diff --git a/lib/Target/X86/X86InstrFormats.td b/lib/Target/X86/X86InstrFormats.td
index 1c4301c7038..5c422e7cd45 100644
--- a/lib/Target/X86/X86InstrFormats.td
+++ b/lib/Target/X86/X86InstrFormats.td
@@ -220,6 +220,7 @@ class Iseg32 <bits<8> o, Format f, dag outs, dag ins, string asm,
 //   PSI   - SSE1 instructions with TB prefix.
 //   PSIi8 - SSE1 instructions with ImmT == Imm8 and TB prefix.
 //   VSSI  - SSE1 instructions with XS prefix in AVX form.
+//   VPSI  - SSE1 instructions with TB prefix in AVX form.
 
 class SSI<bits<8> o, Format F, dag outs, dag ins, string asm, list<dag> pattern>
       : I<o, F, outs, ins, asm, pattern>, XS, Requires<[HasSSE1]>;
@@ -237,6 +238,10 @@ class VSSI<bits<8> o, Format F, dag outs, dag ins, string asm,
            list<dag> pattern>
       : I<o, F, outs, ins, !strconcat("v", asm), pattern>, XS, VEX_4V,
         Requires<[HasAVX, HasSSE1]>;
+class VPSI<bits<8> o, Format F, dag outs, dag ins, string asm,
+           list<dag> pattern>
+      : I<o, F, outs, ins, !strconcat("v", asm), pattern, SSEPackedSingle>,
+        VEX_4V, Requires<[HasAVX, HasSSE1]>;
 
 // SSE2 Instruction Templates:
 // 
@@ -246,6 +251,7 @@ class VSSI<bits<8> o, Format F, dag outs, dag ins, string asm,
 //   PDI    - SSE2 instructions with TB and OpSize prefixes.
 //   PDIi8  - SSE2 instructions with ImmT == Imm8 and TB and OpSize prefixes.
 //   VSDI   - SSE2 instructions with XD prefix in AVX form.
+//   VPDI   - SSE2 instructions with TB and OpSize prefixes in AVX form.
 
 class SDI<bits<8> o, Format F, dag outs, dag ins, string asm, list<dag> pattern>
       : I<o, F, outs, ins, asm, pattern>, XD, Requires<[HasSSE2]>;
@@ -266,6 +272,10 @@ class VSDI<bits<8> o, Format F, dag outs, dag ins, string asm,
            list<dag> pattern>
       : I<o, F, outs, ins, !strconcat("v", asm), pattern>, XD, VEX_4V,
         Requires<[HasAVX, HasSSE2]>;
+class VPDI<bits<8> o, Format F, dag outs, dag ins, string asm,
+           list<dag> pattern>
+      : I<o, F, outs, ins, !strconcat("v", asm), pattern, SSEPackedDouble>,
+        VEX_4V, OpSize, Requires<[HasAVX, HasSSE2]>;
 
 // SSE3 Instruction Templates:
 // 
diff --git a/lib/Target/X86/X86InstrSSE.td b/lib/Target/X86/X86InstrSSE.td
index fe66d8b011e..945d69f9e26 100644
--- a/lib/Target/X86/X86InstrSSE.td
+++ b/lib/Target/X86/X86InstrSSE.td
@@ -737,6 +737,26 @@ multiclass basic_sse12_fp_binop_rm<bits<8> opc, string OpcodeStr,
     let isCommutable = Commutable;
   }
 
+  def V#NAME#PSrr : VPSI<opc, MRMSrcReg, (outs VR128:$dst),
+               (ins VR128:$src1, VR128:$src2),
+               !strconcat(OpcodeStr,
+                          "ps\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
+               []> {
+    let isCommutable = Commutable;
+    let Constraints = "";
+    let isAsmParserOnly = 1;
+  }
+
+  def V#NAME#PDrr : VPDI<opc, MRMSrcReg, (outs VR128:$dst),
+               (ins VR128:$src1, VR128:$src2),
+               !strconcat(OpcodeStr,
+                          "pd\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
+               []> {
+    let isCommutable = Commutable;
+    let Constraints = "";
+    let isAsmParserOnly = 1;
+  }
+
   // Vector operation, reg+mem.
   def PSrm : PSI<opc, MRMSrcMem, (outs VR128:$dst),
                                  (ins VR128:$src1, f128mem:$src2),
diff --git a/lib/Target/X86/X86MCCodeEmitter.cpp b/lib/Target/X86/X86MCCodeEmitter.cpp
index adf4999ae5a..d553b74a6bc 100644
--- a/lib/Target/X86/X86MCCodeEmitter.cpp
+++ b/lib/Target/X86/X86MCCodeEmitter.cpp
@@ -392,12 +392,16 @@ void X86MCCodeEmitter::EmitVEXOpcodePrefix(uint64_t TSFlags, unsigned &CurByte,
   // functionality of a SIMD prefix
   //
   //  0b00: None
-  //  0b01: 66 (not handled yet)
+  //  0b01: 66
   //  0b10: F3
   //  0b11: F2
   //
   unsigned char VEX_PP = 0;
 
