *N) {
   case NVPTXISD::Tex1DArrayFloatFloatGrad:
     Opc = NVPTX::TEX_1D_ARRAY_F32_F32_GRAD;
     break;
-  case NVPTXISD::Tex1DArrayI32I32:
-    Opc = NVPTX::TEX_1D_ARRAY_I32_I32;
+  case NVPTXISD::Tex1DArrayS32S32:
+    Opc = NVPTX::TEX_1D_ARRAY_S32_S32;
+    break;
+  case NVPTXISD::Tex1DArrayS32Float:
+    Opc = NVPTX::TEX_1D_ARRAY_S32_F32;
+    break;
+  case NVPTXISD::Tex1DArrayS32FloatLevel:
+    Opc = NVPTX::TEX_1D_ARRAY_S32_F32_LEVEL;
+    break;
+  case NVPTXISD::Tex1DArrayS32FloatGrad:
+    Opc = NVPTX::TEX_1D_ARRAY_S32_F32_GRAD;
     break;
-  case NVPTXISD::Tex1DArrayI32Float:
-    Opc = NVPTX::TEX_1D_ARRAY_I32_F32;
+  case NVPTXISD::Tex1DArrayU32S32:
+    Opc = NVPTX::TEX_1D_ARRAY_U32_S32;
     break;
-  case NVPTXISD::Tex1DArrayI32FloatLevel:
-    Opc = NVPTX::TEX_1D_ARRAY_I32_F32_LEVEL;
+  case NVPTXISD::Tex1DArrayU32Float:
+    Opc = NVPTX::TEX_1D_ARRAY_U32_F32;
     break;
-  case NVPTXISD::Tex1DArrayI32FloatGrad:
-    Opc = NVPTX::TEX_1D_ARRAY_I32_F32_GRAD;
+  case NVPTXISD::Tex1DArrayU32FloatLevel:
+    Opc = NVPTX::TEX_1D_ARRAY_U32_F32_LEVEL;
     break;
-  case NVPTXISD::Tex2DFloatI32:
-    Opc = NVPTX::TEX_2D_F32_I32;
+  case NVPTXISD::Tex1DArrayU32FloatGrad:
+    Opc = NVPTX::TEX_1D_ARRAY_U32_F32_GRAD;
+    break;
+  case NVPTXISD::Tex2DFloatS32:
+    Opc = NVPTX::TEX_2D_F32_S32;
     break;
   case NVPTXISD::Tex2DFloatFloat:
     Opc = NVPTX::TEX_2D_F32_F32;
@@ -2549,20 +3105,32 @@ SDNode *NVPTXDAGToDAGISel::SelectTextureIntrinsic(SDNode *N) {
   case NVPTXISD::Tex2DFloatFloatGrad:
     Opc = NVPTX::TEX_2D_F32_F32_GRAD;
     break;
-  case NVPTXISD::Tex2DI32I32:
-    Opc = NVPTX::TEX_2D_I32_I32;
+  case NVPTXISD::Tex2DS32S32:
+    Opc = NVPTX::TEX_2D_S32_S32;
+    break;
+  case NVPTXISD::Tex2DS32Float:
+    Opc = NVPTX::TEX_2D_S32_F32;
+    break;
+  case NVPTXISD::Tex2DS32FloatLevel:
+    Opc = NVPTX::TEX_2D_S32_F32_LEVEL;
+    break;
+  case NVPTXISD::Tex2DS32FloatGrad:
+    Opc = NVPTX::TEX_2D_S32_F32_GRAD;
     break;
-  case NVPTXISD::Tex2DI32Float:
-    Opc = NVPTX::TEX_2D_I32_F32;
+  case NVPTXISD::Tex2DU32S32:
+    Opc = NVPTX::TEX_2D_U32_S32;
     break;
-  case NVPTXISD::Tex2DI32FloatLevel:
-    Opc = NVPTX::TEX_2D_I32_F32_LEVEL;
+  case NVPTXISD::Tex2DU32Float:
+    Opc = NVPTX::TEX_2D_U32_F32;
     break;
-  case NVPTXISD::Tex2DI32FloatGrad:
-    Opc = NVPTX::TEX_2D_I32_F32_GRAD;
+  case NVPTXISD::Tex2DU32FloatLevel:
+    Opc = NVPTX::TEX_2D_U32_F32_LEVEL;
     break;
-  case NVPTXISD::Tex2DArrayFloatI32:
-    Opc = NVPTX::TEX_2D_ARRAY_F32_I32;
+  case NVPTXISD::Tex2DU32FloatGrad:
+    Opc = NVPTX::TEX_2D_U32_F32_GRAD;
+    break;
+  case NVPTXISD::Tex2DArrayFloatS32:
+    Opc = NVPTX::TEX_2D_ARRAY_F32_S32;
     break;
   case NVPTXISD::Tex2DArrayFloatFloat:
     Opc = NVPTX::TEX_2D_ARRAY_F32_F32;
@@ -2573,20 +3141,32 @@ SDNode *NVPTXDAGToDAGISel::SelectTextureIntrinsic(SDNode *N) {
   case NVPTXISD::Tex2DArrayFloatFloatGrad:
     Opc = NVPTX::TEX_2D_ARRAY_F32_F32_GRAD;
     break;
-  case NVPTXISD::Tex2DArrayI32I32:
-    Opc = NVPTX::TEX_2D_ARRAY_I32_I32;
+  case NVPTXISD::Tex2DArrayS32S32:
+    Opc = NVPTX::TEX_2D_ARRAY_S32_S32;
+    break;
+  case NVPTXISD::Tex2DArrayS32Float:
+    Opc = NVPTX::TEX_2D_ARRAY_S32_F32;
+    break;
+  case NVPTXISD::Tex2DArrayS32FloatLevel:
+    Opc = NVPTX::TEX_2D_ARRAY_S32_F32_LEVEL;
+    break;
+  case NVPTXISD::Tex2DArrayS32FloatGrad:
+    Opc = NVPTX::TEX_2D_ARRAY_S32_F32_GRAD;
     break;
-  case NVPTXISD::Tex2DArrayI32Float:
-    Opc = NVPTX::TEX_2D_ARRAY_I32_F32;
+  case NVPTXISD::Tex2DArrayU32S32:
+    Opc = NVPTX::TEX_2D_ARRAY_U32_S32;
     break;
-  case NVPTXISD::Tex2DArrayI32FloatLevel:
-    Opc = NVPTX::TEX_2D_ARRAY_I32_F32_LEVEL;
+  case NVPTXISD::Tex2DArrayU32Float:
+    Opc = NVPTX::TEX_2D_ARRAY_U32_F32;
     break;
-  case NVPTXISD::Tex2DArrayI32FloatGrad:
-    Opc = NVPTX::TEX_2D_ARRAY_I32_F32_GRAD;
+  case NVPTXISD::Tex2DArrayU32FloatLevel:
+    Opc = NVPTX::TEX_2D_ARRAY_U32_F32_LEVEL;
     break;
-  case NVPTXISD::Tex3DFloatI32:
-    Opc = NVPTX::TEX_3D_F32_I32;
+  case NVPTXISD::Tex2DArrayU32FloatGrad:
+    Opc = NVPTX::TEX_2D_ARRAY_U32_F32_GRAD;
+    break;
+  case NVPTXISD::Tex3DFloatS32:
+    Opc = NVPTX::TEX_3D_F32_S32;
     break;
   case NVPTXISD::Tex3DFloatFloat:
     Opc = NVPTX::TEX_3D_F32_F32;
@@ -2597,359 +3177,1556 @@ SDNode *NVPTXDAGToDAGISel::SelectTextureIntrinsic(SDNode *N) {
   case NVPTXISD::Tex3DFloatFloatGrad:
     Opc = NVPTX::TEX_3D_F32_F32_GRAD;
     break;
-  case NVPTXISD::Tex3DI32I32:
-    Opc = NVPTX::TEX_3D_I32_I32;
+  case NVPTXISD::Tex3DS32S32:
+    Opc = NVPTX::TEX_3D_S32_S32;
     break;
-  case NVPTXISD::Tex3DI32Float:
-    Opc = NVPTX::TEX_3D_I32_F32;
+  case NVPTXISD::Tex3DS32Float:
+    Opc = NVPTX::TEX_3D_S32_F32;
     break;
-  case NVPTXISD::Tex3DI32FloatLevel:
-    Opc = NVPTX::TEX_3D_I32_F32_LEVEL;
+  case NVPTXISD::Tex3DS32FloatLevel:
+    Opc = NVPTX::TEX_3D_S32_F32_LEVEL;
     break;
-  case NVPTXISD::Tex3DI32FloatGrad:
-    Opc = NVPTX::TEX_3D_I32_F32_GRAD;
+  case NVPTXISD::Tex3DS32FloatGrad:
+    Opc = NVPTX::TEX_3D_S32_F32_GRAD;
     break;
-  }
-
-  Ops.push_back(TexRef);
-  Ops.push_back(SampRef);
-
-  // Copy over indices
-  for (unsigned i = 3; i < N->getNumOperands(); ++i) {
-    Ops.push_back(N->getOperand(i));
-  }
-
-  Ops.push_back(Chain);
-  Ret = CurDAG->getMachineNode(Opc, SDLoc(N), N->getVTList(), Ops);
-  return Ret;
-}
-
-SDNode *NVPTXDAGToDAGISel::SelectSurfaceIntrinsic(SDNode *N) {
-  SDValue Chain = N->getOperand(0);
-  SDValue TexHandle = N->getOperand(1);
-  SDNode *Ret = NULL;
-  unsigned Opc = 0;
-  SmallVector<SDValue, 8> Ops;
-  switch (N->getOpcode()) {
-  default: return NULL;
-  case NVPTXISD::Suld1DI8Trap:
-    Opc = NVPTX::SULD_1D_I8_TRAP;
-    Ops.push_back(TexHandle);
-    Ops.push_back(N->getOperand(2));
-    Ops.push_back(Chain);
+  case NVPTXISD::Tex3DU32S32:
+    Opc = NVPTX::TEX_3D_U32_S32;
     break;
-  case NVPTXISD::Suld1DI16Trap:
