AVX-512: Added VMOVD, VMOVQ, VMOVSS, VMOVSD instructions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188637 91177308-0d34-0410-b5e6-96231b3b80d8

1 files changed, 36 insertions, 12 deletions

diff --git a/lib/Target/X86/X86InstrInfo.cpp b/lib/Target/X86/X86InstrInfo.cpp
index b773768..71df2bb 100644
--- a/lib/Target/X86/X86InstrInfo.cpp
+++ b/lib/Target/X86/X86InstrInfo.cpp
@@ -2910,17 +2910,20 @@ static unsigned CopyToFromAsymmetricReg(unsigned DestReg, unsigned SrcReg,
   // SrcReg(GR64)  -> DestReg(VR64)
 
   bool HasAVX = Subtarget.hasAVX();
+  bool HasAVX512 = Subtarget.hasAVX512();
   if (X86::GR64RegClass.contains(DestReg)) {
-    if (X86::VR128RegClass.contains(SrcReg))
+    if (X86::VR128XRegClass.contains(SrcReg))
       // Copy from a VR128 register to a GR64 register.
-      return HasAVX ? X86::VMOVPQIto64rr : X86::MOVPQIto64rr;
+      return HasAVX512 ? X86::VMOVPQIto64Zrr: (HasAVX ? X86::VMOVPQIto64rr :
+                                               X86::MOVPQIto64rr);
     if (X86::VR64RegClass.contains(SrcReg))
       // Copy from a VR64 register to a GR64 register.
       return X86::MOVSDto64rr;
   } else if (X86::GR64RegClass.contains(SrcReg)) {
     // Copy from a GR64 register to a VR128 register.
-    if (X86::VR128RegClass.contains(DestReg))
-      return HasAVX ? X86::VMOV64toPQIrr : X86::MOV64toPQIrr;
+    if (X86::VR128XRegClass.contains(DestReg))
+      return HasAVX512 ? X86::VMOV64toPQIZrr: (HasAVX ? X86::VMOV64toPQIrr :
+                                               X86::MOV64toPQIrr);
     // Copy from a GR64 register to a VR64 register.
     if (X86::VR64RegClass.contains(DestReg))
       return X86::MOV64toSDrr;
@@ -2929,13 +2932,13 @@ static unsigned CopyToFromAsymmetricReg(unsigned DestReg, unsigned SrcReg,
   // SrcReg(FR32) -> DestReg(GR32)
   // SrcReg(GR32) -> DestReg(FR32)
 
-  if (X86::GR32RegClass.contains(DestReg) && X86::FR32RegClass.contains(SrcReg))
+  if (X86::GR32RegClass.contains(DestReg) && X86::FR32XRegClass.contains(SrcReg))
     // Copy from a FR32 register to a GR32 register.
-    return HasAVX ? X86::VMOVSS2DIrr : X86::MOVSS2DIrr;
+    return HasAVX512 ? X86::VMOVSS2DIZrr : (HasAVX ? X86::VMOVSS2DIrr : X86::MOVSS2DIrr);
 
-  if (X86::FR32RegClass.contains(DestReg) && X86::GR32RegClass.contains(SrcReg))
+  if (X86::FR32XRegClass.contains(DestReg) && X86::GR32RegClass.contains(SrcReg))
     // Copy from a GR32 register to a FR32 register.
-    return HasAVX ? X86::VMOVDI2SSrr : X86::MOVDI2SSrr;
+    return HasAVX512 ? X86::VMOVDI2SSZrr : (HasAVX ? X86::VMOVDI2SSrr : X86::MOVDI2SSrr);
 
   return 0;
 }
@@ -3040,6 +3043,21 @@ static unsigned getLoadStoreRegOpcode(unsigned Reg,
                                       bool isStackAligned,
                                       const TargetMachine &TM,
                                       bool load) {
+  if (TM.getSubtarget<X86Subtarget>().hasAVX512()) {
+    if (X86::VK8RegClass.hasSubClassEq(RC)  || 
+      X86::VK16RegClass.hasSubClassEq(RC))
+      return load ? X86::KMOVWkm : X86::KMOVWmk;
+
+    if (X86::FR32XRegClass.hasSubClassEq(RC))
+      return load ? X86::VMOVSSZrm : X86::VMOVSSZmr;
+    if (X86::FR64XRegClass.hasSubClassEq(RC))
+      return load ? X86::VMOVSDZrm : X86::VMOVSDZmr;
+    if (X86::VR128XRegClass.hasSubClassEq(RC) ||
+        X86::VR256XRegClass.hasSubClassEq(RC) ||
+        X86::VR512RegClass.hasSubClassEq(RC))
+      return load ? X86::VMOVUPSZrm : X86::VMOVUPSZmr;
+  }
+
   bool HasAVX = TM.getSubtarget<X86Subtarget>().hasAVX();
   switch (RC->getSize()) {
   default:
@@ -3099,6 +3117,12 @@ static unsigned getLoadStoreRegOpcode(unsigned Reg,
       return load ? X86::VMOVAPSYrm : X86::VMOVAPSYmr;
     else
       return load ? X86::VMOVUPSYrm : X86::VMOVUPSYmr;
+  case 64:
+    assert(X86::VR512RegClass.hasSubClassEq(RC) && "Unknown 64-byte regclass");
+    if (isStackAligned)
+      return load ? X86::VMOVAPSZrm : X86::VMOVAPSZmr;
+    else
+      return load ? X86::VMOVUPSZrm : X86::VMOVUPSZmr;
   }
 }
 
@@ -3125,7 +3149,7 @@ void X86InstrInfo::storeRegToStackSlot(MachineBasicBlock &MBB,
   const MachineFunction &MF = *MBB.getParent();
   assert(MF.getFrameInfo()->getObjectSize(FrameIdx) >= RC->getSize() &&
          "Stack slot too small for store");
-  unsigned Alignment = RC->getSize() == 32 ? 32 : 16;
+  unsigned Alignment = std::max<uint32_t>(RC->getSize(), 16);
   bool isAligned = (TM.getFrameLowering()->getStackAlignment() >= Alignment) ||
     RI.canRealignStack(MF);
   unsigned Opc = getStoreRegOpcode(SrcReg, RC, isAligned, TM);
@@ -3141,7 +3165,7 @@ void X86InstrInfo::storeRegToAddr(MachineFunction &MF, unsigned SrcReg,
                                   MachineInstr::mmo_iterator MMOBegin,
                                   MachineInstr::mmo_iterator MMOEnd,
                                   SmallVectorImpl<MachineInstr*> &NewMIs) const {
-  unsigned Alignment = RC->getSize() == 32 ? 32 : 16;
+  unsigned Alignment = std::max<uint32_t>(RC->getSize(), 16);
   bool isAligned = MMOBegin != MMOEnd &&
                    (*MMOBegin)->getAlignment() >= Alignment;
   unsigned Opc = getStoreRegOpcode(SrcReg, RC, isAligned, TM);
@@ -3161,7 +3185,7 @@ void X86InstrInfo::loadRegFromStackSlot(MachineBasicBlock &MBB,
                                         const TargetRegisterClass *RC,
                                         const TargetRegisterInfo *TRI) const {
   const MachineFunction &MF = *MBB.getParent();
-  unsigned Alignment = RC->getSize() == 32 ? 32 : 16;
+  unsigned Alignment = std::max<uint32_t>(RC->getSize(), 16);
   bool isAligned = (TM.getFrameLowering()->getStackAlignment() >= Alignment) ||
     RI.canRealignStack(MF);
   unsigned Opc = getLoadRegOpcode(DestReg, RC, isAligned, TM);
@@ -3175,7 +3199,7 @@ void X86InstrInfo::loadRegFromAddr(MachineFunction &MF, unsigned DestReg,
                                  MachineInstr::mmo_iterator MMOBegin,
                                  MachineInstr::mmo_iterator MMOEnd,
                                  SmallVectorImpl<MachineInstr*> &NewMIs) const {
-  unsigned Alignment = RC->getSize() == 32 ? 32 : 16;
+  unsigned Alignment = std::max<uint32_t>(RC->getSize(), 16);
   bool isAligned = MMOBegin != MMOEnd &&
                    (*MMOBegin)->getAlignment() >= Alignment;
   unsigned Opc = getLoadRegOpcode(DestReg, RC, isAligned, TM);