[x86] Use AVX in Crankshaft when available.

src/x64/lithium-codegen-x64.cc

Index: src/x64/lithium-codegen-x64.cc
diff --git a/src/x64/lithium-codegen-x64.cc b/src/x64/lithium-codegen-x64.cc
index 8e1f65c771800d5e501d0e8989dc9af8b9741ca2..ffe1a5291000382daced4ccefc326de2dbe9a8a8 100644
--- a/src/x64/lithium-codegen-x64.cc
+++ b/src/x64/lithium-codegen-x64.cc
@@ -2015,23 +2015,45 @@ void LCodeGen::DoArithmeticD(LArithmeticD* instr) {
   XMMRegister left = ToDoubleRegister(instr->left());
   XMMRegister right = ToDoubleRegister(instr->right());
   XMMRegister result = ToDoubleRegister(instr->result());
-  // All operations except MOD are computed in-place.
-  DCHECK(instr->op() == Token::MOD || left.is(result));
   switch (instr->op()) {
     case Token::ADD:
-      __ addsd(left, right);
+      if (CpuFeatures::IsSupported(AVX)) {
+        CpuFeatureScope scope(masm(), AVX);
+        __ vaddsd(result, left, right);
+      } else {
+        DCHECK(result.is(left));
+        __ addsd(left, right);
+      }
       break;
     case Token::SUB:
-       __ subsd(left, right);
+      if (CpuFeatures::IsSupported(AVX)) {
+        CpuFeatureScope scope(masm(), AVX);
+        __ vsubsd(result, left, right);
+      } else {
+        DCHECK(result.is(left));
+        __ subsd(left, right);
+      }
        break;
     case Token::MUL:
-      __ mulsd(left, right);
+      if (CpuFeatures::IsSupported(AVX)) {
+        CpuFeatureScope scope(masm(), AVX);
+        __ vmulsd(result, left, right);
+      } else {
+        DCHECK(result.is(left));
+        __ mulsd(left, right);
+      }
       break;
     case Token::DIV:
-      __ divsd(left, right);
-      // Don't delete this mov. It may improve performance on some CPUs,
-      // when there is a mulsd depending on the result
-      __ movaps(left, left);
+      if (CpuFeatures::IsSupported(AVX)) {
+        CpuFeatureScope scope(masm(), AVX);
+        __ vdivsd(result, left, right);
+      } else {
+        DCHECK(result.is(left));
+        __ divsd(left, right);
+        // Don't delete this mov. It may improve performance on some CPUs,
+        // when there is a mulsd depending on the result
+        __ movaps(left, left);
+      }
       break;
     case Token::MOD: {
       XMMRegister xmm_scratch = double_scratch0();