-    Opc = NVPTX::SULD_1D_I16_TRAP;
-    Ops.push_back(TexHandle);
-    Ops.push_back(N->getOperand(2));
-    Ops.push_back(Chain);
+  case NVPTXISD::Tex3DU32Float:
+    Opc = NVPTX::TEX_3D_U32_F32;
     break;
-  case NVPTXISD::Suld1DI32Trap:
-    Opc = NVPTX::SULD_1D_I32_TRAP;
-    Ops.push_back(TexHandle);
-    Ops.push_back(N->getOperand(2));
-    Ops.push_back(Chain);
+  case NVPTXISD::Tex3DU32FloatLevel:
+    Opc = NVPTX::TEX_3D_U32_F32_LEVEL;
     break;
-  case NVPTXISD::Suld1DV2I8Trap:
-    Opc = NVPTX::SULD_1D_V2I8_TRAP;
-    Ops.push_back(TexHandle);
-    Ops.push_back(N->getOperand(2));
-    Ops.push_back(Chain);
+  case NVPTXISD::Tex3DU32FloatGrad:
+    Opc = NVPTX::TEX_3D_U32_F32_GRAD;
     break;
-  case NVPTXISD::Suld1DV2I16Trap:
-    Opc = NVPTX::SULD_1D_V2I16_TRAP;
-    Ops.push_back(TexHandle);
-    Ops.push_back(N->getOperand(2));
-    Ops.push_back(Chain);
+  case NVPTXISD::TexCubeFloatFloat:
+    Opc = NVPTX::TEX_CUBE_F32_F32;
     break;
-  case NVPTXISD::Suld1DV2I32Trap:
-    Opc = NVPTX::SULD_1D_V2I32_TRAP;
-    Ops.push_back(TexHandle);
-    Ops.push_back(N->getOperand(2));
-    Ops.push_back(Chain);
+  case NVPTXISD::TexCubeFloatFloatLevel:
+    Opc = NVPTX::TEX_CUBE_F32_F32_LEVEL;
     break;
-  case NVPTXISD::Suld1DV4I8Trap:
-    Opc = NVPTX::SULD_1D_V4I8_TRAP;
-    Ops.push_back(TexHandle);
-    Ops.push_back(N->getOperand(2));
-    Ops.push_back(Chain);
+  case NVPTXISD::TexCubeS32Float:
+    Opc = NVPTX::TEX_CUBE_S32_F32;
     break;
-  case NVPTXISD::Suld1DV4I16Trap:
-    Opc = NVPTX::SULD_1D_V4I16_TRAP;
-    Ops.push_back(TexHandle);
-    Ops.push_back(N->getOperand(2));
-    Ops.push_back(Chain);
+  case NVPTXISD::TexCubeS32FloatLevel:
+    Opc = NVPTX::TEX_CUBE_S32_F32_LEVEL;
     break;
-  case NVPTXISD::Suld1DV4I32Trap:
-    Opc = NVPTX::SULD_1D_V4I32_TRAP;
-    Ops.push_back(TexHandle);
-    Ops.push_back(N->getOperand(2));
-    Ops.push_back(Chain);
+  case NVPTXISD::TexCubeU32Float:
+    Opc = NVPTX::TEX_CUBE_U32_F32;
     break;
-  case NVPTXISD::Suld1DArrayI8Trap:
-    Opc = NVPTX::SULD_1D_ARRAY_I8_TRAP;
-    Ops.push_back(TexHandle);
-    Ops.push_back(N->getOperand(2));
-    Ops.push_back(N->getOperand(3));
-    Ops.push_back(Chain);
+  case NVPTXISD::TexCubeU32FloatLevel:
+    Opc = NVPTX::TEX_CUBE_U32_F32_LEVEL;
     break;
-  case NVPTXISD::Suld1DArrayI16Trap:
-    Opc = NVPTX::SULD_1D_ARRAY_I16_TRAP;
-    Ops.push_back(TexHandle);
+  case NVPTXISD::TexCubeArrayFloatFloat:
+    Opc = NVPTX::TEX_CUBE_ARRAY_F32_F32;
+    break;
+  case NVPTXISD::TexCubeArrayFloatFloatLevel:
+    Opc = NVPTX::TEX_CUBE_ARRAY_F32_F32_LEVEL;
+    break;
+  case NVPTXISD::TexCubeArrayS32Float:
+    Opc = NVPTX::TEX_CUBE_ARRAY_S32_F32;
+    break;
+  case NVPTXISD::TexCubeArrayS32FloatLevel:
+    Opc = NVPTX::TEX_CUBE_ARRAY_S32_F32_LEVEL;
+    break;
+  case NVPTXISD::TexCubeArrayU32Float:
+    Opc = NVPTX::TEX_CUBE_ARRAY_U32_F32;
+    break;
+  case NVPTXISD::TexCubeArrayU32FloatLevel:
+    Opc = NVPTX::TEX_CUBE_ARRAY_U32_F32_LEVEL;
+    break;
+  case NVPTXISD::Tld4R2DFloatFloat:
+    Opc = NVPTX::TLD4_R_2D_F32_F32;
+    break;
+  case NVPTXISD::Tld4G2DFloatFloat:
+    Opc = NVPTX::TLD4_G_2D_F32_F32;
+    break;
+  case NVPTXISD::Tld4B2DFloatFloat:
+    Opc = NVPTX::TLD4_B_2D_F32_F32;
+    break;
+  case NVPTXISD::Tld4A2DFloatFloat:
+    Opc = NVPTX::TLD4_A_2D_F32_F32;
+    break;
+  case NVPTXISD::Tld4R2DS64Float:
+    Opc = NVPTX::TLD4_R_2D_S32_F32;
+    break;
+  case NVPTXISD::Tld4G2DS64Float:
+    Opc = NVPTX::TLD4_G_2D_S32_F32;
+    break;
+  case NVPTXISD::Tld4B2DS64Float:
+    Opc = NVPTX::TLD4_B_2D_S32_F32;
+    break;
+  case NVPTXISD::Tld4A2DS64Float:
+    Opc = NVPTX::TLD4_A_2D_S32_F32;
+    break;
+  case NVPTXISD::Tld4R2DU64Float:
+    Opc = NVPTX::TLD4_R_2D_U32_F32;
+    break;
+  case NVPTXISD::Tld4G2DU64Float:
+    Opc = NVPTX::TLD4_G_2D_U32_F32;
+    break;
+  case NVPTXISD::Tld4B2DU64Float:
+    Opc = NVPTX::TLD4_B_2D_U32_F32;
+    break;
+  case NVPTXISD::Tld4A2DU64Float:
+    Opc = NVPTX::TLD4_A_2D_U32_F32;
+    break;
+  case NVPTXISD::TexUnified1DFloatS32:
+    Opc = NVPTX::TEX_UNIFIED_1D_F32_S32;
+    break;
+  case NVPTXISD::TexUnified1DFloatFloat:
+    Opc = NVPTX::TEX_UNIFIED_1D_F32_F32;
+    break;
+  case NVPTXISD::TexUnified1DFloatFloatLevel:
+    Opc = NVPTX::TEX_UNIFIED_1D_F32_F32_LEVEL;
+    break;
+  case NVPTXISD::TexUnified1DFloatFloatGrad:
+    Opc = NVPTX::TEX_UNIFIED_1D_F32_F32_GRAD;
+    break;
+  case NVPTXISD::TexUnified1DS32S32:
+    Opc = NVPTX::TEX_UNIFIED_1D_S32_S32;
+    break;
+  case NVPTXISD::TexUnified1DS32Float:
+    Opc = NVPTX::TEX_UNIFIED_1D_S32_F32;
+    break;
+  case NVPTXISD::TexUnified1DS32FloatLevel:
+    Opc = NVPTX::TEX_UNIFIED_1D_S32_F32_LEVEL;
+    break;
+  case NVPTXISD::TexUnified1DS32FloatGrad:
+    Opc = NVPTX::TEX_UNIFIED_1D_S32_F32_GRAD;
+    break;
+  case NVPTXISD::TexUnified1DU32S32:
+    Opc = NVPTX::TEX_UNIFIED_1D_U32_S32;
+    break;
+  case NVPTXISD::TexUnified1DU32Float:
+    Opc = NVPTX::TEX_UNIFIED_1D_U32_F32;
+    break;
+  case NVPTXISD::TexUnified1DU32FloatLevel:
+    Opc = NVPTX::TEX_UNIFIED_1D_U32_F32_LEVEL;
+    break;
+  case NVPTXISD::TexUnified1DU32FloatGrad:
+    Opc = NVPTX::TEX_UNIFIED_1D_U32_F32_GRAD;
+    break;
+  case NVPTXISD::TexUnified1DArrayFloatS32:
+    Opc = NVPTX::TEX_UNIFIED_1D_ARRAY_F32_S32;
+    break;
+  case NVPTXISD::TexUnified1DArrayFloatFloat:
+    Opc = NVPTX::TEX_UNIFIED_1D_ARRAY_F32_F32;
+    break;
+  case NVPTXISD::TexUnified1DArrayFloatFloatLevel:
+    Opc = NVPTX::TEX_UNIFIED_1D_ARRAY_F32_F32_LEVEL;
+    break;
+  case NVPTXISD::TexUnified1DArrayFloatFloatGrad:
+    Opc = NVPTX::TEX_UNIFIED_1D_ARRAY_F32_F32_GRAD;
+    break;
+  case NVPTXISD::TexUnified1DArrayS32S32:
+    Opc = NVPTX::TEX_UNIFIED_1D_ARRAY_S32_S32;
+    break;
+  case NVPTXISD::TexUnified1DArrayS32Float:
+    Opc = NVPTX::TEX_UNIFIED_1D_ARRAY_S32_F32;
+    break;
+  case NVPTXISD::TexUnified1DArrayS32FloatLevel:
+    Opc = NVPTX::TEX_UNIFIED_1D_ARRAY_S32_F32_LEVEL;
+    break;
+  case NVPTXISD::TexUnified1DArrayS32FloatGrad:
+    Opc = NVPTX::TEX_UNIFIED_1D_ARRAY_S32_F32_GRAD;
+    break;
+  case NVPTXISD::TexUnified1DArrayU32S32:
+    Opc = NVPTX::TEX_UNIFIED_1D_ARRAY_U32_S32;
+    break;
+  case NVPTXISD::TexUnified1DArrayU32Float:
+    Opc = NVPTX::TEX_UNIFIED_1D_ARRAY_U32_F32;
+    break;
+  case NVPTXISD::TexUnified1DArrayU32FloatLevel:
+    Opc = NVPTX::TEX_UNIFIED_1D_ARRAY_U32_F32_LEVEL;
+    break;
+  case NVPTXISD::TexUnified1DArrayU32FloatGrad:
+    Opc = NVPTX::TEX_UNIFIED_1D_ARRAY_U32_F32_GRAD;
+    break;
+  case NVPTXISD::TexUnified2DFloatS32:
+    Opc = NVPTX::TEX_UNIFIED_2D_F32_S32;
+    break;
+  case NVPTXISD::TexUnified2DFloatFloat:
+    Opc = NVPTX::TEX_UNIFIED_2D_F32_F32;
+    break;
+  case NVPTXISD::TexUnified2DFloatFloatLevel:
+    Opc = NVPTX::TEX_UNIFIED_2D_F32_F32_LEVEL;
+    break;
+  case NVPTXISD::TexUnified2DFloatFloatGrad:
+    Opc = NVPTX::TEX_UNIFIED_2D_F32_F32_GRAD;
+    break;
+  case NVPTXISD::TexUnified2DS32S32:
+    Opc = NVPTX::TEX_UNIFIED_2D_S32_S32;
+    break;
+  case NVPTXISD::TexUnified2DS32Float:
+    Opc = NVPTX::TEX_UNIFIED_2D_S32_F32;
+    break;
+  case NVPTXISD::TexUnified2DS32FloatLevel:
+    Opc = NVPTX::TEX_UNIFIED_2D_S32_F32_LEVEL;
+    break;
+  case NVPTXISD::TexUnified2DS32FloatGrad:
+    Opc = NVPTX::TEX_UNIFIED_2D_S32_F32_GRAD;
+    break;
+  case NVPTXISD::TexUnified2DU32S32:
+    Opc = NVPTX::TEX_UNIFIED_2D_U32_S32;
+    break;
+  case NVPTXISD::TexUnified2DU32Float:
+    Opc = NVPTX::TEX_UNIFIED_2D_U32_F32;
+    break;
+  case NVPTXISD::TexUnified2DU32FloatLevel:
+    Opc = NVPTX::TEX_UNIFIED_2D_U32_F32_LEVEL;
+    break;
+  case NVPTXISD::TexUnified2DU32FloatGrad:
+    Opc = NVPTX::TEX_UNIFIED_2D_U32_F32_GRAD;
+    break;
+  case NVPTXISD::TexUnified2DArrayFloatS32:
+    Opc = NVPTX::TEX_UNIFIED_2D_ARRAY_F32_S32;
+    break;
+  case NVPTXISD::TexUnified2DArrayFloatFloat:
+    Opc = NVPTX::TEX_UNIFIED_2D_ARRAY_F32_F32;
+    break;
+  case NVPTXISD::TexUnified2DArrayFloatFloatLevel:
+    Opc = NVPTX::TEX_UNIFIED_2D_ARRAY_F32_F32_LEVEL;
+    break;
+  case NVPTXISD::TexUnified2DArrayFloatFloatGrad:
+    Opc = NVPTX::TEX_UNIFIED_2D_ARRAY_F32_F32_GRAD;
+    break;
+  case NVPTXISD::TexUnified2DArrayS32S32:
+    Opc = NVPTX::TEX_UNIFIED_2D_ARRAY_S32_S32;
+    break;
+  case NVPTXISD::TexUnified2DArrayS32Float:
+    Opc = NVPTX::TEX_UNIFIED_2D_ARRAY_S32_F32;
+    break;
+  case NVPTXISD::TexUnified2DArrayS32FloatLevel:
+    Opc = NVPTX::TEX_UNIFIED_2D_ARRAY_S32_F32_LEVEL;
+    break;
+  case NVPTXISD::TexUnified2DArrayS32FloatGrad:
+    Opc = NVPTX::TEX_UNIFIED_2D_ARRAY_S32_F32_GRAD;
+    break;
+  case NVPTXISD::TexUnified2DArrayU32S32:
+    Opc = NVPTX::TEX_UNIFIED_2D_ARRAY_U32_S32;
+    break;
+  case NVPTXISD::TexUnified2DArrayU32Float:
+    Opc = NVPTX::TEX_UNIFIED_2D_ARRAY_U32_F32;
+    break;
+  case NVPTXISD::TexUnified2DArrayU32FloatLevel:
+    Opc = NVPTX::TEX_UNIFIED_2D_ARRAY_U32_F32_LEVEL;
+    break;
+  case NVPTXISD::TexUnified2DArrayU32FloatGrad:
+    Opc = NVPTX::TEX_UNIFIED_2D_ARRAY_U32_F32_GRAD;
+    break;
+  case NVPTXISD::TexUnified3DFloatS32:
+    Opc = NVPTX::TEX_UNIFIED_3D_F32_S32;
+    break;
+  case NVPTXISD::TexUnified3DFloatFloat:
+    Opc = NVPTX::TEX_UNIFIED_3D_F32_F32;
+    break;
+  case NVPTXISD::TexUnified3DFloatFloatLevel:
+    Opc = NVPTX::TEX_UNIFIED_3D_F32_F32_LEVEL;
+    break;
+  case NVPTXISD::TexUnified3DFloatFloatGrad:
+    Opc = NVPTX::TEX_UNIFIED_3D_F32_F32_GRAD;
+    break;
+  case NVPTXISD::TexUnified3DS32S32:
+    Opc = NVPTX::TEX_UNIFIED_3D_S32_S32;
+    break;
+  case NVPTXISD::TexUnified3DS32Float:
+    Opc = NVPTX::TEX_UNIFIED_3D_S32_F32;
+    break;
+  case NVPTXISD::TexUnified3DS32FloatLevel:
+    Opc = NVPTX::TEX_UNIFIED_3D_S32_F32_LEVEL;
+    break;
+  case NVPTXISD::TexUnified3DS32FloatGrad:
+    Opc = NVPTX::TEX_UNIFIED_3D_S32_F32_GRAD;
+    break;
+  case NVPTXISD::TexUnified3DU32S32:
+    Opc = NVPTX::TEX_UNIFIED_3D_U32_S32;
+    break;
+  case NVPTXISD::TexUnified3DU32Float:
+    Opc = NVPTX::TEX_UNIFIED_3D_U32_F32;
+    break;
+  case NVPTXISD::TexUnified3DU32FloatLevel:
+    Opc = NVPTX::TEX_UNIFIED_3D_U32_F32_LEVEL;
+    break;
+  case NVPTXISD::TexUnified3DU32FloatGrad:
+    Opc = NVPTX::TEX_UNIFIED_3D_U32_F32_GRAD;
+    break;
+  case NVPTXISD::TexUnifiedCubeFloatFloat:
+    Opc = NVPTX::TEX_UNIFIED_CUBE_F32_F32;
+    break;
+  case NVPTXISD::TexUnifiedCubeFloatFloatLevel:
+    Opc = NVPTX::TEX_UNIFIED_CUBE_F32_F32_LEVEL;
+    break;
+  case NVPTXISD::TexUnifiedCubeS32Float:
+    Opc = NVPTX::TEX_UNIFIED_CUBE_S32_F32;
+    break;
+  case NVPTXISD::TexUnifiedCubeS32FloatLevel:
+    Opc = NVPTX::TEX_UNIFIED_CUBE_S32_F32_LEVEL;
+    break;
+  case NVPTXISD::TexUnifiedCubeU32Float:
+    Opc = NVPTX::TEX_UNIFIED_CUBE_U32_F32;
+    break;
+  case NVPTXISD::TexUnifiedCubeU32FloatLevel:
+    Opc = NVPTX::TEX_UNIFIED_CUBE_U32_F32_LEVEL;
+    break;
+  case NVPTXISD::TexUnifiedCubeArrayFloatFloat:
+    Opc = NVPTX::TEX_UNIFIED_CUBE_ARRAY_F32_F32;
+    break;
+  case NVPTXISD::TexUnifiedCubeArrayFloatFloatLevel:
+    Opc = NVPTX::TEX_UNIFIED_CUBE_ARRAY_F32_F32_LEVEL;
+    break;
+  case NVPTXISD::TexUnifiedCubeArrayS32Float:
+    Opc = NVPTX::TEX_UNIFIED_CUBE_ARRAY_S32_F32;
+    break;
+  case NVPTXISD::TexUnifiedCubeArrayS32FloatLevel:
+    Opc = NVPTX::TEX_UNIFIED_CUBE_ARRAY_S32_F32_LEVEL;
+    break;
+  case NVPTXISD::TexUnifiedCubeArrayU32Float:
+    Opc = NVPTX::TEX_UNIFIED_CUBE_ARRAY_U32_F32;
+    break;
+  case NVPTXISD::TexUnifiedCubeArrayU32FloatLevel:
+    Opc = NVPTX::TEX_UNIFIED_CUBE_ARRAY_U32_F32_LEVEL;
+    break;
+  case NVPTXISD::Tld4UnifiedR2DFloatFloat:
+    Opc = NVPTX::TLD4_UNIFIED_R_2D_F32_F32;
+    break;
+  case NVPTXISD::Tld4UnifiedG2DFloatFloat:
+    Opc = NVPTX::TLD4_UNIFIED_G_2D_F32_F32;
+    break;
+  case NVPTXISD::Tld4UnifiedB2DFloatFloat:
+    Opc = NVPTX::TLD4_UNIFIED_B_2D_F32_F32;
+    break;
+  case NVPTXISD::Tld4UnifiedA2DFloatFloat:
+    Opc = NVPTX::TLD4_UNIFIED_A_2D_F32_F32;
+    break;
+  case NVPTXISD::Tld4UnifiedR2DS64Float:
+    Opc = NVPTX::TLD4_UNIFIED_R_2D_S32_F32;
+    break;
+  case NVPTXISD::Tld4UnifiedG2DS64Float:
+    Opc = NVPTX::TLD4_UNIFIED_G_2D_S32_F32;
+    break;
+  case NVPTXISD::Tld4UnifiedB2DS64Float:
+    Opc = NVPTX::TLD4_UNIFIED_B_2D_S32_F32;
+    break;
+  case NVPTXISD::Tld4UnifiedA2DS64Float:
+    Opc = NVPTX::TLD4_UNIFIED_A_2D_S32_F32;
+    break;
+  case NVPTXISD::Tld4UnifiedR2DU64Float:
+    Opc = NVPTX::TLD4_UNIFIED_R_2D_U32_F32;
+    break;
+  case NVPTXISD::Tld4UnifiedG2DU64Float:
+    Opc = NVPTX::TLD4_UNIFIED_G_2D_U32_F32;
+    break;
+  case NVPTXISD::Tld4UnifiedB2DU64Float:
+    Opc = NVPTX::TLD4_UNIFIED_B_2D_U32_F32;
+    break;
+  case NVPTXISD::Tld4UnifiedA2DU64Float:
+    Opc = NVPTX::TLD4_UNIFIED_A_2D_U32_F32;
+    break;
+  }
+
+  // Copy over operands
+  for (unsigned i = 1; i < N->getNumOperands(); ++i) {
+    Ops.push_back(N->getOperand(i));
+  }
+
+  Ops.push_back(Chain);
+  Ret = CurDAG->getMachineNode(Opc, SDLoc(N), N->getVTList(), Ops);
+  return Ret;
+}
+
+SDNode *NVPTXDAGToDAGISel::SelectSurfaceIntrinsic(SDNode *N) {
+  SDValue Chain = N->getOperand(0);
+  SDValue TexHandle = N->getOperand(1);
+  SDNode *Ret = nullptr;
+  unsigned Opc = 0;
+  SmallVector<SDValue, 8> Ops;
+  switch (N->getOpcode()) {
+  default: return nullptr;
+  case NVPTXISD::Suld1DI8Clamp:
+    Opc = NVPTX::SULD_1D_I8_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DI16Clamp:
+    Opc = NVPTX::SULD_1D_I16_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DI32Clamp:
+    Opc = NVPTX::SULD_1D_I32_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DI64Clamp:
+    Opc = NVPTX::SULD_1D_I64_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DV2I8Clamp:
+    Opc = NVPTX::SULD_1D_V2I8_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DV2I16Clamp:
+    Opc = NVPTX::SULD_1D_V2I16_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DV2I32Clamp:
+    Opc = NVPTX::SULD_1D_V2I32_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DV2I64Clamp:
+    Opc = NVPTX::SULD_1D_V2I64_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DV4I8Clamp:
+    Opc = NVPTX::SULD_1D_V4I8_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DV4I16Clamp:
+    Opc = NVPTX::SULD_1D_V4I16_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DV4I32Clamp:
+    Opc = NVPTX::SULD_1D_V4I32_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DArrayI8Clamp:
+    Opc = NVPTX::SULD_1D_ARRAY_I8_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DArrayI16Clamp:
+    Opc = NVPTX::SULD_1D_ARRAY_I16_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DArrayI32Clamp:
+    Opc = NVPTX::SULD_1D_ARRAY_I32_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DArrayI64Clamp:
+    Opc = NVPTX::SULD_1D_ARRAY_I64_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DArrayV2I8Clamp:
+    Opc = NVPTX::SULD_1D_ARRAY_V2I8_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DArrayV2I16Clamp:
+    Opc = NVPTX::SULD_1D_ARRAY_V2I16_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DArrayV2I32Clamp:
+    Opc = NVPTX::SULD_1D_ARRAY_V2I32_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DArrayV2I64Clamp:
+    Opc = NVPTX::SULD_1D_ARRAY_V2I64_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DArrayV4I8Clamp:
+    Opc = NVPTX::SULD_1D_ARRAY_V4I8_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DArrayV4I16Clamp:
+    Opc = NVPTX::SULD_1D_ARRAY_V4I16_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DArrayV4I32Clamp:
+    Opc = NVPTX::SULD_1D_ARRAY_V4I32_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DI8Clamp:
+    Opc = NVPTX::SULD_2D_I8_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DI16Clamp:
+    Opc = NVPTX::SULD_2D_I16_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DI32Clamp:
+    Opc = NVPTX::SULD_2D_I32_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DI64Clamp:
+    Opc = NVPTX::SULD_2D_I64_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DV2I8Clamp:
+    Opc = NVPTX::SULD_2D_V2I8_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DV2I16Clamp:
+    Opc = NVPTX::SULD_2D_V2I16_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DV2I32Clamp:
+    Opc = NVPTX::SULD_2D_V2I32_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DV2I64Clamp:
+    Opc = NVPTX::SULD_2D_V2I64_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DV4I8Clamp:
+    Opc = NVPTX::SULD_2D_V4I8_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DV4I16Clamp:
+    Opc = NVPTX::SULD_2D_V4I16_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DV4I32Clamp:
+    Opc = NVPTX::SULD_2D_V4I32_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DArrayI8Clamp:
+    Opc = NVPTX::SULD_2D_ARRAY_I8_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DArrayI16Clamp:
+    Opc = NVPTX::SULD_2D_ARRAY_I16_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DArrayI32Clamp:
+    Opc = NVPTX::SULD_2D_ARRAY_I32_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DArrayI64Clamp:
+    Opc = NVPTX::SULD_2D_ARRAY_I64_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DArrayV2I8Clamp:
+    Opc = NVPTX::SULD_2D_ARRAY_V2I8_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DArrayV2I16Clamp:
+    Opc = NVPTX::SULD_2D_ARRAY_V2I16_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DArrayV2I32Clamp:
+    Opc = NVPTX::SULD_2D_ARRAY_V2I32_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DArrayV2I64Clamp:
+    Opc = NVPTX::SULD_2D_ARRAY_V2I64_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DArrayV4I8Clamp:
+    Opc = NVPTX::SULD_2D_ARRAY_V4I8_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DArrayV4I16Clamp:
+    Opc = NVPTX::SULD_2D_ARRAY_V4I16_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DArrayV4I32Clamp:
+    Opc = NVPTX::SULD_2D_ARRAY_V4I32_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld3DI8Clamp:
+    Opc = NVPTX::SULD_3D_I8_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld3DI16Clamp:
+    Opc = NVPTX::SULD_3D_I16_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld3DI32Clamp:
+    Opc = NVPTX::SULD_3D_I32_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld3DI64Clamp:
+    Opc = NVPTX::SULD_3D_I64_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld3DV2I8Clamp:
+    Opc = NVPTX::SULD_3D_V2I8_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld3DV2I16Clamp:
+    Opc = NVPTX::SULD_3D_V2I16_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld3DV2I32Clamp:
+    Opc = NVPTX::SULD_3D_V2I32_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld3DV2I64Clamp:
+    Opc = NVPTX::SULD_3D_V2I64_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld3DV4I8Clamp:
+    Opc = NVPTX::SULD_3D_V4I8_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld3DV4I16Clamp:
+    Opc = NVPTX::SULD_3D_V4I16_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld3DV4I32Clamp:
+    Opc = NVPTX::SULD_3D_V4I32_CLAMP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DI8Trap:
+    Opc = NVPTX::SULD_1D_I8_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DI16Trap:
+    Opc = NVPTX::SULD_1D_I16_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DI32Trap:
+    Opc = NVPTX::SULD_1D_I32_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DI64Trap:
+    Opc = NVPTX::SULD_1D_I64_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DV2I8Trap:
+    Opc = NVPTX::SULD_1D_V2I8_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DV2I16Trap:
+    Opc = NVPTX::SULD_1D_V2I16_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DV2I32Trap:
+    Opc = NVPTX::SULD_1D_V2I32_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DV2I64Trap:
+    Opc = NVPTX::SULD_1D_V2I64_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DV4I8Trap:
+    Opc = NVPTX::SULD_1D_V4I8_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DV4I16Trap:
+    Opc = NVPTX::SULD_1D_V4I16_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DV4I32Trap:
+    Opc = NVPTX::SULD_1D_V4I32_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DArrayI8Trap:
+    Opc = NVPTX::SULD_1D_ARRAY_I8_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DArrayI16Trap:
+    Opc = NVPTX::SULD_1D_ARRAY_I16_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DArrayI32Trap:
+    Opc = NVPTX::SULD_1D_ARRAY_I32_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DArrayI64Trap:
+    Opc = NVPTX::SULD_1D_ARRAY_I64_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DArrayV2I8Trap:
+    Opc = NVPTX::SULD_1D_ARRAY_V2I8_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DArrayV2I16Trap:
+    Opc = NVPTX::SULD_1D_ARRAY_V2I16_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DArrayV2I32Trap:
+    Opc = NVPTX::SULD_1D_ARRAY_V2I32_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DArrayV2I64Trap:
+    Opc = NVPTX::SULD_1D_ARRAY_V2I64_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DArrayV4I8Trap:
+    Opc = NVPTX::SULD_1D_ARRAY_V4I8_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DArrayV4I16Trap:
+    Opc = NVPTX::SULD_1D_ARRAY_V4I16_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DArrayV4I32Trap:
+    Opc = NVPTX::SULD_1D_ARRAY_V4I32_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DI8Trap:
+    Opc = NVPTX::SULD_2D_I8_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DI16Trap:
+    Opc = NVPTX::SULD_2D_I16_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DI32Trap:
+    Opc = NVPTX::SULD_2D_I32_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DI64Trap:
+    Opc = NVPTX::SULD_2D_I64_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DV2I8Trap:
+    Opc = NVPTX::SULD_2D_V2I8_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DV2I16Trap:
+    Opc = NVPTX::SULD_2D_V2I16_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DV2I32Trap:
+    Opc = NVPTX::SULD_2D_V2I32_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DV2I64Trap:
+    Opc = NVPTX::SULD_2D_V2I64_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DV4I8Trap:
+    Opc = NVPTX::SULD_2D_V4I8_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DV4I16Trap:
+    Opc = NVPTX::SULD_2D_V4I16_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DV4I32Trap:
+    Opc = NVPTX::SULD_2D_V4I32_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DArrayI8Trap:
+    Opc = NVPTX::SULD_2D_ARRAY_I8_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DArrayI16Trap:
+    Opc = NVPTX::SULD_2D_ARRAY_I16_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DArrayI32Trap:
+    Opc = NVPTX::SULD_2D_ARRAY_I32_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DArrayI64Trap:
+    Opc = NVPTX::SULD_2D_ARRAY_I64_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DArrayV2I8Trap:
+    Opc = NVPTX::SULD_2D_ARRAY_V2I8_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DArrayV2I16Trap:
+    Opc = NVPTX::SULD_2D_ARRAY_V2I16_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DArrayV2I32Trap:
+    Opc = NVPTX::SULD_2D_ARRAY_V2I32_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DArrayV2I64Trap:
+    Opc = NVPTX::SULD_2D_ARRAY_V2I64_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DArrayV4I8Trap:
+    Opc = NVPTX::SULD_2D_ARRAY_V4I8_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DArrayV4I16Trap:
+    Opc = NVPTX::SULD_2D_ARRAY_V4I16_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DArrayV4I32Trap:
+    Opc = NVPTX::SULD_2D_ARRAY_V4I32_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld3DI8Trap:
+    Opc = NVPTX::SULD_3D_I8_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld3DI16Trap:
+    Opc = NVPTX::SULD_3D_I16_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld3DI32Trap:
+    Opc = NVPTX::SULD_3D_I32_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld3DI64Trap:
+    Opc = NVPTX::SULD_3D_I64_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld3DV2I8Trap:
+    Opc = NVPTX::SULD_3D_V2I8_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld3DV2I16Trap:
