[turbofan] Add backend support for float32 operations.

src/compiler/ia32/instruction-codes-ia32.h

Index: src/compiler/ia32/instruction-codes-ia32.h
diff --git a/src/compiler/ia32/instruction-codes-ia32.h b/src/compiler/ia32/instruction-codes-ia32.h
index b7a9e82f6b66dd5abf4106e9c5afe8c0dcca0d6f..2439463bcacf0e890839abe0c6499a35193aae26 100644
--- a/src/compiler/ia32/instruction-codes-ia32.h
+++ b/src/compiler/ia32/instruction-codes-ia32.h
@@ -31,6 +31,14 @@ namespace compiler {
   V(IA32Sar)                       \
   V(IA32Ror)                       \
   V(IA32Lzcnt)                     \
+  V(SSEFloat32Cmp)                 \
+  V(SSEFloat32Add)                 \
+  V(SSEFloat32Sub)                 \
+  V(SSEFloat32Mul)                 \
+  V(SSEFloat32Div)                 \
+  V(SSEFloat32Max)                 \
+  V(SSEFloat32Min)                 \
+  V(SSEFloat32Sqrt)                \
   V(SSEFloat64Cmp)                 \
   V(SSEFloat64Add)                 \
   V(SSEFloat64Sub)                 \
@@ -41,8 +49,8 @@ namespace compiler {
   V(SSEFloat64Min)                 \
   V(SSEFloat64Sqrt)                \
   V(SSEFloat64Round)               \
-  V(SSECvtss2sd)                   \
-  V(SSECvtsd2ss)                   \
+  V(SSEFloat32ToFloat64)           \
+  V(SSEFloat64ToFloat32)           \
   V(SSEFloat64ToInt32)             \
   V(SSEFloat64ToUint32)            \
   V(SSEInt32ToFloat64)             \
@@ -52,6 +60,12 @@ namespace compiler {
   V(SSEFloat64InsertLowWord32)     \
   V(SSEFloat64InsertHighWord32)    \
   V(SSEFloat64LoadLowWord32)       \
+  V(AVXFloat32Add)                 \
+  V(AVXFloat32Sub)                 \
+  V(AVXFloat32Mul)                 \
+  V(AVXFloat32Div)                 \
+  V(AVXFloat32Max)                 \
+  V(AVXFloat32Min)                 \
   V(AVXFloat64Add)                 \
   V(AVXFloat64Sub)                 \
   V(AVXFloat64Mul)                 \