[X86][InstCombine] Teach SimplifyDemandedVectorElts to handle masked scalar add/sub/mul/div/max/min intrinsics better.

Now we can remove these intrinsics if element 0 isn't used. Also fix undef element tracking.

llvm-svn: 289636

2 files changed, 14 insertions, 20 deletions

diff --git a/llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp b/llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp
index 6aa9c36ef6a..db5ac7fa570 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp
@@ -1754,6 +1754,7 @@ Instruction *InstCombiner::visitCallInst(CallInst &CI) {
     break;
   }
 
+  // X86 scalar intrinsics simplified with SimplifyDemandedVectorElts.
   case Intrinsic::x86_avx512_mask_add_ss_round:
   case Intrinsic::x86_avx512_mask_div_ss_round:
   case Intrinsic::x86_avx512_mask_mul_ss_round:
@@ -1765,26 +1766,7 @@ Instruction *InstCombiner::visitCallInst(CallInst &CI) {
   case Intrinsic::x86_avx512_mask_mul_sd_round:
   case Intrinsic::x86_avx512_mask_sub_sd_round:
   case Intrinsic::x86_avx512_mask_max_sd_round:
-  case Intrinsic::x86_avx512_mask_min_sd_round: {
-    // These intrinsics only demand the lowest element of the second and third
-    // input vector.
-    bool MadeChange = false;
-    Value *Arg1 = II->getArgOperand(1);
-    Value *Arg2 = II->getArgOperand(2);
-    unsigned VWidth = Arg1->getType()->getVectorNumElements();
-    if (Value *V = SimplifyDemandedVectorEltsLow(Arg1, VWidth, 1)) {
-      II->setArgOperand(1, V);
-      MadeChange = true;
-    }
-    if (Value *V = SimplifyDemandedVectorEltsLow(Arg2, VWidth, 1)) {
-      II->setArgOperand(2, V);
-      MadeChange = true;
-    }
-    if (MadeChange)
-      return II;
-    break;
-  }
-
+  case Intrinsic::x86_avx512_mask_min_sd_round:
   case Intrinsic::x86_fma_vfmadd_ss:
   case Intrinsic::x86_fma_vfmsub_ss:
   case Intrinsic::x86_fma_vfnmadd_ss:
diff --git a/llvm/lib/Transforms/InstCombine/InstCombineSimplifyDemanded.cpp b/llvm/lib/Transforms/InstCombine/InstCombineSimplifyDemanded.cpp
index 1e8432afe15..e9d83b38f54 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineSimplifyDemanded.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineSimplifyDemanded.cpp
@@ -1352,6 +1352,18 @@ Value *InstCombiner::SimplifyDemandedVectorElts(Value *V, APInt DemandedElts,
     // Three input scalar-as-vector operations that work column-wise. The high
     // elements come from operand 0 and the low element is a function of all
     // three inputs.
+    case Intrinsic::x86_avx512_mask_add_ss_round:
+    case Intrinsic::x86_avx512_mask_div_ss_round:
+    case Intrinsic::x86_avx512_mask_mul_ss_round:
+    case Intrinsic::x86_avx512_mask_sub_ss_round:
+    case Intrinsic::x86_avx512_mask_max_ss_round:
+    case Intrinsic::x86_avx512_mask_min_ss_round:
+    case Intrinsic::x86_avx512_mask_add_sd_round:
+    case Intrinsic::x86_avx512_mask_div_sd_round:
+    case Intrinsic::x86_avx512_mask_mul_sd_round:
+    case Intrinsic::x86_avx512_mask_sub_sd_round:
+    case Intrinsic::x86_avx512_mask_max_sd_round:
+    case Intrinsic::x86_avx512_mask_min_sd_round:
     case Intrinsic::x86_fma_vfmadd_ss:
     case Intrinsic::x86_fma_vfmsub_ss:
     case Intrinsic::x86_fma_vfnmadd_ss: