Simplify a vpermil* with constant mask.

With a constant mask a vpermil* is just a shufflevector. This patch implements
that simplification. This allows us to produce denser code. It should also
allow more folding down the line.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@206801 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Transforms/InstCombine/InstCombineCalls.cpp b/lib/Transforms/InstCombine/InstCombineCalls.cpp
index 19e28fe393f5a5647c68b9c8611059c904a60288..d9b5da3ddc488cc88f70690f51d3a2448cc4f8ec 100644 (file)
--- a/lib/Transforms/InstCombine/InstCombineCalls.cpp
+++ b/lib/Transforms/InstCombine/InstCombineCalls.cpp
@@ -577,6 +577,21 @@ Instruction *InstCombiner::visitCallInst(CallInst &CI) {
     break;
   }
 
+  case Intrinsic::x86_avx_vpermilvar_ps:
+  case Intrinsic::x86_avx_vpermilvar_ps_256:
+  case Intrinsic::x86_avx_vpermilvar_pd:
+  case Intrinsic::x86_avx_vpermilvar_pd_256: {
+    // Convert vpermil* to shufflevector if the mask is constant.
+    Value *V = II->getArgOperand(1);
+    if (auto C = dyn_cast<ConstantDataVector>(V)) {
+      auto V1 = II->getArgOperand(0);
+      auto V2 = UndefValue::get(V1->getType());
+      auto Shuffle = Builder->CreateShuffleVector(V1, V2, C);
+      return ReplaceInstUsesWith(CI, Shuffle);
+    }
+    break;
+  }
+
   case Intrinsic::ppc_altivec_vperm:
     // Turn vperm(V1,V2,mask) -> shuffle(V1,V2,mask) if mask is a constant.
     if (Constant *Mask = dyn_cast<Constant>(II->getArgOperand(2))) {

diff --git a/test/Transforms/InstCombine/vec_demanded_elts.ll b/test/Transforms/InstCombine/vec_demanded_elts.ll
index d12412a9297718c3cf99d1234309a058d4240a14..f75ea032735d8893bf5d7c67130dc044f92ce948 100644 (file)
--- a/test/Transforms/InstCombine/vec_demanded_elts.ll
+++ b/test/Transforms/InstCombine/vec_demanded_elts.ll
@@ -209,4 +209,34 @@ define <4 x float> @test_select(float %f, float %g) {
   ret <4 x float> %ret
 }
 
+declare <4 x float> @llvm.x86.avx.vpermilvar.ps(<4 x float>, <4 x i32>)
+define <4 x float> @test_vpermilvar_ps(<4 x float> %v) {
+; CHECK-LABEL: @test_vpermilvar_ps(
+; CHECK: shufflevector <4 x float> %v, <4 x float> undef, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
+  %a = tail call <4 x float> @llvm.x86.avx.vpermilvar.ps(<4 x float> %v, <4 x i32> <i32 3, i32 2, i32 1, i32 0>)
+  ret <4 x float> %a
+}
+
+declare <8 x float> @llvm.x86.avx.vpermilvar.ps.256(<8 x float>, <8 x i32>)
+define <8 x float> @test_vpermilvar_ps_256(<8 x float> %v) {
+; CHECK-LABEL: @test_vpermilvar_ps_256(
+; CHECK: shufflevector <8 x float> %v, <8 x float> undef, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+  %a = tail call <8 x float> @llvm.x86.avx.vpermilvar.ps.256(<8 x float> %v, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>)
+  ret <8 x float> %a
+}
 
+declare <2 x double> @llvm.x86.avx.vpermilvar.pd(<2 x double>, <2 x i32>)
+define <2 x double> @test_vpermilvar_pd(<2 x double> %v) {
+; CHECK-LABEL: @test_vpermilvar_pd(
+; CHECK: shufflevector <2 x double> %v, <2 x double> undef, <2 x i32> <i32 1, i32 0>
+  %a = tail call <2 x double> @llvm.x86.avx.vpermilvar.pd(<2 x double> %v, <2 x i32> <i32 1, i32 0>)
+  ret <2 x double> %a
+}
+
+declare <4 x double> @llvm.x86.avx.vpermilvar.pd.256(<4 x double>, <4 x i32>)
+define <4 x double> @test_vpermilvar_pd_256(<4 x double> %v) {
+; CHECK-LABEL: @test_vpermilvar_pd_256(
+; CHECK: shufflevector <4 x double> %v, <4 x double> undef, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
+  %a = tail call <4 x double> @llvm.x86.avx.vpermilvar.pd.256(<4 x double> %v, <4 x i32> <i32 3, i32 2, i32 1, i32 0>)
+  ret <4 x double> %a
+}