[X86] Use Ld scheduler classes for instructions with folded loads.

llvm-svn: 320459

4 files changed, 73 insertions, 73 deletions

diff --git a/llvm/lib/Target/X86/X86InstrAVX512.td b/llvm/lib/Target/X86/X86InstrAVX512.td
index 8cb1cb8a416..9753464ad76 100644
--- a/llvm/lib/Target/X86/X86InstrAVX512.td
+++ b/llvm/lib/Target/X86/X86InstrAVX512.td
@@ -6037,7 +6037,7 @@ multiclass avx512_fma3p_213_rm<bits<8> opc, string OpcodeStr, SDNode OpNode,
           (ins _.RC:$src2, _.MemOp:$src3),
           OpcodeStr, "$src3, $src2", "$src2, $src3",
           (_.VT (OpNode _.RC:$src2, _.RC:$src1, (_.LdFrag addr:$src3))),
-          NoItinerary, 1, 0>, AVX512FMA3Base, Sched<[WriteFMA, ReadAfterLd]>;
+          NoItinerary, 1, 0>, AVX512FMA3Base, Sched<[WriteFMALd, ReadAfterLd]>;
 
   defm mb: AVX512_maskable_3src<opc, MRMSrcMem, _, (outs _.RC:$dst),
             (ins _.RC:$src2, _.ScalarMemOp:$src3),
@@ -6046,7 +6046,7 @@ multiclass avx512_fma3p_213_rm<bits<8> opc, string OpcodeStr, SDNode OpNode,
             (OpNode _.RC:$src2,
              _.RC:$src1,(_.VT (X86VBroadcast (_.ScalarLdFrag addr:$src3)))),
              NoItinerary, 1, 0>, AVX512FMA3Base, EVEX_B,
-             Sched<[WriteFMA, ReadAfterLd]>;
+             Sched<[WriteFMALd, ReadAfterLd]>;
   }
 }
 
@@ -6105,7 +6105,7 @@ multiclass avx512_fma3p_231_rm<bits<8> opc, string OpcodeStr, SDNode OpNode,
           (ins _.RC:$src2, _.MemOp:$src3),
           OpcodeStr, "$src3, $src2", "$src2, $src3",
           (_.VT (OpNode _.RC:$src2, (_.LdFrag addr:$src3), _.RC:$src1)),
-          NoItinerary, 1, 0>, AVX512FMA3Base, Sched<[WriteFMA, ReadAfterLd]>;
+          NoItinerary, 1, 0>, AVX512FMA3Base, Sched<[WriteFMALd, ReadAfterLd]>;
 
   defm mb: AVX512_maskable_3src<opc, MRMSrcMem, _, (outs _.RC:$dst),
          (ins _.RC:$src2, _.ScalarMemOp:$src3),
@@ -6114,7 +6114,7 @@ multiclass avx512_fma3p_231_rm<bits<8> opc, string OpcodeStr, SDNode OpNode,
          (_.VT (OpNode _.RC:$src2,
                       (_.VT (X86VBroadcast(_.ScalarLdFrag addr:$src3))),
                       _.RC:$src1)), NoItinerary, 1, 0>, AVX512FMA3Base, EVEX_B,
-         Sched<[WriteFMA, ReadAfterLd]>;
+         Sched<[WriteFMALd, ReadAfterLd]>;
   }
 }
 
@@ -6175,7 +6175,7 @@ multiclass avx512_fma3p_132_rm<bits<8> opc, string OpcodeStr, SDNode OpNode,
           (ins _.RC:$src2, _.MemOp:$src3),
           OpcodeStr, "$src3, $src2", "$src2, $src3",
           (_.VT (OpNode (_.LdFrag addr:$src3), _.RC:$src1, _.RC:$src2)),
-          NoItinerary, 1, 0>, AVX512FMA3Base, Sched<[WriteFMA, ReadAfterLd]>;
+          NoItinerary, 1, 0>, AVX512FMA3Base, Sched<[WriteFMALd, ReadAfterLd]>;
 
   // Pattern is 312 order so that the load is in a different place from the
   // 213 and 231 patterns this helps tablegen's duplicate pattern detection.
@@ -6185,7 +6185,7 @@ multiclass avx512_fma3p_132_rm<bits<8> opc, string OpcodeStr, SDNode OpNode,
          "$src2, ${src3}"##_.BroadcastStr,
          (_.VT (OpNode (_.VT (X86VBroadcast(_.ScalarLdFrag addr:$src3))),
                        _.RC:$src1, _.RC:$src2)), NoItinerary, 1, 0>,
-         AVX512FMA3Base, EVEX_B, Sched<[WriteFMA, ReadAfterLd]>;
+         AVX512FMA3Base, EVEX_B, Sched<[WriteFMALd, ReadAfterLd]>;
   }
 }
 
@@ -6244,7 +6244,7 @@ let Constraints = "$src1 = $dst", hasSideEffects = 0 in {
   defm m_Int: AVX512_maskable_3src_scalar<opc, MRMSrcMem, _, (outs _.RC:$dst),
           (ins _.RC:$src2, _.IntScalarMemOp:$src3), OpcodeStr,
           "$src3, $src2", "$src2, $src3", RHS_VEC_m, NoItinerary, 1, 1>,
-          AVX512FMA3Base, Sched<[WriteFMA, ReadAfterLd]>;
+          AVX512FMA3Base, Sched<[WriteFMALd, ReadAfterLd]>;
 
   defm rb_Int: AVX512_maskable_3src_scalar<opc, MRMSrcReg, _, (outs _.RC:$dst),
          (ins _.RC:$src2, _.RC:$src3, AVX512RC:$rc),
@@ -6262,7 +6262,7 @@ let Constraints = "$src1 = $dst", hasSideEffects = 0 in {
                     (ins _.FRC:$src1, _.FRC:$src2, _.ScalarMemOp:$src3),
                     !strconcat(OpcodeStr,
                                "\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
-                    [RHS_m]>, Sched<[WriteFMA, ReadAfterLd]>;
+                    [RHS_m]>, Sched<[WriteFMALd, ReadAfterLd]>;
   }// isCodeGenOnly = 1
 }// Constraints = "$src1 = $dst"
 }
diff --git a/llvm/lib/Target/X86/X86InstrFMA.td b/llvm/lib/Target/X86/X86InstrFMA.td
index 1b706674a4d..35fa45590fc 100644
--- a/llvm/lib/Target/X86/X86InstrFMA.td
+++ b/llvm/lib/Target/X86/X86InstrFMA.td
@@ -51,7 +51,7 @@ multiclass fma3p_rm_213<bits<8> opc, string OpcodeStr, RegisterClass RC,
                               "\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
                    [(set RC:$dst, (VT (Op RC:$src2, RC:$src1,
                                           (MemFrag addr:$src3))))]>,
-                   Sched<[WriteFMA, ReadAfterLd]>;
+                   Sched<[WriteFMALd, ReadAfterLd]>;
 }
 
