adding autoUpgrade support to broadcast[f|i]32x2 intrinsics

llvm-svn: 312879

3 files changed, 20 insertions, 31 deletions

diff --git a/llvm/include/llvm/IR/IntrinsicsX86.td b/llvm/include/llvm/IR/IntrinsicsX86.td
index c9c2d3ec11b..44455b6527f 100644
--- a/llvm/include/llvm/IR/IntrinsicsX86.td
+++ b/llvm/include/llvm/IR/IntrinsicsX86.td
@@ -4414,26 +4414,6 @@ let TargetPrefix = "x86" in {  // All intrinsics start with "llvm.x86.".
   def int_x86_avx512_vbroadcast_sd_512 :
         Intrinsic<[llvm_v8f64_ty], [llvm_ptr_ty], [IntrReadMem, IntrArgMemOnly]>;
 
-  def int_x86_avx512_mask_broadcastf32x2_256 :
-          Intrinsic<[llvm_v8f32_ty],
-                    [llvm_v4f32_ty, llvm_v8f32_ty, llvm_i8_ty], [IntrNoMem]>;
-
-  def int_x86_avx512_mask_broadcastf32x2_512 :
-          Intrinsic<[llvm_v16f32_ty],
-                    [llvm_v4f32_ty, llvm_v16f32_ty, llvm_i16_ty], [IntrNoMem]>;
-
-  def int_x86_avx512_mask_broadcasti32x2_128 :
-          Intrinsic<[llvm_v4i32_ty],
-                    [llvm_v4i32_ty, llvm_v4i32_ty, llvm_i8_ty], [IntrNoMem]>;
-
-  def int_x86_avx512_mask_broadcasti32x2_256 :
-          Intrinsic<[llvm_v8i32_ty],
-                    [llvm_v4i32_ty, llvm_v8i32_ty, llvm_i8_ty], [IntrNoMem]>;
-
-  def int_x86_avx512_mask_broadcasti32x2_512 :
-          Intrinsic<[llvm_v16i32_ty],
-                    [llvm_v4i32_ty, llvm_v16i32_ty, llvm_i16_ty], [IntrNoMem]>;
-
    def int_x86_avx512_broadcastmw_512 :
           GCCBuiltin<"__builtin_ia32_broadcastmw512">,
           Intrinsic<[llvm_v16i32_ty], [llvm_i16_ty], [IntrNoMem]>;
diff --git a/llvm/lib/IR/AutoUpgrade.cpp b/llvm/lib/IR/AutoUpgrade.cpp
index 6184ce59dc1..188fdcf6cd5 100644
--- a/llvm/lib/IR/AutoUpgrade.cpp
+++ b/llvm/lib/IR/AutoUpgrade.cpp
@@ -72,7 +72,9 @@ static bool ShouldUpgradeX86Intrinsic(Function *F, StringRef Name) {
   // like to use this information to remove upgrade code for some older
   // intrinsics. It is currently undecided how we will determine that future
   // point.
-  if (Name.startswith("sse2.pcmpeq.") || // Added in 3.1
+  if (Name.startswith("avx512.mask.broadcastf32x2") || // Added in 6.0
+      Name.startswith("avx512.mask.broadcasti32x2") || // Added in 6.0
+      Name.startswith("sse2.pcmpeq.") || // Added in 3.1
       Name.startswith("sse2.pcmpgt.") || // Added in 3.1
       Name.startswith("avx2.pcmpeq.") || // Added in 3.1
       Name.startswith("avx2.pcmpgt.") || // Added in 3.1
@@ -803,6 +805,20 @@ static Value *upgradeIntMinMax(IRBuilder<> &Builder, CallInst &CI,
   return Res;
 }
 
+static Value *upgradeBroadcastf32x2(IRBuilder<> &Builder, CallInst &CI) {
+  Value *Op0 = CI.getArgOperand(0);
+  Value *RetArg = CI.getReturnedArgOperand();
+
+  llvm::VectorType *Ty = RetArg->getType();
+  unsigned NumElts = Ty->getVectorNumElements();
+  uint32_t Indices[NumElts];
+  for(unsigned i = 0; i < NumElts; ++i)
+    Indices[i] = i % 2;
+
+  Value *Res = Builder.CreateShuffleVector(Op0,UndefValue::get(Ty),Indices);
+  return EmitX86Select(Builder, CI.getArgOperand(2), Res, CI.getArgOperand(1));
+}
+
 static Value *upgradeMaskedCompare(IRBuilder<> &Builder, CallInst &CI,
                                    unsigned CC, bool Signed) {
   Value *Op0 = CI.getArgOperand(0);
@@ -1059,6 +1075,9 @@ void llvm::UpgradeIntrinsicCall(CallInst *CI, Function *NewFn) {
                          Name.startswith("avx2.pmaxs") ||
                          Name.startswith("avx512.mask.pmaxs"))) {
       Rep = upgradeIntMinMax(Builder, *CI, ICmpInst::ICMP_SGT);
+    } else if (IsX86 && (Name.startswith("avx512.mask.broadcastf32x2") ||
+                         Name.startswith("avx512.mask.broadcasti32x2"))) {
+        Rep =upgradeBroadcastf32x2(Builder, *CI);
     } else if (IsX86 && (Name == "sse2.pmaxu.b" ||
                          Name == "sse41.pmaxuw" ||
                          Name == "sse41.pmaxud" ||
diff --git a/llvm/lib/Target/X86/X86IntrinsicsInfo.h b/llvm/lib/Target/X86/X86IntrinsicsInfo.h
index 699b6effac3..4672c95e9fa 100644
--- a/llvm/lib/Target/X86/X86IntrinsicsInfo.h
+++ b/llvm/lib/Target/X86/X86IntrinsicsInfo.h
@@ -478,16 +478,6 @@ static const IntrinsicData  IntrinsicsWithoutChain[] = {
                      X86ISD::FADDS_RND, 0),
   X86_INTRINSIC_DATA(avx512_mask_add_ss_round, INTR_TYPE_SCALAR_MASK_RM,
                      X86ISD::FADDS_RND, 0),
-  X86_INTRINSIC_DATA(avx512_mask_broadcastf32x2_256, BRCST32x2_TO_VEC,
-                     X86ISD::VBROADCAST, 0),
-  X86_INTRINSIC_DATA(avx512_mask_broadcastf32x2_512, BRCST32x2_TO_VEC,
-                     X86ISD::VBROADCAST, 0),
-  X86_INTRINSIC_DATA(avx512_mask_broadcasti32x2_128, BRCST32x2_TO_VEC,
-                     X86ISD::VBROADCAST, 0),
-  X86_INTRINSIC_DATA(avx512_mask_broadcasti32x2_256, BRCST32x2_TO_VEC,
-                     X86ISD::VBROADCAST, 0),
-  X86_INTRINSIC_DATA(avx512_mask_broadcasti32x2_512, BRCST32x2_TO_VEC,
-                     X86ISD::VBROADCAST, 0),
   X86_INTRINSIC_DATA(avx512_mask_cmp_pd_128, CMP_MASK_CC, X86ISD::CMPM, 0),
   X86_INTRINSIC_DATA(avx512_mask_cmp_pd_256, CMP_MASK_CC, X86ISD::CMPM, 0),
   X86_INTRINSIC_DATA(avx512_mask_cmp_pd_512, CMP_MASK_CC, X86ISD::CMPM,