Optimize shufflevector that copies an i64/f64 and zeros the rest.

Summary:
Also ran clang-format on the function. The code added is the last else
if block.

Reviewers: nadav, craig.topper

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D3518

llvm-svn: 207992

2 files changed, 16 insertions, 0 deletions

diff --git a/llvm/test/CodeGen/X86/avx-shuffle.ll b/llvm/test/CodeGen/X86/avx-shuffle.ll
index 02aa617c56c..e472042b2ce 100644
--- a/llvm/test/CodeGen/X86/avx-shuffle.ll
+++ b/llvm/test/CodeGen/X86/avx-shuffle.ll
@@ -306,3 +306,11 @@ define void @test20() {
   store <3 x double> %a1, <3 x double>* undef, align 1
   ret void
 }
+
+define <2 x i64> @test_insert_64_zext(<2 x i64> %i) {
+; CHECK-LABEL: test_insert_64_zext
+; CHECK-NOT: xor
+; CHECK: vmovq
+  %1 = shufflevector <2 x i64> %i, <2 x i64> <i64 0, i64 undef>, <2 x i32> <i32 0, i32 2>
+  ret <2 x i64> %1
+}
diff --git a/llvm/test/CodeGen/X86/sse2.ll b/llvm/test/CodeGen/X86/sse2.ll
index 628dba0b101..0a17eaca411 100644
--- a/llvm/test/CodeGen/X86/sse2.ll
+++ b/llvm/test/CodeGen/X86/sse2.ll
@@ -221,3 +221,11 @@ entry:
  %double2float.i = fptrunc <4 x double> %0 to <4 x float>
  ret <4 x float> %double2float.i
 }
+
+define <2 x i64> @test_insert_64_zext(<2 x i64> %i) {
+; CHECK-LABEL: test_insert_64_zext
+; CHECK-NOT: xor
+; CHECK: movq
+  %1 = shufflevector <2 x i64> %i, <2 x i64> <i64 0, i64 undef>, <2 x i32> <i32 0, i32 2>
+  ret <2 x i64> %1
+}