 multiclass fma3p_rm_231<bits<8> opc, string OpcodeStr, RegisterClass RC,
@@ -70,7 +70,7 @@ multiclass fma3p_rm_231<bits<8> opc, string OpcodeStr, RegisterClass RC,
                    !strconcat(OpcodeStr,
                               "\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
                    [(set RC:$dst, (VT (Op RC:$src2, (MemFrag addr:$src3),
-                                          RC:$src1)))]>, Sched<[WriteFMA, ReadAfterLd]>;
+                                          RC:$src1)))]>, Sched<[WriteFMALd, ReadAfterLd]>;
 }
 
 multiclass fma3p_rm_132<bits<8> opc, string OpcodeStr, RegisterClass RC,
@@ -91,7 +91,7 @@ multiclass fma3p_rm_132<bits<8> opc, string OpcodeStr, RegisterClass RC,
                    !strconcat(OpcodeStr,
                               "\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
                    [(set RC:$dst, (VT (Op (MemFrag addr:$src3), RC:$src1,
-                                          RC:$src2)))]>, Sched<[WriteFMA, ReadAfterLd]>;
+                                          RC:$src2)))]>, Sched<[WriteFMALd, ReadAfterLd]>;
 }
 
 let Constraints = "$src1 = $dst", hasSideEffects = 0, isCommutable = 1 in
@@ -184,7 +184,7 @@ multiclass fma3s_rm_213<bits<8> opc, string OpcodeStr,
                            "\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
                 [(set RC:$dst,
                   (OpNode RC:$src2, RC:$src1, (load addr:$src3)))]>,
-                Sched<[WriteFMA, ReadAfterLd]>;
+                Sched<[WriteFMALd, ReadAfterLd]>;
 }
 
 multiclass fma3s_rm_231<bits<8> opc, string OpcodeStr,
@@ -204,7 +204,7 @@ multiclass fma3s_rm_231<bits<8> opc, string OpcodeStr,
                            "\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
                 [(set RC:$dst,
                   (OpNode RC:$src2, (load addr:$src3), RC:$src1))]>,
-                Sched<[WriteFMA, ReadAfterLd]>;
+                Sched<[WriteFMALd, ReadAfterLd]>;
 }
 
 multiclass fma3s_rm_132<bits<8> opc, string OpcodeStr,
@@ -226,7 +226,7 @@ multiclass fma3s_rm_132<bits<8> opc, string OpcodeStr,
                            "\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
                 [(set RC:$dst,
                   (OpNode (load addr:$src3), RC:$src1, RC:$src2))]>,
-                Sched<[WriteFMA, ReadAfterLd]>;
+                Sched<[WriteFMALd, ReadAfterLd]>;
 }
 
 let Constraints = "$src1 = $dst", isCommutable = 1, hasSideEffects = 0 in
@@ -270,7 +270,7 @@ multiclass fma3s_rm_int<bits<8> opc, string OpcodeStr,
                         (ins RC:$src1, RC:$src2, memopr:$src3),
                         !strconcat(OpcodeStr,
                                    "\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
-                        []>, Sched<[WriteFMA, ReadAfterLd]>;
+                        []>, Sched<[WriteFMALd, ReadAfterLd]>;
 }
 