+    Opc = NVPTX::SULD_3D_V2I16_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld3DV2I32Trap:
+    Opc = NVPTX::SULD_3D_V2I32_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld3DV2I64Trap:
+    Opc = NVPTX::SULD_3D_V2I64_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld3DV4I8Trap:
+    Opc = NVPTX::SULD_3D_V4I8_TRAP;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld3DV4I16Trap:
+    Opc = NVPTX::SULD_3D_V4I16_TRAP;
+    Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld1DArrayI32Trap:
-    Opc = NVPTX::SULD_1D_ARRAY_I32_TRAP;
+  case NVPTXISD::Suld3DV4I32Trap:
+    Opc = NVPTX::SULD_3D_V4I32_TRAP;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld1DArrayV2I8Trap:
-    Opc = NVPTX::SULD_1D_ARRAY_V2I8_TRAP;
+  case NVPTXISD::Suld1DI8Zero:
+    Opc = NVPTX::SULD_1D_I8_ZERO;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DI16Zero:
+    Opc = NVPTX::SULD_1D_I16_ZERO;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DI32Zero:
+    Opc = NVPTX::SULD_1D_I32_ZERO;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DI64Zero:
+    Opc = NVPTX::SULD_1D_I64_ZERO;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DV2I8Zero:
+    Opc = NVPTX::SULD_1D_V2I8_ZERO;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DV2I16Zero:
+    Opc = NVPTX::SULD_1D_V2I16_ZERO;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DV2I32Zero:
+    Opc = NVPTX::SULD_1D_V2I32_ZERO;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DV2I64Zero:
+    Opc = NVPTX::SULD_1D_V2I64_ZERO;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DV4I8Zero:
+    Opc = NVPTX::SULD_1D_V4I8_ZERO;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DV4I16Zero:
+    Opc = NVPTX::SULD_1D_V4I16_ZERO;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DV4I32Zero:
+    Opc = NVPTX::SULD_1D_V4I32_ZERO;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld1DArrayI8Zero:
+    Opc = NVPTX::SULD_1D_ARRAY_I8_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld1DArrayV2I16Trap:
-    Opc = NVPTX::SULD_1D_ARRAY_V2I16_TRAP;
+  case NVPTXISD::Suld1DArrayI16Zero:
+    Opc = NVPTX::SULD_1D_ARRAY_I16_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld1DArrayV2I32Trap:
-    Opc = NVPTX::SULD_1D_ARRAY_V2I32_TRAP;
+  case NVPTXISD::Suld1DArrayI32Zero:
+    Opc = NVPTX::SULD_1D_ARRAY_I32_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld1DArrayV4I8Trap:
-    Opc = NVPTX::SULD_1D_ARRAY_V4I8_TRAP;
+  case NVPTXISD::Suld1DArrayI64Zero:
+    Opc = NVPTX::SULD_1D_ARRAY_I64_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld1DArrayV4I16Trap:
-    Opc = NVPTX::SULD_1D_ARRAY_V4I16_TRAP;
+  case NVPTXISD::Suld1DArrayV2I8Zero:
+    Opc = NVPTX::SULD_1D_ARRAY_V2I8_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld1DArrayV4I32Trap:
-    Opc = NVPTX::SULD_1D_ARRAY_V4I32_TRAP;
+  case NVPTXISD::Suld1DArrayV2I16Zero:
+    Opc = NVPTX::SULD_1D_ARRAY_V2I16_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld2DI8Trap:
-    Opc = NVPTX::SULD_2D_I8_TRAP;
+  case NVPTXISD::Suld1DArrayV2I32Zero:
+    Opc = NVPTX::SULD_1D_ARRAY_V2I32_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld2DI16Trap:
-    Opc = NVPTX::SULD_2D_I16_TRAP;
+  case NVPTXISD::Suld1DArrayV2I64Zero:
+    Opc = NVPTX::SULD_1D_ARRAY_V2I64_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld2DI32Trap:
-    Opc = NVPTX::SULD_2D_I32_TRAP;
+  case NVPTXISD::Suld1DArrayV4I8Zero:
+    Opc = NVPTX::SULD_1D_ARRAY_V4I8_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld2DV2I8Trap:
-    Opc = NVPTX::SULD_2D_V2I8_TRAP;
+  case NVPTXISD::Suld1DArrayV4I16Zero:
+    Opc = NVPTX::SULD_1D_ARRAY_V4I16_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld2DV2I16Trap:
-    Opc = NVPTX::SULD_2D_V2I16_TRAP;
+  case NVPTXISD::Suld1DArrayV4I32Zero:
+    Opc = NVPTX::SULD_1D_ARRAY_V4I32_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld2DV2I32Trap:
-    Opc = NVPTX::SULD_2D_V2I32_TRAP;
+  case NVPTXISD::Suld2DI8Zero:
+    Opc = NVPTX::SULD_2D_I8_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld2DV4I8Trap:
-    Opc = NVPTX::SULD_2D_V4I8_TRAP;
+  case NVPTXISD::Suld2DI16Zero:
+    Opc = NVPTX::SULD_2D_I16_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld2DV4I16Trap:
-    Opc = NVPTX::SULD_2D_V4I16_TRAP;
+  case NVPTXISD::Suld2DI32Zero:
+    Opc = NVPTX::SULD_2D_I32_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld2DV4I32Trap:
-    Opc = NVPTX::SULD_2D_V4I32_TRAP;
+  case NVPTXISD::Suld2DI64Zero:
+    Opc = NVPTX::SULD_2D_I64_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld2DArrayI8Trap:
-    Opc = NVPTX::SULD_2D_ARRAY_I8_TRAP;
+  case NVPTXISD::Suld2DV2I8Zero:
+    Opc = NVPTX::SULD_2D_V2I8_ZERO;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DV2I16Zero:
+    Opc = NVPTX::SULD_2D_V2I16_ZERO;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DV2I32Zero:
+    Opc = NVPTX::SULD_2D_V2I32_ZERO;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DV2I64Zero:
+    Opc = NVPTX::SULD_2D_V2I64_ZERO;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DV4I8Zero:
+    Opc = NVPTX::SULD_2D_V4I8_ZERO;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DV4I16Zero:
+    Opc = NVPTX::SULD_2D_V4I16_ZERO;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DV4I32Zero:
+    Opc = NVPTX::SULD_2D_V4I32_ZERO;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld2DArrayI8Zero:
+    Opc = NVPTX::SULD_2D_ARRAY_I8_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
     Ops.push_back(N->getOperand(4));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld2DArrayI16Trap:
-    Opc = NVPTX::SULD_2D_ARRAY_I16_TRAP;
+  case NVPTXISD::Suld2DArrayI16Zero:
+    Opc = NVPTX::SULD_2D_ARRAY_I16_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
     Ops.push_back(N->getOperand(4));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld2DArrayI32Trap:
-    Opc = NVPTX::SULD_2D_ARRAY_I32_TRAP;
+  case NVPTXISD::Suld2DArrayI32Zero:
+    Opc = NVPTX::SULD_2D_ARRAY_I32_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
     Ops.push_back(N->getOperand(4));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld2DArrayV2I8Trap:
-    Opc = NVPTX::SULD_2D_ARRAY_V2I8_TRAP;
+  case NVPTXISD::Suld2DArrayI64Zero:
+    Opc = NVPTX::SULD_2D_ARRAY_I64_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
     Ops.push_back(N->getOperand(4));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld2DArrayV2I16Trap:
-    Opc = NVPTX::SULD_2D_ARRAY_V2I16_TRAP;
+  case NVPTXISD::Suld2DArrayV2I8Zero:
+    Opc = NVPTX::SULD_2D_ARRAY_V2I8_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
     Ops.push_back(N->getOperand(4));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld2DArrayV2I32Trap:
-    Opc = NVPTX::SULD_2D_ARRAY_V2I32_TRAP;
+  case NVPTXISD::Suld2DArrayV2I16Zero:
+    Opc = NVPTX::SULD_2D_ARRAY_V2I16_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
     Ops.push_back(N->getOperand(4));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld2DArrayV4I8Trap:
-    Opc = NVPTX::SULD_2D_ARRAY_V4I8_TRAP;
+  case NVPTXISD::Suld2DArrayV2I32Zero:
+    Opc = NVPTX::SULD_2D_ARRAY_V2I32_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
     Ops.push_back(N->getOperand(4));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld2DArrayV4I16Trap:
-    Opc = NVPTX::SULD_2D_ARRAY_V4I16_TRAP;
+  case NVPTXISD::Suld2DArrayV2I64Zero:
+    Opc = NVPTX::SULD_2D_ARRAY_V2I64_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
     Ops.push_back(N->getOperand(4));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld2DArrayV4I32Trap:
-    Opc = NVPTX::SULD_2D_ARRAY_V4I32_TRAP;
+  case NVPTXISD::Suld2DArrayV4I8Zero:
+    Opc = NVPTX::SULD_2D_ARRAY_V4I8_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
     Ops.push_back(N->getOperand(4));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld3DI8Trap:
-    Opc = NVPTX::SULD_3D_I8_TRAP;
+  case NVPTXISD::Suld2DArrayV4I16Zero:
+    Opc = NVPTX::SULD_2D_ARRAY_V4I16_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
     Ops.push_back(N->getOperand(4));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld3DI16Trap:
-    Opc = NVPTX::SULD_3D_I16_TRAP;
+  case NVPTXISD::Suld2DArrayV4I32Zero:
+    Opc = NVPTX::SULD_2D_ARRAY_V4I32_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
     Ops.push_back(N->getOperand(4));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld3DI32Trap:
-    Opc = NVPTX::SULD_3D_I32_TRAP;
+  case NVPTXISD::Suld3DI8Zero:
+    Opc = NVPTX::SULD_3D_I8_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
     Ops.push_back(N->getOperand(4));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld3DV2I8Trap:
-    Opc = NVPTX::SULD_3D_V2I8_TRAP;
+  case NVPTXISD::Suld3DI16Zero:
+    Opc = NVPTX::SULD_3D_I16_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
     Ops.push_back(N->getOperand(4));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld3DV2I16Trap:
-    Opc = NVPTX::SULD_3D_V2I16_TRAP;
+  case NVPTXISD::Suld3DI32Zero:
+    Opc = NVPTX::SULD_3D_I32_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
     Ops.push_back(N->getOperand(4));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld3DV2I32Trap:
-    Opc = NVPTX::SULD_3D_V2I32_TRAP;
+  case NVPTXISD::Suld3DI64Zero:
+    Opc = NVPTX::SULD_3D_I64_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
     Ops.push_back(N->getOperand(4));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld3DV4I8Trap:
-    Opc = NVPTX::SULD_3D_V4I8_TRAP;
+  case NVPTXISD::Suld3DV2I8Zero:
+    Opc = NVPTX::SULD_3D_V2I8_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
     Ops.push_back(N->getOperand(4));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld3DV4I16Trap:
-    Opc = NVPTX::SULD_3D_V4I16_TRAP;
+  case NVPTXISD::Suld3DV2I16Zero:
+    Opc = NVPTX::SULD_3D_V2I16_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
     Ops.push_back(N->getOperand(4));
     Ops.push_back(Chain);
     break;
-  case NVPTXISD::Suld3DV4I32Trap:
-    Opc = NVPTX::SULD_3D_V4I32_TRAP;
+  case NVPTXISD::Suld3DV2I32Zero:
+    Opc = NVPTX::SULD_3D_V2I32_ZERO;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld3DV2I64Zero:
+    Opc = NVPTX::SULD_3D_V2I64_ZERO;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld3DV4I8Zero:
+    Opc = NVPTX::SULD_3D_V4I8_ZERO;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld3DV4I16Zero:
+    Opc = NVPTX::SULD_3D_V4I16_ZERO;
+    Ops.push_back(TexHandle);
+    Ops.push_back(N->getOperand(2));
+    Ops.push_back(N->getOperand(3));
+    Ops.push_back(N->getOperand(4));
+    Ops.push_back(Chain);
+    break;
+  case NVPTXISD::Suld3DV4I32Zero:
+    Opc = NVPTX::SULD_3D_V4I32_ZERO;
     Ops.push_back(TexHandle);
     Ops.push_back(N->getOperand(2));
     Ops.push_back(N->getOperand(3));
@@ -2961,6 +4738,215 @@ SDNode *NVPTXDAGToDAGISel::SelectSurfaceIntrinsic(SDNode *N) {
   return Ret;
 }
 
+
+/// SelectBFE - Look for instruction sequences that can be made more efficient
+/// by using the 'bfe' (bit-field extract) PTX instruction
+SDNode *NVPTXDAGToDAGISel::SelectBFE(SDNode *N) {
+  SDValue LHS = N->getOperand(0);
+  SDValue RHS = N->getOperand(1);
+  SDValue Len;
+  SDValue Start;
+  SDValue Val;
+  bool IsSigned = false;
+
+  if (N->getOpcode() == ISD::AND) {
+    // Canonicalize the operands
+    // We want 'and %val, %mask'
+    if (isa<ConstantSDNode>(LHS) && !isa<ConstantSDNode>(RHS)) {
+      std::swap(LHS, RHS);
+    }
+
+    ConstantSDNode *Mask = dyn_cast<ConstantSDNode>(RHS);
+    if (!Mask) {
+      // We need a constant mask on the RHS of the AND
+      return NULL;
+    }
+
+    // Extract the mask bits
+    uint64_t MaskVal = Mask->getZExtValue();
+    if (!isMask_64(MaskVal)) {
+      // We *could* handle shifted masks here, but doing so would require an
+      // 'and' operation to fix up the low-order bits so we would trade
+      // shr+and for bfe+and, which has the same throughput
+      return NULL;
+    }
+
+    // How many bits are in our mask?
+    uint64_t NumBits = CountTrailingOnes_64(MaskVal);
+    Len = CurDAG->getTargetConstant(NumBits, MVT::i32);
+
+    if (LHS.getOpcode() == ISD::SRL || LHS.getOpcode() == ISD::SRA) {
+      // We have a 'srl/and' pair, extract the effective start bit and length
+      Val = LHS.getNode()->getOperand(0);
+      Start = LHS.getNode()->getOperand(1);
+      ConstantSDNode *StartConst = dyn_cast<ConstantSDNode>(Start);
+      if (StartConst) {
+        uint64_t StartVal = StartConst->getZExtValue();
+        // How many "good" bits do we have left?  "good" is defined here as bits
+        // that exist in the original value, not shifted in.
+        uint64_t GoodBits = Start.getValueType().getSizeInBits() - StartVal;
+        if (NumBits > GoodBits) {
+          // Do not handle the case where bits have been shifted in. In theory
+          // we could handle this, but the cost is likely higher than just
+          // emitting the srl/and pair.
+          return NULL;
+        }
+        Start = CurDAG->getTargetConstant(StartVal, MVT::i32);
+      } else {
+        // Do not handle the case where the shift amount (can be zero if no srl
+        // was found) is not constant. We could handle this case, but it would
+        // require run-time logic that would be more expensive than just
+        // emitting the srl/and pair.
+        return NULL;
+      }
+    } else {
+      // Do not handle the case where the LHS of the and is not a shift. While
+      // it would be trivial to handle this case, it would just transform
+      // 'and' -> 'bfe', but 'and' has higher-throughput.
+      return NULL;
+    }
+  } else if (N->getOpcode() == ISD::SRL || N->getOpcode() == ISD::SRA) {
+    if (LHS->getOpcode() == ISD::AND) {
+      ConstantSDNode *ShiftCnst = dyn_cast<ConstantSDNode>(RHS);
+      if (!ShiftCnst) {
+        // Shift amount must be constant
+        return NULL;
+      }
+
+      uint64_t ShiftAmt = ShiftCnst->getZExtValue();
+
+      SDValue AndLHS = LHS->getOperand(0);
+      SDValue AndRHS = LHS->getOperand(1);
+
+      // Canonicalize the AND to have the mask on the RHS
+      if (isa<ConstantSDNode>(AndLHS)) {
+        std::swap(AndLHS, AndRHS);
+      }
+
+      ConstantSDNode *MaskCnst = dyn_cast<ConstantSDNode>(AndRHS);
+      if (!MaskCnst) {
+        // Mask must be constant
+        return NULL;
+      }
+
+      uint64_t MaskVal = MaskCnst->getZExtValue();
+      uint64_t NumZeros;
+      uint64_t NumBits;
+      if (isMask_64(MaskVal)) {
+        NumZeros = 0;
+        // The number of bits in the result bitfield will be the number of
+        // trailing ones (the AND) minus the number of bits we shift off
+        NumBits = CountTrailingOnes_64(MaskVal) - ShiftAmt;
+      } else if (isShiftedMask_64(MaskVal)) {
+        NumZeros = countTrailingZeros(MaskVal);
+        unsigned NumOnes = CountTrailingOnes_64(MaskVal >> NumZeros);
+        // The number of bits in the result bitfield will be the number of
+        // trailing zeros plus the number of set bits in the mask minus the
+        // number of bits we shift off
+        NumBits = NumZeros + NumOnes - ShiftAmt;
+      } else {
+        // This is not a mask we can handle
+        return NULL;
+      }
+
+      if (ShiftAmt < NumZeros) {
+        // Handling this case would require extra logic that would make this
+        // transformation non-profitable
+        return NULL;
+      }
+
+      Val = AndLHS;
+      Start = CurDAG->getTargetConstant(ShiftAmt, MVT::i32);
+      Len = CurDAG->getTargetConstant(NumBits, MVT::i32);
+    } else if (LHS->getOpcode() == ISD::SHL) {
+      // Here, we have a pattern like:
+      //
+      // (sra (shl val, NN), MM)
+      // or
+      // (srl (shl val, NN), MM)
+      //
+      // If MM >= NN, we can efficiently optimize this with bfe
+      Val = LHS->getOperand(0);
+
+      SDValue ShlRHS = LHS->getOperand(1);
+      ConstantSDNode *ShlCnst = dyn_cast<ConstantSDNode>(ShlRHS);
+      if (!ShlCnst) {
+        // Shift amount must be constant
+        return NULL;
+      }
+      uint64_t InnerShiftAmt = ShlCnst->getZExtValue();
+
+      SDValue ShrRHS = RHS;
+      ConstantSDNode *ShrCnst = dyn_cast<ConstantSDNode>(ShrRHS);
+      if (!ShrCnst) {
+        // Shift amount must be constant
+        return NULL;
+      }
+      uint64_t OuterShiftAmt = ShrCnst->getZExtValue();
+
+      // To avoid extra codegen and be profitable, we need Outer >= Inner
+      if (OuterShiftAmt < InnerShiftAmt) {
+        return NULL;
+      }
+
+      // If the outer shift is more than the type size, we have no bitfield to
+      // extract (since we also check that the inner shift is <= the outer shift
+      // then this also implies that the inner shift is < the type size)
+      if (OuterShiftAmt >= Val.getValueType().getSizeInBits()) {
+        return NULL;
+      }
+
+      Start =
+        CurDAG->getTargetConstant(OuterShiftAmt - InnerShiftAmt, MVT::i32);
+      Len =
+        CurDAG->getTargetConstant(Val.getValueType().getSizeInBits() -
+                                  OuterShiftAmt, MVT::i32);
+
+      if (N->getOpcode() == ISD::SRA) {
+        // If we have a arithmetic right shift, we need to use the signed bfe
+        // variant
+        IsSigned = true;
+      }
+    } else {
+      // No can do...
+      return NULL;
+    }
+  } else {
+    // No can do...
+    return NULL;
+  }
+
+
+  unsigned Opc;
+  // For the BFE operations we form here from "and" and "srl", always use the
+  // unsigned variants.
+  if (Val.getValueType() == MVT::i32) {
+    if (IsSigned) {
+      Opc = NVPTX::BFE_S32rii;
+    } else {
+      Opc = NVPTX::BFE_U32rii;
+    }
+  } else if (Val.getValueType() == MVT::i64) {
+    if (IsSigned) {
+      Opc = NVPTX::BFE_S64rii;
+    } else {
+      Opc = NVPTX::BFE_U64rii;
+    }
+  } else {
+    // We cannot handle this type
+    return NULL;
+  }
+
+  SDValue Ops[] = {
+    Val, Start, Len
+  };
+
+  SDNode *Ret =
+    CurDAG->getMachineNode(Opc, SDLoc(N), N->getVTList(), Ops);
+
+  return Ret;
+}
+
 // SelectDirectAddr - Match a direct address for DAG.
 // A direct address could be a globaladdress or externalsymbol.
 bool NVPTXDAGToDAGISel::SelectDirectAddr(SDValue N, SDValue &Address) {
@@ -3054,18 +5040,11 @@ bool NVPTXDAGToDAGISel::SelectADDRri64(SDNode *OpNode, SDValue Addr,
 
 bool NVPTXDAGToDAGISel::ChkMemSDNodeAddressSpace(SDNode *N,
                                                  unsigned int spN) const {
-  const Value *Src = NULL;
-  // Even though MemIntrinsicSDNode is a subclas of MemSDNode,
-  // the classof() for MemSDNode does not include MemIntrinsicSDNode
-  // (See SelectionDAGNodes.h). So we need to check for both.
+  const Value *Src = nullptr;
   if (MemSDNode *mN = dyn_cast<MemSDNode>(N)) {
     if (spN == 0 && mN->getMemOperand()->getPseudoValue())
       return true;
     Src = mN->getMemOperand()->getValue();
-  } else if (MemSDNode *mN = dyn_cast<MemIntrinsicSDNode>(N)) {
-    if (spN == 0 && mN->getMemOperand()->getPseudoValue())
-      return true;
-    Src = mN->getMemOperand()->getValue();
   }
   if (!Src)
     return false;