+  // Encode the operand size opcode prefix as needed.
+  if (TSFlags & X86II::OpSize)
+    VEX_PP = 0x01;
+
   switch (TSFlags & X86II::Op0Mask) {
   default: assert(0 && "Invalid prefix!");
   case 0: break;  // No prefix!
diff --git a/test/MC/AsmParser/X86/x86_32-encoding.s b/test/MC/AsmParser/X86/x86_32-encoding.s
index 237f67b82b8..61afcdae33d 100644
--- a/test/MC/AsmParser/X86/x86_32-encoding.s
+++ b/test/MC/AsmParser/X86/x86_32-encoding.s
@@ -10117,3 +10117,34 @@ pshufb	CPI1_0(%rip), %xmm1
 // CHECK:  encoding: [0xc5,0xeb,0x5e,0xac,0xcb,0xef,0xbe,0xad,0xde]
           vdivsd  3735928559(%ebx,%ecx,8), %xmm2, %xmm5
 
+// CHECK: vaddps  %xmm4, %xmm6, %xmm2
+// CHECK: encoding: [0xc5,0xc8,0x58,0xd4]
+          vaddps  %xmm4, %xmm6, %xmm2
+
+// CHECK: vsubps  %xmm4, %xmm6, %xmm2
+// CHECK: encoding: [0xc5,0xc8,0x5c,0xd4]
+          vsubps  %xmm4, %xmm6, %xmm2
+
+// CHECK: vmulps  %xmm4, %xmm6, %xmm2
+// CHECK: encoding: [0xc5,0xc8,0x59,0xd4]
+          vmulps  %xmm4, %xmm6, %xmm2
+
+// CHECK: vdivps  %xmm4, %xmm6, %xmm2
+// CHECK: encoding: [0xc5,0xc8,0x5e,0xd4]
+          vdivps  %xmm4, %xmm6, %xmm2
+
+// CHECK: vaddpd  %xmm4, %xmm6, %xmm2
+// CHECK: encoding: [0xc5,0xc9,0x58,0xd4]
+          vaddpd  %xmm4, %xmm6, %xmm2
+
+// CHECK: vsubpd  %xmm4, %xmm6, %xmm2
+// CHECK: encoding: [0xc5,0xc9,0x5c,0xd4]
+          vsubpd  %xmm4, %xmm6, %xmm2
+
+// CHECK: vmulpd  %xmm4, %xmm6, %xmm2
+// CHECK: encoding: [0xc5,0xc9,0x59,0xd4]
+          vmulpd  %xmm4, %xmm6, %xmm2
+
+// CHECK: vdivpd  %xmm4, %xmm6, %xmm2
+// CHECK: encoding: [0xc5,0xc9,0x5e,0xd4]
+          vdivpd  %xmm4, %xmm6, %xmm2
diff --git a/test/MC/AsmParser/X86/x86_64-encoding.s b/test/MC/AsmParser/X86/x86_64-encoding.s
index 592db3595b8..18196f75170 100644
--- a/test/MC/AsmParser/X86/x86_64-encoding.s
+++ b/test/MC/AsmParser/X86/x86_64-encoding.s
@@ -167,3 +167,35 @@ vmulsd  -4(%rcx,%rbx,8), %xmm10, %xmm11
 // CHECK:   vdivsd  -4(%rcx,%rbx,8), %xmm10, %xmm11
 // CHECK:   encoding: [0xc5,0x2b,0x5e,0x5c,0xd9,0xfc]
 vdivsd  -4(%rcx,%rbx,8), %xmm10, %xmm11
+
+// CHECK: vaddps  %xmm10, %xmm11, %xmm15
+// CHECK: encoding: [0xc4,0x41,0x20,0x58,0xfa]
+vaddps  %xmm10, %xmm11, %xmm15
+
+// CHECK: vsubps  %xmm10, %xmm11, %xmm15
+// CHECK: encoding: [0xc4,0x41,0x20,0x5c,0xfa]
+vsubps  %xmm10, %xmm11, %xmm15
+
+// CHECK: vmulps  %xmm10, %xmm11, %xmm15
+// CHECK: encoding: [0xc4,0x41,0x20,0x59,0xfa]
+vmulps  %xmm10, %xmm11, %xmm15
+
+// CHECK: vdivps  %xmm10, %xmm11, %xmm15
+// CHECK: encoding: [0xc4,0x41,0x20,0x5e,0xfa]
+vdivps  %xmm10, %xmm11, %xmm15
+
+// CHECK: vaddpd  %xmm10, %xmm11, %xmm15
+// CHECK: encoding: [0xc4,0x41,0x21,0x58,0xfa]
+vaddpd  %xmm10, %xmm11, %xmm15
+
+// CHECK: vsubpd  %xmm10, %xmm11, %xmm15
+// CHECK: encoding: [0xc4,0x41,0x21,0x5c,0xfa]
+vsubpd  %xmm10, %xmm11, %xmm15
+
+// CHECK: vmulpd  %xmm10, %xmm11, %xmm15
+// CHECK: encoding: [0xc4,0x41,0x21,0x59,0xfa]
+vmulpd  %xmm10, %xmm11, %xmm15
+
+// CHECK: vdivpd  %xmm10, %xmm11, %xmm15
+// CHECK: encoding: [0xc4,0x41,0x21,0x5e,0xfa]
+vdivpd  %xmm10, %xmm11, %xmm15