 // The FMA 213 form is created for lowering of scalar FMA intrinscis
@@ -374,14 +374,14 @@ multiclass fma4s<bits<8> opc, string OpcodeStr, RegisterClass RC,
            "\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"),
            [(set RC:$dst, (OpNode RC:$src1, RC:$src2,
                            (mem_frag addr:$src3)))]>, VEX_W, VEX_LIG,
-           Sched<[WriteFMA, ReadAfterLd]>;
+           Sched<[WriteFMALd, ReadAfterLd]>;
   def mr : FMA4S<opc, MRMSrcMem, (outs RC:$dst),
            (ins RC:$src1, x86memop:$src2, RC:$src3),
            !strconcat(OpcodeStr,
            "\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"),
            [(set RC:$dst,
              (OpNode RC:$src1, (mem_frag addr:$src2), RC:$src3))]>, VEX_LIG,
-           Sched<[WriteFMA, ReadAfterLd]>;
+           Sched<[WriteFMALd, ReadAfterLd]>;
 // For disassembler
 let isCodeGenOnly = 1, ForceDisassemble = 1, hasSideEffects = 0 in
   def rr_REV : FMA4S<opc, MRMSrcReg, (outs RC:$dst),
@@ -407,14 +407,14 @@ let isCodeGenOnly = 1 in {
                "\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"),
                [(set VR128:$dst, (VT (OpNode VR128:$src1, VR128:$src2,
                                   mem_cpat:$src3)))]>, VEX_W, VEX_LIG,
-               Sched<[WriteFMA, ReadAfterLd]>;
+               Sched<[WriteFMALd, ReadAfterLd]>;
   def mr_Int : FMA4S_Int<opc, MRMSrcMem, (outs VR128:$dst),
                (ins VR128:$src1, memop:$src2, VR128:$src3),
                !strconcat(OpcodeStr,
                "\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"),
                [(set VR128:$dst,
                  (VT (OpNode VR128:$src1, mem_cpat:$src2, VR128:$src3)))]>,
-               VEX_LIG, Sched<[WriteFMA, ReadAfterLd]>;
+               VEX_LIG, Sched<[WriteFMALd, ReadAfterLd]>;
 let hasSideEffects = 0 in
   def rr_Int_REV : FMA4S_Int<opc, MRMSrcReg, (outs VR128:$dst),
                (ins VR128:$src1, VR128:$src2, VR128:$src3),
@@ -441,14 +441,14 @@ multiclass fma4p<bits<8> opc, string OpcodeStr, SDNode OpNode,
            "\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"),
            [(set VR128:$dst, (OpNode VR128:$src1, VR128:$src2,
                               (ld_frag128 addr:$src3)))]>, VEX_W,
-           Sched<[WriteFMA, ReadAfterLd]>;
+           Sched<[WriteFMALd, ReadAfterLd]>;
   def mr : FMA4<opc, MRMSrcMem, (outs VR128:$dst),
            (ins VR128:$src1, f128mem:$src2, VR128:$src3),
            !strconcat(OpcodeStr,
            "\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"),
            [(set VR128:$dst,
              (OpNode VR128:$src1, (ld_frag128 addr:$src2), VR128:$src3))]>,
-           Sched<[WriteFMA, ReadAfterLd]>;
+           Sched<[WriteFMALd, ReadAfterLd]>;
   let isCommutable = 1 in
   def Yrr : FMA4<opc, MRMSrcRegOp4, (outs VR256:$dst),
            (ins VR256:$src1, VR256:$src2, VR256:$src3),
@@ -463,14 +463,14 @@ multiclass fma4p<bits<8> opc, string OpcodeStr, SDNode OpNode,
            "\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"),
            [(set VR256:$dst, (OpNode VR256:$src1, VR256:$src2,
                               (ld_frag256 addr:$src3)))]>, VEX_W, VEX_L,
-           Sched<[WriteFMA, ReadAfterLd]>;
+           Sched<[WriteFMALd, ReadAfterLd]>;
   def Ymr : FMA4<opc, MRMSrcMem, (outs VR256:$dst),
            (ins VR256:$src1, f256mem:$src2, VR256:$src3),
            !strconcat(OpcodeStr,
            "\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"),
            [(set VR256:$dst, (OpNode VR256:$src1,
                               (ld_frag256 addr:$src2), VR256:$src3))]>, VEX_L,
-           Sched<[WriteFMA, ReadAfterLd]>;
+           Sched<[WriteFMALd, ReadAfterLd]>;
 // For disassembler
 let isCodeGenOnly = 1, ForceDisassemble = 1, hasSideEffects = 0 in {
   def rr_REV : FMA4<opc, MRMSrcReg, (outs VR128:$dst),
diff --git a/llvm/lib/Target/X86/X86InstrXOP.td b/llvm/lib/Target/X86/X86InstrXOP.td
index 383ffbffb39..c4b8e3e90d2 100644
--- a/llvm/lib/Target/X86/X86InstrXOP.td
+++ b/llvm/lib/Target/X86/X86InstrXOP.td
@@ -18,7 +18,7 @@ multiclass xop2op<bits<8> opc, string OpcodeStr, Intrinsic Int, PatFrag memop> {
   def rm : IXOP<opc, MRMSrcMem, (outs VR128:$dst), (ins i128mem:$src),
            !strconcat(OpcodeStr, "\t{$src, $dst|$dst, $src}"),
            [(set VR128:$dst, (Int (bitconvert (memop addr:$src))))]>, XOP,
-           Sched<[WritePHAdd, ReadAfterLd]>;
+           Sched<[WritePHAddLd, ReadAfterLd]>;
 }
 
 let ExeDomain = SSEPackedInt in {
@@ -48,7 +48,7 @@ multiclass xop2opsld<bits<8> opc, string OpcodeStr, Intrinsic Int,
   def rm : IXOP<opc, MRMSrcMem, (outs VR128:$dst), (ins memop:$src),
            !strconcat(OpcodeStr, "\t{$src, $dst|$dst, $src}"),
            [(set VR128:$dst, (Int (bitconvert mem_cpat:$src)))]>, XOP,
-           Sched<[WriteFAdd, ReadAfterLd]>;
+           Sched<[WriteFAddLd, ReadAfterLd]>;
 }
 
 multiclass xop2op128<bits<8> opc, string OpcodeStr, Intrinsic Int,
@@ -59,7 +59,7 @@ multiclass xop2op128<bits<8> opc, string OpcodeStr, Intrinsic Int,
   def rm : IXOP<opc, MRMSrcMem, (outs VR128:$dst), (ins f128mem:$src),
            !strconcat(OpcodeStr, "\t{$src, $dst|$dst, $src}"),
            [(set VR128:$dst, (Int (bitconvert (memop addr:$src))))]>, XOP,
-           Sched<[WriteFAdd, ReadAfterLd]>;
+           Sched<[WriteFAddLd, ReadAfterLd]>;
 }
 
 multiclass xop2op256<bits<8> opc, string OpcodeStr, Intrinsic Int,
@@ -70,7 +70,7 @@ multiclass xop2op256<bits<8> opc, string OpcodeStr, Intrinsic Int,
   def rmY : IXOP<opc, MRMSrcMem, (outs VR256:$dst), (ins f256mem:$src),
            !strconcat(OpcodeStr, "\t{$src, $dst|$dst, $src}"),
            [(set VR256:$dst, (Int (bitconvert (memop addr:$src))))]>, XOP, VEX_L,
-           Sched<[WriteFAdd, ReadAfterLd]>;
+           Sched<[WriteFAddLd, ReadAfterLd]>;
 }
 
 let ExeDomain = SSEPackedSingle in {
@@ -101,14 +101,14 @@ multiclass xop3op<bits<8> opc, string OpcodeStr, SDNode OpNode,
            [(set VR128:$dst,
               (vt128 (OpNode (vt128 VR128:$src1),
                              (vt128 (bitconvert (loadv2i64 addr:$src2))))))]>,
-           XOP_4V, VEX_W, Sched<[WriteVarVecShift, ReadAfterLd]>;
+           XOP_4V, VEX_W, Sched<[WriteVarVecShiftLd, ReadAfterLd]>;
   def mr : IXOP<opc, MRMSrcMem4VOp3, (outs VR128:$dst),
            (ins i128mem:$src1, VR128:$src2),
            !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
            [(set VR128:$dst,
               (vt128 (OpNode (vt128 (bitconvert (loadv2i64 addr:$src1))),
                              (vt128 VR128:$src2))))]>,
-             XOP, Sched<[WriteVarVecShift, ReadAfterLd]>;
+             XOP, Sched<[WriteVarVecShiftLd, ReadAfterLd]>;
   // For disassembler
   let isCodeGenOnly = 1, ForceDisassemble = 1, hasSideEffects = 0 in
   def rr_REV : IXOP<opc, MRMSrcReg, (outs VR128:$dst),
@@ -146,7 +146,7 @@ multiclass xop3opimm<bits<8> opc, string OpcodeStr, SDNode OpNode,
            !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
            [(set VR128:$dst,
               (vt128 (OpNode (vt128 (bitconvert (loadv2i64 addr:$src1))), imm:$src2)))]>,
-           XOP, Sched<[WriteVecShift, ReadAfterLd]>;
+           XOP, Sched<[WriteVecShiftLd, ReadAfterLd]>;
 }
 
 let ExeDomain = SSEPackedInt in {
@@ -172,7 +172,7 @@ multiclass xop4opm2<bits<8> opc, string OpcodeStr, Intrinsic Int> {
            "\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"),
            [(set VR128:$dst,
               (Int VR128:$src1, (bitconvert (loadv2i64 addr:$src2)),
-              VR128:$src3))]>, XOP_4V, Sched<[WriteVecIMul, ReadAfterLd]>;
+              VR128:$src3))]>, XOP_4V, Sched<[WriteVecIMulLd, ReadAfterLd]>;
 }
 
 let ExeDomain = SSEPackedInt in {
@@ -221,7 +221,7 @@ multiclass xopvpcom<bits<8> opc, string Suffix, SDNode OpNode, ValueType vt128>
            [(set VR128:$dst,
               (vt128 (OpNode (vt128 VR128:$src1), (vt128 VR128:$src2),
                              imm:$cc)))]>,
-           XOP_4V, Sched<[WriteVecALU, ReadAfterLd]>;
+           XOP_4V, Sched<[WriteVecALULd, ReadAfterLd]>;
   def mi : IXOPi8<opc, MRMSrcMem, (outs VR128:$dst),
            (ins VR128:$src1, i128mem:$src2, XOPCC:$cc),
            !strconcat("vpcom${cc}", Suffix,
@@ -230,19 +230,19 @@ multiclass xopvpcom<bits<8> opc, string Suffix, SDNode OpNode, ValueType vt128>
               (vt128 (OpNode (vt128 VR128:$src1),
                              (vt128 (bitconvert (loadv2i64 addr:$src2))),
                               imm:$cc)))]>,
-           XOP_4V, Sched<[WriteVecALU, ReadAfterLd]>;
+           XOP_4V, Sched<[WriteVecALULd, ReadAfterLd]>;
   let isAsmParserOnly = 1, hasSideEffects = 0 in {
     def ri_alt : IXOPi8<opc, MRMSrcReg, (outs VR128:$dst),
                  (ins VR128:$src1, VR128:$src2, u8imm:$src3),
                  !strconcat("vpcom", Suffix,
                  "\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"),
-                 []>, XOP_4V, Sched<[WriteVecALU, ReadAfterLd]>;
+                 []>, XOP_4V, Sched<[WriteVecALULd, ReadAfterLd]>;
     let mayLoad = 1 in
     def mi_alt : IXOPi8<opc, MRMSrcMem, (outs VR128:$dst),
                  (ins VR128:$src1, i128mem:$src2, u8imm:$src3),
                  !strconcat("vpcom", Suffix,
                  "\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"),
-                 []>, XOP_4V, Sched<[WriteVecALU, ReadAfterLd]>;
+                 []>, XOP_4V, Sched<[WriteVecALULd, ReadAfterLd]>;
   }
 }
 
@@ -274,7 +274,7 @@ multiclass xop4op<bits<8> opc, string OpcodeStr, SDNode OpNode,
             [(set VR128:$dst,
               (vt128 (OpNode (vt128 VR128:$src1), (vt128 VR128:$src2),
                              (vt128 (bitconvert (loadv2i64 addr:$src3))))))]>,
-            XOP_4V, VEX_W, Sched<[WriteShuffle, ReadAfterLd]>;
+            XOP_4V, VEX_W, Sched<[WriteShuffleLd, ReadAfterLd]>;
   def rmr : IXOPi8Reg<opc, MRMSrcMem, (outs VR128:$dst),
             (ins VR128:$src1, i128mem:$src2, VR128:$src3),
             !strconcat(OpcodeStr,
@@ -282,7 +282,7 @@ multiclass xop4op<bits<8> opc, string OpcodeStr, SDNode OpNode,
             [(set VR128:$dst,
               (v16i8 (OpNode (vt128 VR128:$src1), (vt128 (bitconvert (loadv2i64 addr:$src2))),
                              (vt128 VR128:$src3))))]>,
-            XOP_4V, Sched<[WriteShuffle, ReadAfterLd]>;
+            XOP_4V, Sched<[WriteShuffleLd, ReadAfterLd]>;
   // For disassembler
   let isCodeGenOnly = 1, ForceDisassemble = 1, hasSideEffects = 0 in
   def rrr_REV : IXOPi8Reg<opc, MRMSrcRegOp4, (outs VR128:$dst),
@@ -312,14 +312,14 @@ multiclass xop4op_int<bits<8> opc, string OpcodeStr, RegisterClass RC,
             "\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"),
             [(set RC:$dst, (VT (or (and (load addr:$src3), RC:$src1),
                                    (X86andnp (load addr:$src3), RC:$src2))))]>,
-            XOP_4V, VEX_W, Sched<[WriteShuffle, ReadAfterLd]>;
+            XOP_4V, VEX_W, Sched<[WriteShuffleLd, ReadAfterLd]>;
   def rmr : IXOPi8Reg<opc, MRMSrcMem, (outs RC:$dst),
             (ins RC:$src1, x86memop:$src2, RC:$src3),
             !strconcat(OpcodeStr,
             "\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"),
             [(set RC:$dst, (VT (or (and RC:$src3, RC:$src1),
                                    (X86andnp RC:$src3, (load addr:$src2)))))]>,
-            XOP_4V, Sched<[WriteShuffle, ReadAfterLd]>;
+            XOP_4V, Sched<[WriteShuffleLd, ReadAfterLd]>;
   // For disassembler
   let isCodeGenOnly = 1, ForceDisassemble = 1, hasSideEffects = 0 in
   def rrr_REV : IXOPi8Reg<opc, MRMSrcRegOp4, (outs RC:$dst),
@@ -353,7 +353,7 @@ multiclass xop_vpermil2<bits<8> Opc, string OpcodeStr, RegisterClass RC,
           (VT (X86vpermil2 RC:$src1, RC:$src2,
                            (bitconvert (IntLdFrag addr:$src3)),
                            (i8 imm:$src4))))]>, VEX_W,
-        Sched<[WriteFShuffle, ReadAfterLd]>;
+        Sched<[WriteFShuffleLd, ReadAfterLd]>;
   def mr : IXOP5<Opc, MRMSrcMem, (outs RC:$dst),
         (ins RC:$src1, fpmemop:$src2, RC:$src3, u8imm:$src4),
         !strconcat(OpcodeStr,
@@ -361,7 +361,7 @@ multiclass xop_vpermil2<bits<8> Opc, string OpcodeStr, RegisterClass RC,
         [(set RC:$dst,
           (VT (X86vpermil2 RC:$src1, (FPLdFrag addr:$src2),
                            RC:$src3, (i8 imm:$src4))))]>,
-        Sched<[WriteFShuffle, ReadAfterLd]>;
+        Sched<[WriteFShuffleLd, ReadAfterLd]>;
   // For disassembler
   let isCodeGenOnly = 1, ForceDisassemble = 1, hasSideEffects = 0 in
   def rr_REV : IXOP5<Opc, MRMSrcRegOp4, (outs RC:$dst),
diff --git a/llvm/test/CodeGen/X86/fma-schedule.ll b/llvm/test/CodeGen/X86/fma-schedule.ll
index 3f540ff1041..f4db8bbd9bc 100644
--- a/llvm/test/CodeGen/X86/fma-schedule.ll
+++ b/llvm/test/CodeGen/X86/fma-schedule.ll
@@ -19,7 +19,7 @@ define <2 x double> @test_vfmadd213pd(<2 x double> %a0, <2 x double> %a1, <2 x d
 ; GENERIC-LABEL: test_vfmadd213pd:
 ; GENERIC:       # %bb.0:
 ; GENERIC-NEXT:    vfmadd213pd %xmm2, %xmm1, %xmm0 # sched: [5:0.50]
-; GENERIC-NEXT:    vfmadd213pd (%rdi), %xmm1, %xmm0 # sched: [5:0.50]
+; GENERIC-NEXT:    vfmadd213pd (%rdi), %xmm1, %xmm0 # sched: [9:0.50]
 ; GENERIC-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: test_vfmadd213pd:
@@ -67,7 +67,7 @@ define <4 x double> @test_vfmadd213pd_ymm(<4 x double> %a0, <4 x double> %a1, <4
 ; GENERIC-LABEL: test_vfmadd213pd_ymm:
 ; GENERIC:       # %bb.0:
 ; GENERIC-NEXT:    vfmadd213pd %ymm2, %ymm1, %ymm0 # sched: [5:0.50]
-; GENERIC-NEXT:    vfmadd213pd (%rdi), %ymm1, %ymm0 # sched: [5:0.50]
+; GENERIC-NEXT:    vfmadd213pd (%rdi), %ymm1, %ymm0 # sched: [9:0.50]
 ; GENERIC-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: test_vfmadd213pd_ymm:
@@ -115,7 +115,7 @@ define <4 x float> @test_vfmadd213ps(<4 x float> %a0, <4 x float> %a1, <4 x floa
 ; GENERIC-LABEL: test_vfmadd213ps:
 ; GENERIC:       # %bb.0:
 ; GENERIC-NEXT:    vfmadd213ps %xmm2, %xmm1, %xmm0 # sched: [5:0.50]
-; GENERIC-NEXT:    vfmadd213ps (%rdi), %xmm1, %xmm0 # sched: [5:0.50]
+; GENERIC-NEXT:    vfmadd213ps (%rdi), %xmm1, %xmm0 # sched: [9:0.50]
 ; GENERIC-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: test_vfmadd213ps:
@@ -163,7 +163,7 @@ define <8 x float> @test_vfmadd213ps_ymm(<8 x float> %a0, <8 x float> %a1, <8 x
 ; GENERIC-LABEL: test_vfmadd213ps_ymm:
 ; GENERIC:       # %bb.0:
 ; GENERIC-NEXT:    vfmadd213ps %ymm2, %ymm1, %ymm0 # sched: [5:0.50]
-; GENERIC-NEXT:    vfmadd213ps (%rdi), %ymm1, %ymm0 # sched: [5:0.50]
+; GENERIC-NEXT:    vfmadd213ps (%rdi), %ymm1, %ymm0 # sched: [9:0.50]
 ; GENERIC-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: test_vfmadd213ps_ymm:
@@ -211,7 +211,7 @@ define <2 x double> @test_vfmadd213sd(<2 x double> %a0, <2 x double> %a1, <2 x d
 ; GENERIC-LABEL: test_vfmadd213sd:
 ; GENERIC:       # %bb.0:
 ; GENERIC-NEXT:    vfmadd213sd %xmm2, %xmm1, %xmm0 # sched: [5:0.50]
-; GENERIC-NEXT:    vfmadd213sd (%rdi), %xmm1, %xmm0 # sched: [5:0.50]
+; GENERIC-NEXT:    vfmadd213sd (%rdi), %xmm1, %xmm0 # sched: [9:0.50]
 ; GENERIC-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: test_vfmadd213sd:
@@ -259,7 +259,7 @@ define <4 x float> @test_vfmadd213ss(<4 x float> %a0, <4 x float> %a1, <4 x floa
 ; GENERIC-LABEL: test_vfmadd213ss:
 ; GENERIC:       # %bb.0:
 ; GENERIC-NEXT:    vfmadd213ss %xmm2, %xmm1, %xmm0 # sched: [5:0.50]
-; GENERIC-NEXT:    vfmadd213ss (%rdi), %xmm1, %xmm0 # sched: [5:0.50]
+; GENERIC-NEXT:    vfmadd213ss (%rdi), %xmm1, %xmm0 # sched: [9:0.50]
 ; GENERIC-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: test_vfmadd213ss:
@@ -319,7 +319,7 @@ define <2 x double> @test_vfmaddsubpd(<2 x double> %a0, <2 x double> %a1, <2 x d
 ; GENERIC-LABEL: test_vfmaddsubpd:
 ; GENERIC:       # %bb.0:
 ; GENERIC-NEXT:    vfmaddsub213pd %xmm2, %xmm1, %xmm0 # sched: [5:0.50]
-; GENERIC-NEXT:    vfmaddsub213pd (%rdi), %xmm1, %xmm0 # sched: [5:0.50]
+; GENERIC-NEXT:    vfmaddsub213pd (%rdi), %xmm1, %xmm0 # sched: [9:0.50]
 ; GENERIC-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: test_vfmaddsubpd:
@@ -367,7 +367,7 @@ define <4 x double> @test_vfmaddsubpd_ymm(<4 x double> %a0, <4 x double> %a1, <4
 ; GENERIC-LABEL: test_vfmaddsubpd_ymm:
 ; GENERIC:       # %bb.0:
 ; GENERIC-NEXT:    vfmaddsub213pd %ymm2, %ymm1, %ymm0 # sched: [5:0.50]
-; GENERIC-NEXT:    vfmaddsub213pd (%rdi), %ymm1, %ymm0 # sched: [5:0.50]
+; GENERIC-NEXT:    vfmaddsub213pd (%rdi), %ymm1, %ymm0 # sched: [9:0.50]
 ; GENERIC-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: test_vfmaddsubpd_ymm:
@@ -415,7 +415,7 @@ define <4 x float> @test_vfmaddsubps(<4 x float> %a0, <4 x float> %a1, <4 x floa
 ; GENERIC-LABEL: test_vfmaddsubps:
 ; GENERIC:       # %bb.0:
 ; GENERIC-NEXT:    vfmaddsub213ps %xmm2, %xmm1, %xmm0 # sched: [5:0.50]
-; GENERIC-NEXT:    vfmaddsub213ps (%rdi), %xmm1, %xmm0 # sched: [5:0.50]
+; GENERIC-NEXT:    vfmaddsub213ps (%rdi), %xmm1, %xmm0 # sched: [9:0.50]
 ; GENERIC-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: test_vfmaddsubps:
@@ -463,7 +463,7 @@ define <8 x float> @test_vfmaddsubps_ymm(<8 x float> %a0, <8 x float> %a1, <8 x
 ; GENERIC-LABEL: test_vfmaddsubps_ymm:
 ; GENERIC:       # %bb.0:
 ; GENERIC-NEXT:    vfmaddsub213ps %ymm2, %ymm1, %ymm0 # sched: [5:0.50]
-; GENERIC-NEXT:    vfmaddsub213ps (%rdi), %ymm1, %ymm0 # sched: [5:0.50]
+; GENERIC-NEXT:    vfmaddsub213ps (%rdi), %ymm1, %ymm0 # sched: [9:0.50]
 ; GENERIC-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: test_vfmaddsubps_ymm:
@@ -523,7 +523,7 @@ define <2 x double> @test_vfmsubaddpd(<2 x double> %a0, <2 x double> %a1, <2 x d
 ; GENERIC-LABEL: test_vfmsubaddpd:
 ; GENERIC:       # %bb.0:
 ; GENERIC-NEXT:    vfmsubadd213pd %xmm2, %xmm1, %xmm0 # sched: [5:0.50]
-; GENERIC-NEXT:    vfmsubadd213pd (%rdi), %xmm1, %xmm0 # sched: [5:0.50]
+; GENERIC-NEXT:    vfmsubadd213pd (%rdi), %xmm1, %xmm0 # sched: [9:0.50]
 ; GENERIC-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: test_vfmsubaddpd:
@@ -571,7 +571,7 @@ define <4 x double> @test_vfmsubaddpd_ymm(<4 x double> %a0, <4 x double> %a1, <4
 ; GENERIC-LABEL: test_vfmsubaddpd_ymm:
 ; GENERIC:       # %bb.0:
 ; GENERIC-NEXT:    vfmsubadd213pd %ymm2, %ymm1, %ymm0 # sched: [5:0.50]
-; GENERIC-NEXT:    vfmsubadd213pd (%rdi), %ymm1, %ymm0 # sched: [5:0.50]
+; GENERIC-NEXT:    vfmsubadd213pd (%rdi), %ymm1, %ymm0 # sched: [9:0.50]
 ; GENERIC-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: test_vfmsubaddpd_ymm:
@@ -619,7 +619,7 @@ define <4 x float> @test_vfmsubaddps(<4 x float> %a0, <4 x float> %a1, <4 x floa
 ; GENERIC-LABEL: test_vfmsubaddps:
 ; GENERIC:       # %bb.0:
 ; GENERIC-NEXT:    vfmsubadd213ps %xmm2, %xmm1, %xmm0 # sched: [5:0.50]
-; GENERIC-NEXT:    vfmsubadd213ps (%rdi), %xmm1, %xmm0 # sched: [5:0.50]
+; GENERIC-NEXT:    vfmsubadd213ps (%rdi), %xmm1, %xmm0 # sched: [9:0.50]
 ; GENERIC-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: test_vfmsubaddps:
@@ -667,7 +667,7 @@ define <8 x float> @test_vfmsubaddps_ymm(<8 x float> %a0, <8 x float> %a1, <8 x
 ; GENERIC-LABEL: test_vfmsubaddps_ymm:
 ; GENERIC:       # %bb.0:
 ; GENERIC-NEXT:    vfmsubadd213ps %ymm2, %ymm1, %ymm0 # sched: [5:0.50]
-; GENERIC-NEXT:    vfmsubadd213ps (%rdi), %ymm1, %ymm0 # sched: [5:0.50]
+; GENERIC-NEXT:    vfmsubadd213ps (%rdi), %ymm1, %ymm0 # sched: [9:0.50]
 ; GENERIC-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: test_vfmsubaddps_ymm:
@@ -727,7 +727,7 @@ define <2 x double> @test_vfmsub213pd(<2 x double> %a0, <2 x double> %a1, <2 x d
 ; GENERIC-LABEL: test_vfmsub213pd:
 ; GENERIC:       # %bb.0:
 ; GENERIC-NEXT:    vfmsub213pd %xmm2, %xmm1, %xmm0 # sched: [5:0.50]
-; GENERIC-NEXT:    vfmsub213pd (%rdi), %xmm1, %xmm0 # sched: [5:0.50]
+; GENERIC-NEXT:    vfmsub213pd (%rdi), %xmm1, %xmm0 # sched: [9:0.50]
 ; GENERIC-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: test_vfmsub213pd:
@@ -775,7 +775,7 @@ define <4 x double> @test_vfmsub213pd_ymm(<4 x double> %a0, <4 x double> %a1, <4
 ; GENERIC-LABEL: test_vfmsub213pd_ymm:
 ; GENERIC:       # %bb.0:
 ; GENERIC-NEXT:    vfmsub213pd %ymm2, %ymm1, %ymm0 # sched: [5:0.50]
-; GENERIC-NEXT:    vfmsub213pd (%rdi), %ymm1, %ymm0 # sched: [5:0.50]
+; GENERIC-NEXT:    vfmsub213pd (%rdi), %ymm1, %ymm0 # sched: [9:0.50]
 ; GENERIC-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: test_vfmsub213pd_ymm:
@@ -823,7 +823,7 @@ define <4 x float> @test_vfmsub213ps(<4 x float> %a0, <4 x float> %a1, <4 x floa
 ; GENERIC-LABEL: test_vfmsub213ps:
 ; GENERIC:       # %bb.0:
 ; GENERIC-NEXT:    vfmsub213ps %xmm2, %xmm1, %xmm0 # sched: [5:0.50]
-; GENERIC-NEXT:    vfmsub213ps (%rdi), %xmm1, %xmm0 # sched: [5:0.50]
+; GENERIC-NEXT:    vfmsub213ps (%rdi), %xmm1, %xmm0 # sched: [9:0.50]
 ; GENERIC-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: test_vfmsub213ps:
@@ -871,7 +871,7 @@ define <8 x float> @test_vfmsub213ps_ymm(<8 x float> %a0, <8 x float> %a1, <8 x
 ; GENERIC-LABEL: test_vfmsub213ps_ymm:
 ; GENERIC:       # %bb.0:
 ; GENERIC-NEXT:    vfmsub213ps %ymm2, %ymm1, %ymm0 # sched: [5:0.50]
-; GENERIC-NEXT:    vfmsub213ps (%rdi), %ymm1, %ymm0 # sched: [5:0.50]
+; GENERIC-NEXT:    vfmsub213ps (%rdi), %ymm1, %ymm0 # sched: [9:0.50]
 ; GENERIC-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: test_vfmsub213ps_ymm:
@@ -919,7 +919,7 @@ define <2 x double> @test_vfmsub213sd(<2 x double> %a0, <2 x double> %a1, <2 x d
 ; GENERIC-LABEL: test_vfmsub213sd:
 ; GENERIC:       # %bb.0:
 ; GENERIC-NEXT:    vfmsub213sd %xmm2, %xmm1, %xmm0 # sched: [5:0.50]
-; GENERIC-NEXT:    vfmsub213sd (%rdi), %xmm1, %xmm0 # sched: [5:0.50]
+; GENERIC-NEXT:    vfmsub213sd (%rdi), %xmm1, %xmm0 # sched: [9:0.50]
 ; GENERIC-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: test_vfmsub213sd:
@@ -967,7 +967,7 @@ define <4 x float> @test_vfmsub213ss(<4 x float> %a0, <4 x float> %a1, <4 x floa
 ; GENERIC-LABEL: test_vfmsub213ss:
 ; GENERIC:       # %bb.0:
 ; GENERIC-NEXT:    vfmsub213ss %xmm2, %xmm1, %xmm0 # sched: [5:0.50]
-; GENERIC-NEXT:    vfmsub213ss (%rdi), %xmm1, %xmm0 # sched: [5:0.50]
+; GENERIC-NEXT:    vfmsub213ss (%rdi), %xmm1, %xmm0 # sched: [9:0.50]
 ; GENERIC-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: test_vfmsub213ss:
@@ -1027,7 +1027,7 @@ define <2 x double> @test_vfnmadd213pd(<2 x double> %a0, <2 x double> %a1, <2 x
 ; GENERIC-LABEL: test_vfnmadd213pd:
 ; GENERIC:       # %bb.0:
 ; GENERIC-NEXT:    vfnmadd213pd %xmm2, %xmm1, %xmm0 # sched: [5:0.50]
-; GENERIC-NEXT:    vfnmadd213pd (%rdi), %xmm1, %xmm0 # sched: [5:0.50]
+; GENERIC-NEXT:    vfnmadd213pd (%rdi), %xmm1, %xmm0 # sched: [9:0.50]
 ; GENERIC-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: test_vfnmadd213pd:
@@ -1075,7 +1075,7 @@ define <4 x double> @test_vfnmadd213pd_ymm(<4 x double> %a0, <4 x double> %a1, <
 ; GENERIC-LABEL: test_vfnmadd213pd_ymm:
 ; GENERIC:       # %bb.0:
 ; GENERIC-NEXT:    vfnmadd213pd %ymm2, %ymm1, %ymm0 # sched: [5:0.50]
-; GENERIC-NEXT:    vfnmadd213pd (%rdi), %ymm1, %ymm0 # sched: [5:0.50]
+; GENERIC-NEXT:    vfnmadd213pd (%rdi), %ymm1, %ymm0 # sched: [9:0.50]
 ; GENERIC-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: test_vfnmadd213pd_ymm:
@@ -1123,7 +1123,7 @@ define <4 x float> @test_vfnmadd213ps(<4 x float> %a0, <4 x float> %a1, <4 x flo
 ; GENERIC-LABEL: test_vfnmadd213ps:
 ; GENERIC:       # %bb.0:
 ; GENERIC-NEXT:    vfnmadd213ps %xmm2, %xmm1, %xmm0 # sched: [5:0.50]
-; GENERIC-NEXT:    vfnmadd213ps (%rdi), %xmm1, %xmm0 # sched: [5:0.50]
+; GENERIC-NEXT:    vfnmadd213ps (%rdi), %xmm1, %xmm0 # sched: [9:0.50]
 ; GENERIC-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: test_vfnmadd213ps:
@@ -1171,7 +1171,7 @@ define <8 x float> @test_vfnmadd213ps_ymm(<8 x float> %a0, <8 x float> %a1, <8 x
 ; GENERIC-LABEL: test_vfnmadd213ps_ymm:
 ; GENERIC:       # %bb.0:
 ; GENERIC-NEXT:    vfnmadd213ps %ymm2, %ymm1, %ymm0 # sched: [5:0.50]
-; GENERIC-NEXT:    vfnmadd213ps (%rdi), %ymm1, %ymm0 # sched: [5:0.50]
+; GENERIC-NEXT:    vfnmadd213ps (%rdi), %ymm1, %ymm0 # sched: [9:0.50]
 ; GENERIC-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: test_vfnmadd213ps_ymm:
@@ -1219,7 +1219,7 @@ define <2 x double> @test_vfnmadd213sd(<2 x double> %a0, <2 x double> %a1, <2 x
 ; GENERIC-LABEL: test_vfnmadd213sd:
 ; GENERIC:       # %bb.0:
 ; GENERIC-NEXT:    vfnmadd213sd %xmm2, %xmm1, %xmm0 # sched: [5:0.50]
-; GENERIC-NEXT:    vfnmadd213sd (%rdi), %xmm1, %xmm0 # sched: [5:0.50]
+; GENERIC-NEXT:    vfnmadd213sd (%rdi), %xmm1, %xmm0 # sched: [9:0.50]
 ; GENERIC-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: test_vfnmadd213sd:
@@ -1267,7 +1267,7 @@ define <4 x float> @test_vfnmadd213ss(<4 x float> %a0, <4 x float> %a1, <4 x flo
 ; GENERIC-LABEL: test_vfnmadd213ss:
 ; GENERIC:       # %bb.0:
 ; GENERIC-NEXT:    vfnmadd213ss %xmm2, %xmm1, %xmm0 # sched: [5:0.50]
-; GENERIC-NEXT:    vfnmadd213ss (%rdi), %xmm1, %xmm0 # sched: [5:0.50]
+; GENERIC-NEXT:    vfnmadd213ss (%rdi), %xmm1, %xmm0 # sched: [9:0.50]
 ; GENERIC-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: test_vfnmadd213ss:
@@ -1327,7 +1327,7 @@ define <2 x double> @test_vfnmsub213pd(<2 x double> %a0, <2 x double> %a1, <2 x
 ; GENERIC-LABEL: test_vfnmsub213pd:
 ; GENERIC:       # %bb.0:
 ; GENERIC-NEXT:    vfnmsub213pd %xmm2, %xmm1, %xmm0 # sched: [5:0.50]
-; GENERIC-NEXT:    vfnmsub213pd (%rdi), %xmm1, %xmm0 # sched: [5:0.50]
+; GENERIC-NEXT:    vfnmsub213pd (%rdi), %xmm1, %xmm0 # sched: [9:0.50]
 ; GENERIC-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: test_vfnmsub213pd:
@@ -1375,7 +1375,7 @@ define <4 x double> @test_vfnmsub213pd_ymm(<4 x double> %a0, <4 x double> %a1, <
 ; GENERIC-LABEL: test_vfnmsub213pd_ymm:
 ; GENERIC:       # %bb.0:
 ; GENERIC-NEXT:    vfnmsub213pd %ymm2, %ymm1, %ymm0 # sched: [5:0.50]
-; GENERIC-NEXT:    vfnmsub213pd (%rdi), %ymm1, %ymm0 # sched: [5:0.50]
+; GENERIC-NEXT:    vfnmsub213pd (%rdi), %ymm1, %ymm0 # sched: [9:0.50]
 ; GENERIC-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: test_vfnmsub213pd_ymm:
@@ -1423,7 +1423,7 @@ define <4 x float> @test_vfnmsub213ps(<4 x float> %a0, <4 x float> %a1, <4 x flo
 ; GENERIC-LABEL: test_vfnmsub213ps:
 ; GENERIC:       # %bb.0:
 ; GENERIC-NEXT:    vfnmsub213ps %xmm2, %xmm1, %xmm0 # sched: [5:0.50]
-; GENERIC-NEXT:    vfnmsub213ps (%rdi), %xmm1, %xmm0 # sched: [5:0.50]
+; GENERIC-NEXT:    vfnmsub213ps (%rdi), %xmm1, %xmm0 # sched: [9:0.50]
 ; GENERIC-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: test_vfnmsub213ps:
@@ -1471,7 +1471,7 @@ define <8 x float> @test_vfnmsub213ps_ymm(<8 x float> %a0, <8 x float> %a1, <8 x
 ; GENERIC-LABEL: test_vfnmsub213ps_ymm:
 ; GENERIC:       # %bb.0:
 ; GENERIC-NEXT:    vfnmsub213ps %ymm2, %ymm1, %ymm0 # sched: [5:0.50]
-; GENERIC-NEXT:    vfnmsub213ps (%rdi), %ymm1, %ymm0 # sched: [5:0.50]
+; GENERIC-NEXT:    vfnmsub213ps (%rdi), %ymm1, %ymm0 # sched: [9:0.50]
 ; GENERIC-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: test_vfnmsub213ps_ymm:
@@ -1519,7 +1519,7 @@ define <2 x double> @test_vfnmsub213sd(<2 x double> %a0, <2 x double> %a1, <2 x
 ; GENERIC-LABEL: test_vfnmsub213sd:
 ; GENERIC:       # %bb.0:
 ; GENERIC-NEXT:    vfnmsub213sd %xmm2, %xmm1, %xmm0 # sched: [5:0.50]
-; GENERIC-NEXT:    vfnmsub213sd (%rdi), %xmm1, %xmm0 # sched: [5:0.50]
+; GENERIC-NEXT:    vfnmsub213sd (%rdi), %xmm1, %xmm0 # sched: [9:0.50]
 ; GENERIC-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: test_vfnmsub213sd:
@@ -1567,7 +1567,7 @@ define <4 x float> @test_vfnmsub213ss(<4 x float> %a0, <4 x float> %a1, <4 x flo
 ; GENERIC-LABEL: test_vfnmsub213ss:
 ; GENERIC:       # %bb.0:
 ; GENERIC-NEXT:    vfnmsub213ss %xmm2, %xmm1, %xmm0 # sched: [5:0.50]
-; GENERIC-NEXT:    vfnmsub213ss (%rdi), %xmm1, %xmm0 # sched: [5:0.50]
+; GENERIC-NEXT:    vfnmsub213ss (%rdi), %xmm1, %xmm0 # sched: [9:0.50]
 ; GENERIC-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: test_vfnmsub213ss: