[X86] Extract latency of fldz/fld1 in separate classes.

[courbet]

Summary

• I've measured the values for Broadwell, Haswell, SandyBridge, Skylake.
• For ZnVer1 and Atom, values were transferred form InstRWs.
• For SLM and BtVer2, values are from Agner.

This is split off from https://reviews.llvm.org/D47377

[RKSimon]

lib/Target/X86/X86InstrFPStack.td

658

Please pull these 5 FLD* entries out and classify as WriteFLDC (see X86Schedule.td)

lib/Target/X86/X86Schedule.td

146

Shorten the names to WriteFLD0, WriteFLD1, - allows you to remove all the indentation changes.

Please add WriteFLDC for the other math constants (FLDPI etc.)

lib/Target/X86/X86ScheduleBtVer2.td

275

Checked with the AMD16h SOG - these should be [JFPU1, JSTC], 3, [1,1,], 1

lib/Target/X86/X86ScheduleZnver1.td

818

Just replace these with SchedAlias instead of entries above - that's what is mainly used in znver1

[courbet]

• Use WriteFLD0/1 instread of WriteF80Constant0/1.
• Fix btver2 data as suggested by Simon.

[courbet]

Thanks

lib/Target/X86/X86Schedule.td

146

Shorten the names to WriteFLD0, WriteFLD1, - allows you to remove all the indentation changes.

Done.

Please add WriteFLDC for the other math constants (FLDPI etc.)

If you don't mind I'll tacle this one in a separate diff.

lib/Target/X86/X86ScheduleZnver1.td

818

I'm not sure what you mean.
Getting rid of the definition for ZnWriteFPU3 and alias it to WriteFLD0 ?

[RKSimon]

lib/Target/X86/X86ScheduleZnver1.td

818

You can remove the X86WriteRes<WriteFLD* entries and update these existing entries here:

// FLDZ.
def : SchedAlias<WriteFLD0, ZnWriteFPU13>;

// FLD1.
def : SchedAlias<WriteFLD1, ZnWriteFPU3>;

[courbet]

use SchedAliases on ZnVer1, update tests for BtVer2.

[RKSimon]

LGTM with 2 minors

lib/Target/X86/X86InstrFPStack.td

569

Drop the braces for single entries like this

lib/Target/X86/X86Schedule.td

146

Really pedantic but please could you put these new entries above WriteFLoad so all the WriteFLoad* tags are together.

[courbet]

Address review comments.

[courbet]

Thanks for the review !

lib/Target/X86/X86InstrFPStack.td

 def LD_Fp164 : FpIf64<(outs RFP64:$dst), (ins), ZeroArgFP,
                 [(set RFP64:$dst, fpimm1)]>;
 def LD_Fp080 : FpI_<(outs RFP80:$dst), (ins), ZeroArgFP,
                 [(set RFP80:$dst, fpimm0)]>;
 def LD_Fp180 : FpI_<(outs RFP80:$dst), (ins), ZeroArgFP,
                 [(set RFP80:$dst, fpimm1)]>;
 }
 
-let SchedRW = [WriteZero] in {
+let SchedRW = [WriteFLD0] in {
 def LD_F0 : FPI<0xD9, MRM_EE, (outs), (ins), "fldz">;
+} // SchedRW
+
+let SchedRW = [WriteFLD1] in {
 def LD_F1 : FPI<0xD9, MRM_E8, (outs), (ins), "fld1">;
-}
+} // SchedRW
 
 // Floating point compares.
 let SchedRW = [WriteFCom] in {
 def UCOM_Fpr32 : FpIf32<(outs), (ins RFP32:$lhs, RFP32:$rhs), CompareFP,
                         [(set FPSW, (trunc (X86cmp RFP32:$lhs, RFP32:$rhs)))]>;
 def UCOM_Fpr64 : FpIf64<(outs), (ins RFP64:$lhs, RFP64:$rhs), CompareFP,
                         [(set FPSW, (trunc (X86cmp RFP64:$lhs, RFP64:$rhs)))]>;
 def UCOM_Fpr80 : FpI_  <(outs), (ins RFP80:$lhs, RFP80:$rhs), CompareFP,

 let SchedRW = [WriteMicrocoded] in {
 let Defs = [FPSW] in {
 def WAIT : I<0x9B, RawFrm, (outs), (ins), "wait", []>;
 def FXAM : I<0xD9, MRM_E5, (outs), (ins), "fxam", []>;
 def FLDL2T : I<0xD9, MRM_E9, (outs), (ins), "fldl2t", []>;
 def FLDL2E : I<0xD9, MRM_EA, (outs), (ins), "fldl2e", []>;
 def FLDPI : I<0xD9, MRM_EB, (outs), (ins), "fldpi", []>;
 def FLDLG2 : I<0xD9, MRM_EC, (outs), (ins), "fldlg2", []>;
 def FLDLN2 : I<0xD9, MRM_ED, (outs), (ins), "fldln2", []>;
 def F2XM1 : I<0xD9, MRM_F0, (outs), (ins), "f2xm1", []>;
 def FYL2X : I<0xD9, MRM_F1, (outs), (ins), "fyl2x", []>;
 def FPTAN : I<0xD9, MRM_F2, (outs), (ins), "fptan", []>;
 def FPATAN : I<0xD9, MRM_F3, (outs), (ins), "fpatan", []>;
 def FXTRACT : I<0xD9, MRM_F4, (outs), (ins), "fxtract", []>;
 def FPREM1 : I<0xD9, MRM_F5, (outs), (ins), "fprem1", []>;
 def FDECSTP : I<0xD9, MRM_F6, (outs), (ins), "fdecstp", []>;
 def FINCSTP : I<0xD9, MRM_F7, (outs), (ins), "fincstp", []>;

lib/Target/X86/X86SchedBroadwell.td

 // Treat misc copies as a move.
 def : InstRW<[WriteMove], (instrs COPY)>;
 
 // Branches don't produce values, so they have no latency, but they still
 // consume resources. Indirect branches can fold loads.
 defm : BWWriteResPair<WriteJump,  [BWPort06],   1>;
 
 // Floating point. This covers both scalar and vector operations.
+defm : X86WriteRes<WriteFLD0,          [BWPort01], 1, [1], 1>;
+defm : X86WriteRes<WriteFLD1,          [BWPort01], 1, [2], 2>;
 defm : X86WriteRes<WriteFLoad,         [BWPort23], 5, [1], 1>;
 defm : X86WriteRes<WriteFLoadX,        [BWPort23], 5, [1], 1>;
 defm : X86WriteRes<WriteFLoadY,        [BWPort23], 6, [1], 1>;
 defm : X86WriteRes<WriteFMaskedLoad,   [BWPort23,BWPort5], 7, [1,2], 3>;
 defm : X86WriteRes<WriteFMaskedLoadY,  [BWPort23,BWPort5], 8, [1,2], 3>;
 defm : X86WriteRes<WriteFStore,        [BWPort237,BWPort4], 1, [1,1], 2>;
 defm : X86WriteRes<WriteFStoreX,       [BWPort237,BWPort4], 1, [1,1], 2>;
 defm : X86WriteRes<WriteFStoreY,       [BWPort237,BWPort4], 1, [1,1], 2>;

lib/Target/X86/X86SchedHaswell.td

 defm : HWWriteResPair<WriteDiv32,  [HWPort0, HWDivider], 25, [1,10], 1, 4>;
 defm : HWWriteResPair<WriteDiv64,  [HWPort0, HWDivider], 25, [1,10], 1, 4>;
 defm : HWWriteResPair<WriteIDiv8,  [HWPort0, HWDivider], 25, [1,10], 1, 4>;
 defm : HWWriteResPair<WriteIDiv16, [HWPort0, HWDivider], 25, [1,10], 1, 4>;
 defm : HWWriteResPair<WriteIDiv32, [HWPort0, HWDivider], 25, [1,10], 1, 4>;
 defm : HWWriteResPair<WriteIDiv64, [HWPort0, HWDivider], 25, [1,10], 1, 4>;
 
 // Scalar and vector floating point.
+defm : X86WriteRes<WriteFLD0,          [HWPort01], 1, [1], 1>;
+defm : X86WriteRes<WriteFLD1,          [HWPort01], 1, [2], 2>;
 defm : X86WriteRes<WriteFLoad,         [HWPort23], 5, [1], 1>;
 defm : X86WriteRes<WriteFLoadX,        [HWPort23], 6, [1], 1>;
 defm : X86WriteRes<WriteFLoadY,        [HWPort23], 7, [1], 1>;
 defm : X86WriteRes<WriteFMaskedLoad,   [HWPort23,HWPort5], 8, [1,2], 3>;
 defm : X86WriteRes<WriteFMaskedLoadY,  [HWPort23,HWPort5], 9, [1,2], 3>;
 defm : X86WriteRes<WriteFStore,        [HWPort237,HWPort4], 1, [1,1], 2>;
 defm : X86WriteRes<WriteFStoreX,       [HWPort237,HWPort4], 1, [1,1], 2>;
 defm : X86WriteRes<WriteFStoreY,       [HWPort237,HWPort4], 1, [1,1], 2>;
def HWWriteFBLD : SchedWriteRes<[]> {
   let NumMicroOps = 43;
 }
 def : InstRW<[HWWriteFBLD], (instregex "FBLDm")>;
 
 // FST(P).
 // r.
 def : InstRW<[HWWriteP01], (instregex "ST_(F|FP)rr")>;
 
-// FLDZ.
-def : InstRW<[HWWriteP01], (instregex "LD_F0")>;
-
 // FLDPI FLDL2E etc.
 def : InstRW<[HWWrite2P01], (instregex "FLDPI", "FLDL2(T|E)", "FLDL(G|N)2")>;
 
 // FFREE.
 def : InstRW<[HWWriteP01], (instregex "FFREE")>;
 
 // FNSAVE.
 def HWWriteFNSAVE : SchedWriteRes<[]> {

lib/Target/X86/X86SchedSandyBridge.td

 defm : SBWriteResPair<WritePOPCNT,  [SBPort1], 3, [1], 1, 6>;
 
 // BMI1 BEXTR, BMI2 BZHI
 // NOTE: These don't exist on Sandy Bridge. Ports are guesses.
 defm : SBWriteResPair<WriteBEXTR, [SBPort05,SBPort1], 2, [1,1], 2>;
 defm : SBWriteResPair<WriteBZHI, [SBPort1], 1>;
 
 // Scalar and vector floating point.
+defm : X86WriteRes<WriteFLD0,          [SBPort5], 1, [1], 1>;
+defm : X86WriteRes<WriteFLD1,          [SBPort0,SBPort5], 1, [1,1], 2>;
 defm : X86WriteRes<WriteFLoad,         [SBPort23], 5, [1], 1>;
 defm : X86WriteRes<WriteFLoadX,        [SBPort23], 6, [1], 1>;
 defm : X86WriteRes<WriteFLoadY,        [SBPort23], 7, [1], 1>;
 defm : X86WriteRes<WriteFMaskedLoad,   [SBPort23,SBPort05], 8, [1,2], 3>;
 defm : X86WriteRes<WriteFMaskedLoadY,  [SBPort23,SBPort05], 9, [1,2], 3>;
 defm : X86WriteRes<WriteFStore,        [SBPort23,SBPort4], 1, [1,1], 1>;
 defm : X86WriteRes<WriteFStoreX,       [SBPort23,SBPort4], 1, [1,1], 1>;
 defm : X86WriteRes<WriteFStoreY,       [SBPort23,SBPort4], 1, [1,1], 1>;

lib/Target/X86/X86SchedSkylakeClient.td

 // These can often bypass execution ports completely.
 def : WriteRes<WriteZero,  []>;
 
 // Branches don't produce values, so they have no latency, but they still
 // consume resources. Indirect branches can fold loads.
 defm : SKLWriteResPair<WriteJump,  [SKLPort06],   1>;
 
 // Floating point. This covers both scalar and vector operations.
+defm : X86WriteRes<WriteFLD0,          [SKLPort05], 1, [1], 1>;
+defm : X86WriteRes<WriteFLD1,          [SKLPort05], 1, [2], 2>;
 defm : X86WriteRes<WriteFLoad,         [SKLPort23], 5, [1], 1>;
 defm : X86WriteRes<WriteFLoadX,        [SKLPort23], 6, [1], 1>;
 defm : X86WriteRes<WriteFLoadY,        [SKLPort23], 7, [1], 1>;
 defm : X86WriteRes<WriteFMaskedLoad,   [SKLPort23,SKLPort015], 7, [1,1], 2>;
 defm : X86WriteRes<WriteFMaskedLoadY,  [SKLPort23,SKLPort015], 8, [1,1], 2>;
 defm : X86WriteRes<WriteFStore,        [SKLPort237,SKLPort4], 1, [1,1], 2>;
 defm : X86WriteRes<WriteFStoreX,       [SKLPort237,SKLPort4], 1, [1,1], 2>;
 defm : X86WriteRes<WriteFStoreY,       [SKLPort237,SKLPort4], 1, [1,1], 2>;

lib/Target/X86/X86SchedSkylakeServer.td

 // These can often bypass execution ports completely.
 def : WriteRes<WriteZero,  []>;
 
 // Branches don't produce values, so they have no latency, but they still
 // consume resources. Indirect branches can fold loads.
 defm : SKXWriteResPair<WriteJump,  [SKXPort06],   1>;
 
 // Floating point. This covers both scalar and vector operations.
+defm : X86WriteRes<WriteFLD0,          [SKXPort05], 1, [1], 1>;
+defm : X86WriteRes<WriteFLD1,          [SKXPort05], 1, [2], 2>;
 defm : X86WriteRes<WriteFLoad,         [SKXPort23], 5, [1], 1>;
 defm : X86WriteRes<WriteFLoadX,        [SKXPort23], 6, [1], 1>;
 defm : X86WriteRes<WriteFLoadY,        [SKXPort23], 7, [1], 1>;
 defm : X86WriteRes<WriteFMaskedLoad,   [SKXPort23,SKXPort015], 7, [1,1], 2>;
 defm : X86WriteRes<WriteFMaskedLoadY,  [SKXPort23,SKXPort015], 8, [1,1], 2>;
 defm : X86WriteRes<WriteFStore,        [SKXPort237,SKXPort4], 1, [1,1], 2>;
 defm : X86WriteRes<WriteFStoreX,       [SKXPort237,SKXPort4], 1, [1,1], 2>;
 defm : X86WriteRes<WriteFStoreY,       [SKXPort237,SKXPort4], 1, [1,1], 2>;

lib/Target/X86/X86Schedule.td

 def WriteZero : SchedWrite;
 
 // Branches don't produce values, so they have no latency, but they still
 // consume resources. Indirect branches can fold loads.
 defm WriteJump : X86SchedWritePair;
 
 // Floating point. This covers both scalar and vector operations.
 def  WriteFLoad         : SchedWrite;
+def  WriteFLD0          : SchedWrite;
+def  WriteFLD1          : SchedWrite;
 def  WriteFLoadX        : SchedWrite;
 def  WriteFLoadY        : SchedWrite;
 def  WriteFMaskedLoad   : SchedWrite;
 def  WriteFMaskedLoadY  : SchedWrite;
 def  WriteFStore        : SchedWrite;
 def  WriteFStoreX       : SchedWrite;
 def  WriteFStoreY       : SchedWrite;
 def  WriteFStoreNT      : SchedWrite;

lib/Target/X86/X86ScheduleAtom.td

 // Nops don't have dependencies, so there's no actual latency, but we set this
 // to '1' to tell the scheduler that the nop uses an ALU slot for a cycle.
 def : WriteRes<WriteNop, [AtomPort01]>;
 
 ////////////////////////////////////////////////////////////////////////////////
 // Floating point. This covers both scalar and vector operations.
 ////////////////////////////////////////////////////////////////////////////////
 
+defm : X86WriteRes<WriteFLD0,       [AtomPort01], 1, [1], 1>;
+defm : X86WriteRes<WriteFLD1,       [AtomPort01], 6, [6], 1>;
 def  : WriteRes<WriteFLoad,         [AtomPort0]>;
 def  : WriteRes<WriteFLoadX,        [AtomPort0]>;
 def  : WriteRes<WriteFLoadY,        [AtomPort0]>;
 def  : WriteRes<WriteFMaskedLoad,   [AtomPort0]>;
 def  : WriteRes<WriteFMaskedLoadY,  [AtomPort0]>;
 
 def  : WriteRes<WriteFStore,        [AtomPort0]>;
 def  : WriteRes<WriteFStoreX,       [AtomPort0]>;
 def  : WriteRes<WriteFStoreY,       [AtomPort0]>;
 def  : WriteRes<WriteFStoreNT,      [AtomPort0]>;
 def  : WriteRes<WriteFStoreNTX,     [AtomPort0]>;
 def  : WriteRes<WriteFStoreNTY,     [AtomPort0]>;
 def  : WriteRes<WriteFMaskedStore,  [AtomPort0]>;
 def  : WriteRes<WriteFMaskedStoreY, [AtomPort0]>;
 
 def  : WriteRes<WriteFMove,         [AtomPort01]>;
 def  : WriteRes<WriteFMoveX,        [AtomPort01]>;
 def  : WriteRes<WriteFMoveY,        [AtomPort01]>;
-                                    
 defm : X86WriteRes<WriteEMMS,       [AtomPort01], 5, [5], 1>;
 
 defm : AtomWriteResPair<WriteFAdd,           [AtomPort0],  [AtomPort0],  5,  5,  [5],  [5]>;
 defm : AtomWriteResPair<WriteFAddX,          [AtomPort0],  [AtomPort0],  5,  5,  [5],  [5]>;
 defm : AtomWriteResPair<WriteFAddY,          [AtomPort0],  [AtomPort0],  5,  5,  [5],  [5]>;
 defm : AtomWriteResPair<WriteFAdd64,         [AtomPort0],  [AtomPort0],  5,  5,  [5],  [5]>;
 defm : AtomWriteResPair<WriteFAdd64X,       [AtomPort01], [AtomPort01],  6,  7,  [6],  [7]>;
 defm : AtomWriteResPair<WriteFAdd64Y,       [AtomPort01], [AtomPort01],  6,  7,  [6],  [7]>;

 def : InstRW<[AtomWrite0_1_5], (instrs MMX_CVTPS2PIirm, MMX_CVTTPS2PIirm)>;
 def : InstRW<[AtomWrite0_1_5], (instregex "ILD_F(16|32|64)")>;
 
 // Port0 or Port1
 def AtomWrite01_1 : SchedWriteRes<[AtomPort01]> {
   let Latency = 1;
   let ResourceCycles = [1];
 }
-def : InstRW<[AtomWrite01_1], (instrs FDECSTP, FFREE, FFREEP, FINCSTP, LD_F0, WAIT,
+def : InstRW<[AtomWrite01_1], (instrs FDECSTP, FFREE, FFREEP, FINCSTP, WAIT,
                                       LFENCE,
                                       STOSB, STOSL, STOSQ, STOSW,
                                       MOVSSrr, MOVSSrr_REV,
                                       PSLLDQri, PSRLDQri)>;
 def : InstRW<[AtomWrite01_1], (instregex "MMX_PACK(SSDW|SSWB|USWB)irr",
                                          "MMX_PUNPCKH(BW|DQ|WD)irr")>;
 
 def AtomWrite01_2 : SchedWriteRes<[AtomPort01]> {

 }
 def : InstRW<[AtomWrite01_5], (instrs FLDCW16m, ST_FP80m)>;
 def : InstRW<[AtomWrite01_5], (instregex "MMX_PH(ADD|SUB)S?Wrr")>;
 
 def AtomWrite01_6 : SchedWriteRes<[AtomPort01]> {
   let Latency = 6;
   let ResourceCycles = [6];
 }
-def : InstRW<[AtomWrite01_6], (instrs LD_F1, CMPXCHG8rm, INTO, XLAT,
+def : InstRW<[AtomWrite01_6], (instrs CMPXCHG8rm, INTO, XLAT,
                                       SHLD16rrCL, SHRD16rrCL,
                                       SHLD16rri8, SHRD16rri8,
                                       SHLD16mrCL, SHRD16mrCL,
                                       SHLD16mri8, SHRD16mri8)>;
 def : InstRW<[AtomWrite01_6], (instregex "IMUL16rr",
                                          "IST_F(P)?(16|32|64)?m",
                                          "MMX_PH(ADD|SUB)S?Wrm")>;
 

lib/Target/X86/X86ScheduleBtVer2.td

 // Nops don't have dependencies, so there's no actual latency, but we set this
 // to '1' to tell the scheduler that the nop uses an ALU slot for a cycle.
 def : WriteRes<WriteNop, [JALU01]> { let Latency = 1; }
 
 ////////////////////////////////////////////////////////////////////////////////
 // Floating point. This covers both scalar and vector operations.
 ////////////////////////////////////////////////////////////////////////////////
 
+defm : X86WriteRes<WriteFLD0,          [JFPU1, JSTC], 3, [1,1], 1>;
+defm : X86WriteRes<WriteFLD1,          [JFPU1, JSTC], 3, [1,1], 1>;
 defm : X86WriteRes<WriteFLoad,         [JLAGU, JFPU01, JFPX], 5, [1, 1, 1], 1>;
 defm : X86WriteRes<WriteFLoadX,        [JLAGU, JFPU01, JFPX], 5, [1, 1, 1], 1>;
 defm : X86WriteRes<WriteFLoadY,        [JLAGU, JFPU01, JFPX], 5, [1, 1, 1], 1>;
 defm : X86WriteRes<WriteFMaskedLoad,   [JLAGU, JFPU01, JFPX], 6, [1, 1, 2], 1>;
 defm : X86WriteRes<WriteFMaskedLoadY,  [JLAGU, JFPU01, JFPX], 6, [2, 2, 4], 2>;
 
 defm : X86WriteRes<WriteFStore,        [JSAGU, JFPU1,  JSTC], 2, [1, 1, 1], 1>;
 defm : X86WriteRes<WriteFStoreX,       [JSAGU, JFPU1,  JSTC], 1, [1, 1, 1], 1>;

lib/Target/X86/X86ScheduleSLM.td

 defm : SLMWriteResPair<WriteDiv32,  [SLM_IEC_RSV01, SLMDivider], 25, [1,25], 1, 4>;
 defm : SLMWriteResPair<WriteDiv64,  [SLM_IEC_RSV01, SLMDivider], 25, [1,25], 1, 4>;
 defm : SLMWriteResPair<WriteIDiv8,  [SLM_IEC_RSV01, SLMDivider], 25, [1,25], 1, 4>;
 defm : SLMWriteResPair<WriteIDiv16, [SLM_IEC_RSV01, SLMDivider], 25, [1,25], 1, 4>;
 defm : SLMWriteResPair<WriteIDiv32, [SLM_IEC_RSV01, SLMDivider], 25, [1,25], 1, 4>;
 defm : SLMWriteResPair<WriteIDiv64, [SLM_IEC_RSV01, SLMDivider], 25, [1,25], 1, 4>;
 
 // Scalar and vector floating point.
+defm : X86WriteRes<WriteFLD0,       [SLM_FPC_RSV01], 1, [1], 1>;
+defm : X86WriteRes<WriteFLD1,       [SLM_FPC_RSV01], 1, [1], 1>;
 def  : WriteRes<WriteFLoad,         [SLM_MEC_RSV]> { let Latency = 3; }
 def  : WriteRes<WriteFLoadX,        [SLM_MEC_RSV]> { let Latency = 3; }
 def  : WriteRes<WriteFLoadY,        [SLM_MEC_RSV]> { let Latency = 3; }
 def  : WriteRes<WriteFMaskedLoad,   [SLM_MEC_RSV]> { let Latency = 3; }
 def  : WriteRes<WriteFMaskedLoadY,  [SLM_MEC_RSV]> { let Latency = 3; }
 def  : WriteRes<WriteFStore,        [SLM_MEC_RSV]>;
 def  : WriteRes<WriteFStoreX,       [SLM_MEC_RSV]>;
 def  : WriteRes<WriteFStoreY,       [SLM_MEC_RSV]>;

lib/Target/X86/X86ScheduleZnver1.td

 defm : ZnWriteResPair<WriteIDiv64, [ZnALU2, ZnDivider], 41, [1,41], 2>;
 
 // IMULH
 def  : WriteRes<WriteIMulH, [ZnALU1, ZnMultiplier]>{
   let Latency = 4;
 }
 
 // Floating point operations
+defm : X86WriteRes<WriteFLD0,          [ZnAGU,ZnFPU13], 8, [1,1], 2>;
+defm : X86WriteRes<WriteFLD1,          [ZnAGU,ZnFPU3], 11, [1,1], 2>;
 defm : X86WriteRes<WriteFLoad,         [ZnAGU], 8, [1], 1>;
 defm : X86WriteRes<WriteFLoadX,        [ZnAGU], 8, [1], 1>;
 defm : X86WriteRes<WriteFLoadY,        [ZnAGU], 8, [1], 1>;
 defm : X86WriteRes<WriteFMaskedLoad,   [ZnAGU,ZnFPU01], 8, [1,1], 1>;
 defm : X86WriteRes<WriteFMaskedLoadY,  [ZnAGU,ZnFPU01], 8, [1,2], 2>;
 defm : X86WriteRes<WriteFStore,        [ZnAGU], 1, [1,1], 1>;
 defm : X86WriteRes<WriteFStoreX,       [ZnAGU], 1, [1,1], 1>;
 defm : X86WriteRes<WriteFStoreY,       [ZnAGU], 1, [1,1], 1>;

 def ZnWriteFPU13 : SchedWriteRes<[ZnAGU, ZnFPU13]> {
   let Latency = 8;
 }
 
 def ZnWriteFPU3 : SchedWriteRes<[ZnAGU, ZnFPU3]> {
   let Latency = 11;
 }
 
-// FLDZ.
-def : InstRW<[ZnWriteFPU13], (instregex "LD_F0")>;
-
-// FLD1.
-def : InstRW<[ZnWriteFPU3], (instregex "LD_F1")>;
-
 // FLDPI FLDL2E etc.
 def : InstRW<[ZnWriteFPU3], (instregex "FLDPI", "FLDL2(T|E)", "FLDL(G|N)2")>;
 
 // FNSTSW.
 // AX.
 def : InstRW<[WriteMicrocoded], (instrs FNSTSW16r)>;
 
 // m16.

test/CodeGen/X86/x87-schedule.ll

 ; ATOM-NEXT:    fldpi # sched: [10:5.00]
 ; ATOM-NEXT:    fldz # sched: [1:0.50]
 ; ATOM-NEXT:    #NO_APP
 ; ATOM-NEXT:    retl # sched: [79:39.50]
 ;
 ; SLM-LABEL: test_fld1_fldl2e_fldl2t_fldlg2_fldln2_fldpi_fldz:
 ; SLM:       # %bb.0:
 ; SLM-NEXT:    #APP
-; SLM-NEXT:    fld1 # sched: [1:?]
+; SLM-NEXT:    fld1 # sched: [1:0.50]
 ; SLM-NEXT:    fldl2e # sched: [100:1.00]
 ; SLM-NEXT:    fldl2t # sched: [100:1.00]
 ; SLM-NEXT:    fldlg2 # sched: [100:1.00]
 ; SLM-NEXT:    fldln2 # sched: [100:1.00]
 ; SLM-NEXT:    fldpi # sched: [100:1.00]
-; SLM-NEXT:    fldz # sched: [1:?]
+; SLM-NEXT:    fldz # sched: [1:0.50]
 ; SLM-NEXT:    #NO_APP
 ; SLM-NEXT:    retl # sched: [4:1.00]
 ;
 ; SANDY-LABEL: test_fld1_fldl2e_fldl2t_fldlg2_fldln2_fldpi_fldz:
 ; SANDY:       # %bb.0:
 ; SANDY-NEXT:    #APP
-; SANDY-NEXT:    fld1 # sched: [1:?]
+; SANDY-NEXT:    fld1 # sched: [1:1.00]
 ; SANDY-NEXT:    fldl2e # sched: [100:0.33]
 ; SANDY-NEXT:    fldl2t # sched: [100:0.33]
 ; SANDY-NEXT:    fldlg2 # sched: [100:0.33]
 ; SANDY-NEXT:    fldln2 # sched: [100:0.33]
 ; SANDY-NEXT:    fldpi # sched: [100:0.33]
-; SANDY-NEXT:    fldz # sched: [1:?]
+; SANDY-NEXT:    fldz # sched: [1:1.00]
 ; SANDY-NEXT:    #NO_APP
 ; SANDY-NEXT:    retl # sched: [6:1.00]
 ;
 ; HASWELL-LABEL: test_fld1_fldl2e_fldl2t_fldlg2_fldln2_fldpi_fldz:
 ; HASWELL:       # %bb.0:
 ; HASWELL-NEXT:    #APP
-; HASWELL-NEXT:    fld1 # sched: [1:?]
+; HASWELL-NEXT:    fld1 # sched: [1:1.00]
 ; HASWELL-NEXT:    fldl2e # sched: [1:0.50]
 ; HASWELL-NEXT:    fldl2t # sched: [1:0.50]
 ; HASWELL-NEXT:    fldlg2 # sched: [1:0.50]
 ; HASWELL-NEXT:    fldln2 # sched: [1:0.50]
 ; HASWELL-NEXT:    fldpi # sched: [1:0.50]
 ; HASWELL-NEXT:    fldz # sched: [1:0.50]
 ; HASWELL-NEXT:    #NO_APP
 ; HASWELL-NEXT:    retl # sched: [7:1.00]
 ;
 ; BROADWELL-LABEL: test_fld1_fldl2e_fldl2t_fldlg2_fldln2_fldpi_fldz:
 ; BROADWELL:       # %bb.0:
 ; BROADWELL-NEXT:    #APP
-; BROADWELL-NEXT:    fld1 # sched: [1:?]
+; BROADWELL-NEXT:    fld1 # sched: [1:1.00]
 ; BROADWELL-NEXT:    fldl2e # sched: [100:0.25]
 ; BROADWELL-NEXT:    fldl2t # sched: [100:0.25]
 ; BROADWELL-NEXT:    fldlg2 # sched: [100:0.25]
 ; BROADWELL-NEXT:    fldln2 # sched: [100:0.25]
 ; BROADWELL-NEXT:    fldpi # sched: [100:0.25]
-; BROADWELL-NEXT:    fldz # sched: [1:?]
+; BROADWELL-NEXT:    fldz # sched: [1:0.50]
 ; BROADWELL-NEXT:    #NO_APP
 ; BROADWELL-NEXT:    retl # sched: [6:0.50]
 ;
 ; SKYLAKE-LABEL: test_fld1_fldl2e_fldl2t_fldlg2_fldln2_fldpi_fldz:
 ; SKYLAKE:       # %bb.0:
 ; SKYLAKE-NEXT:    #APP
-; SKYLAKE-NEXT:    fld1 # sched: [1:?]
+; SKYLAKE-NEXT:    fld1 # sched: [1:1.00]
 ; SKYLAKE-NEXT:    fldl2e # sched: [100:0.25]
 ; SKYLAKE-NEXT:    fldl2t # sched: [100:0.25]
 ; SKYLAKE-NEXT:    fldlg2 # sched: [100:0.25]
 ; SKYLAKE-NEXT:    fldln2 # sched: [100:0.25]
 ; SKYLAKE-NEXT:    fldpi # sched: [100:0.25]
-; SKYLAKE-NEXT:    fldz # sched: [1:?]
+; SKYLAKE-NEXT:    fldz # sched: [1:0.50]
 ; SKYLAKE-NEXT:    #NO_APP
 ; SKYLAKE-NEXT:    retl # sched: [6:0.50]
 ;
 ; SKX-LABEL: test_fld1_fldl2e_fldl2t_fldlg2_fldln2_fldpi_fldz:
 ; SKX:       # %bb.0:
 ; SKX-NEXT:    #APP
-; SKX-NEXT:    fld1 # sched: [1:?]
+; SKX-NEXT:    fld1 # sched: [1:1.00]
 ; SKX-NEXT:    fldl2e # sched: [100:0.25]
 ; SKX-NEXT:    fldl2t # sched: [100:0.25]
 ; SKX-NEXT:    fldlg2 # sched: [100:0.25]
 ; SKX-NEXT:    fldln2 # sched: [100:0.25]
 ; SKX-NEXT:    fldpi # sched: [100:0.25]
-; SKX-NEXT:    fldz # sched: [1:?]
+; SKX-NEXT:    fldz # sched: [1:0.50]
 ; SKX-NEXT:    #NO_APP
 ; SKX-NEXT:    retl # sched: [6:0.50]
 ;
 ; BTVER2-LABEL: test_fld1_fldl2e_fldl2t_fldlg2_fldln2_fldpi_fldz:
 ; BTVER2:       # %bb.0:
 ; BTVER2-NEXT:    #APP
-; BTVER2-NEXT:    fld1 # sched: [1:?]
+; BTVER2-NEXT:    fld1 # sched: [1:1.00]
 ; BTVER2-NEXT:    fldl2e # sched: [100:0.50]
 ; BTVER2-NEXT:    fldl2t # sched: [100:0.50]
 ; BTVER2-NEXT:    fldlg2 # sched: [100:0.50]
 ; BTVER2-NEXT:    fldln2 # sched: [100:0.50]
 ; BTVER2-NEXT:    fldpi # sched: [100:0.50]
-; BTVER2-NEXT:    fldz # sched: [1:?]
+; BTVER2-NEXT:    fldz # sched: [1:1.00]
 ; BTVER2-NEXT:    #NO_APP
 ; BTVER2-NEXT:    retl # sched: [4:1.00]
 ;
 ; ZNVER1-LABEL: test_fld1_fldl2e_fldl2t_fldlg2_fldln2_fldpi_fldz:
 ; ZNVER1:       # %bb.0:
 ; ZNVER1-NEXT:    #APP
 ; ZNVER1-NEXT:    fld1 # sched: [11:1.00]
 ; ZNVER1-NEXT:    fldl2e # sched: [11:1.00]

test/tools/llvm-mca/X86/Broadwell/resources-x87.s

 # CHECK-NEXT:  3      4     1.00           *      *     fisttpl (%ecx)
 # CHECK-NEXT:  3      4     1.00           *      *     fisttpll  (%eax)
 # CHECK-NEXT:  1      1     0.25                  *     fld %st(0)
 # CHECK-NEXT:  1      6     0.50    *             *     flds  (%edx)
 # CHECK-NEXT:  1      6     0.50    *             *     fldl  (%ecx)
 # CHECK-NEXT:  1      6     0.50    *             *     fldt  (%eax)
 # CHECK-NEXT:  3      7     1.00    *             *     fldcw (%eax)
 # CHECK-NEXT:  64     60    14.00                 *     fldenv  (%eax)
-# CHECK-NEXT:  1      1      -                    *     fld1
+# CHECK-NEXT:  2      1     1.00                  *     fld1
 # CHECK-NEXT:  1      100   0.25                  *     fldl2e
 # CHECK-NEXT:  1      100   0.25                  *     fldl2t
 # CHECK-NEXT:  1      100   0.25                  *     fldlg2
 # CHECK-NEXT:  1      100   0.25                  *     fldln2
 # CHECK-NEXT:  1      100   0.25                  *     fldpi
-# CHECK-NEXT:  1      1      -                    *     fldz
+# CHECK-NEXT:  1      1     0.50                  *     fldz
 # CHECK-NEXT:  1      5     1.00                  *     fmul  %st(0), %st(1)
 # CHECK-NEXT:  1      5     1.00                  *     fmul  %st(2)
 # CHECK-NEXT:  2      11    1.00    *             *     fmuls (%ecx)
 # CHECK-NEXT:  2      11    1.00    *             *     fmull (%eax)
 # CHECK-NEXT:  1      5     1.00                  *     fmulp %st(1)
 # CHECK-NEXT:  1      5     1.00                  *     fmulp %st(2)
 # CHECK-NEXT:  3      14    1.00    *             *     fimuls  (%ecx)
 # CHECK-NEXT:  3      14    1.00    *             *     fimull  (%eax)

 # CHECK-NEXT: [5]   - BWPort3
 # CHECK-NEXT: [6]   - BWPort4
 # CHECK-NEXT: [7]   - BWPort5
 # CHECK-NEXT: [8]   - BWPort6
 # CHECK-NEXT: [9]   - BWPort7
 
 # CHECK:      Resource pressure per iteration:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    [8]    [9]
-# CHECK-NEXT:  -     9.00   111.67 140.67 49.00  49.00  27.00  60.67  71.00  9.00
+# CHECK-NEXT:  -     9.00   113.17 142.17 49.00  49.00  27.00  60.67  71.00  9.00
 
 # CHECK:      Resource pressure by instruction:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    [8]    [9]    Instructions:
 # CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     f2xm1
 # CHECK-NEXT:  -      -      -      -      -      -      -     1.00    -      -     fabs
 # CHECK-NEXT:  -      -      -     1.00    -      -      -      -      -      -     fadd  %st(0), %st(1)
 # CHECK-NEXT:  -      -      -     1.00    -      -      -      -      -      -     fadd  %st(2)
 # CHECK-NEXT:  -      -      -     1.00   0.50   0.50    -      -      -      -     fadds (%ecx)

 # CHECK-NEXT:  -      -      -     1.00   0.33   0.33   1.00    -      -     0.33   fisttpl (%ecx)
 # CHECK-NEXT:  -      -      -     1.00   0.33   0.33   1.00    -      -     0.33   fisttpll  (%eax)
 # CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     fld %st(0)
 # CHECK-NEXT:  -      -      -      -     0.50   0.50    -      -      -      -     flds  (%edx)
 # CHECK-NEXT:  -      -      -      -     0.50   0.50    -      -      -      -     fldl  (%ecx)
 # CHECK-NEXT:  -      -      -      -     0.50   0.50    -      -      -      -     fldt  (%eax)
 # CHECK-NEXT:  -      -     1.50   0.50   0.50   0.50    -      -      -      -     fldcw (%eax)
 # CHECK-NEXT:  -      -     18.92  11.42  4.00   4.00    -     10.92  14.75   -     fldenv  (%eax)
-# CHECK-NEXT:  -      -      -      -      -      -      -      -      -      -     fld1
+# CHECK-NEXT:  -      -     1.00   1.00    -      -      -      -      -      -     fld1
 # CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     fldl2e
 # CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     fldl2t
 # CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     fldlg2
 # CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     fldln2
 # CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     fldpi
-# CHECK-NEXT:  -      -      -      -      -      -      -      -      -      -     fldz
+# CHECK-NEXT:  -      -     0.50   0.50    -      -      -      -      -      -     fldz
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -     fmul  %st(0), %st(1)
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -     fmul  %st(2)
 # CHECK-NEXT:  -      -     1.00    -     0.50   0.50    -      -      -      -     fmuls (%ecx)
 # CHECK-NEXT:  -      -     1.00    -     0.50   0.50    -      -      -      -     fmull (%eax)
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -     fmulp %st(1)
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -     fmulp %st(2)
 # CHECK-NEXT:  -      -     1.00   1.00   0.50   0.50    -      -      -      -     fimuls  (%ecx)
 # CHECK-NEXT:  -      -     1.00   1.00   0.50   0.50    -      -      -      -     fimull  (%eax)

test/tools/llvm-mca/X86/BtVer2/resources-x87.s

 # CHECK-NEXT:  1      1     1.00           *      *     fisttpl (%ecx)
 # CHECK-NEXT:  1      1     1.00           *      *     fisttpll  (%eax)
 # CHECK-NEXT:  1      1     0.50                  *     fld %st(0)
 # CHECK-NEXT:  1      5     1.00    *             *     flds  (%edx)
 # CHECK-NEXT:  1      5     1.00    *             *     fldl  (%ecx)
 # CHECK-NEXT:  1      5     1.00    *             *     fldt  (%eax)
 # CHECK-NEXT:  1      5     1.00    *             *     fldcw (%eax)
 # CHECK-NEXT:  1      100   0.50                  *     fldenv  (%eax)
-# CHECK-NEXT:  1      1      -                    *     fld1
+# CHECK-NEXT:  1      1     1.00                  *     fld1
 # CHECK-NEXT:  1      100   0.50                  *     fldl2e
 # CHECK-NEXT:  1      100   0.50                  *     fldl2t
 # CHECK-NEXT:  1      100   0.50                  *     fldlg2
 # CHECK-NEXT:  1      100   0.50                  *     fldln2
 # CHECK-NEXT:  1      100   0.50                  *     fldpi
-# CHECK-NEXT:  1      1      -                    *     fldz
+# CHECK-NEXT:  1      1     1.00                  *     fldz
 # CHECK-NEXT:  1      2     1.00                  *     fmul  %st(0), %st(1)
 # CHECK-NEXT:  1      2     1.00                  *     fmul  %st(2)
 # CHECK-NEXT:  1      7     1.00    *             *     fmuls (%ecx)
 # CHECK-NEXT:  1      7     1.00    *             *     fmull (%eax)
 # CHECK-NEXT:  1      2     1.00                  *     fmulp %st(1)
 # CHECK-NEXT:  1      2     1.00                  *     fmulp %st(2)
 # CHECK-NEXT:  1      7     1.00    *             *     fimuls  (%ecx)
 # CHECK-NEXT:  1      7     1.00    *             *     fimull  (%eax)

 # CHECK-NEXT: [9]   - JSAGU
 # CHECK-NEXT: [10]  - JSTC
 # CHECK-NEXT: [11]  - JVALU0
 # CHECK-NEXT: [12]  - JVALU1
 # CHECK-NEXT: [13]  - JVIMUL
 
 # CHECK:      Resource pressure per iteration:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    [8]    [9]    [10]   [11]   [12]   [13]
-# CHECK-NEXT: 44.50  22.50   -     54.00  349.00 54.00  27.00  39.00   -     13.00   -      -      -      -
+# CHECK-NEXT: 44.50  22.50   -     54.00  349.00 54.00  29.00  39.00   -     13.00   -      -      -      -
 
 # CHECK:      Resource pressure by instruction:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    [8]    [9]    [10]   [11]   [12]   [13]   Instructions:
 # CHECK-NEXT: 0.50   0.50    -      -      -      -      -      -      -      -      -      -      -      -     f2xm1
 # CHECK-NEXT:  -      -      -      -     1.00    -     1.00    -      -      -      -      -      -      -     fabs
 # CHECK-NEXT:  -      -      -     1.00    -     1.00    -      -      -      -      -      -      -      -     fadd  %st(0), %st(1)
 # CHECK-NEXT:  -      -      -     1.00    -     1.00    -      -      -      -      -      -      -      -     fadd  %st(2)
 # CHECK-NEXT:  -      -      -     1.00    -     1.00    -     1.00    -      -      -      -      -      -     fadds (%ecx)

 # CHECK-NEXT:  -      -      -      -      -      -      -      -      -     1.00    -      -      -      -     fisttpl (%ecx)
 # CHECK-NEXT:  -      -      -      -      -      -      -      -      -     1.00    -      -      -      -     fisttpll  (%eax)
 # CHECK-NEXT: 0.50   0.50    -      -      -      -      -      -      -      -      -      -      -      -     fld %st(0)
 # CHECK-NEXT:  -      -      -      -      -      -      -     1.00    -      -      -      -      -      -     flds  (%edx)
 # CHECK-NEXT:  -      -      -      -      -      -      -     1.00    -      -      -      -      -      -     fldl  (%ecx)
 # CHECK-NEXT:  -      -      -      -      -      -      -     1.00    -      -      -      -      -      -     fldt  (%eax)
 # CHECK-NEXT:  -      -      -      -      -      -      -     1.00    -      -      -      -      -      -     fldcw (%eax)
 # CHECK-NEXT: 0.50   0.50    -      -      -      -      -      -      -      -      -      -      -      -     fldenv  (%eax)
-# CHECK-NEXT:  -      -      -      -      -      -      -      -      -      -      -      -      -      -     fld1
+# CHECK-NEXT:  -      -      -      -      -      -     1.00    -      -      -      -      -      -      -     fld1
 # CHECK-NEXT: 0.50   0.50    -      -      -      -      -      -      -      -      -      -      -      -     fldl2e
 # CHECK-NEXT: 0.50   0.50    -      -      -      -      -      -      -      -      -      -      -      -     fldl2t
 # CHECK-NEXT: 0.50   0.50    -      -      -      -      -      -      -      -      -      -      -      -     fldlg2
 # CHECK-NEXT: 0.50   0.50    -      -      -      -      -      -      -      -      -      -      -      -     fldln2
 # CHECK-NEXT: 0.50   0.50    -      -      -      -      -      -      -      -      -      -      -      -     fldpi
-# CHECK-NEXT:  -      -      -      -      -      -      -      -      -      -      -      -      -      -     fldz
+# CHECK-NEXT:  -      -      -      -      -      -     1.00    -      -      -      -      -      -      -     fldz
 # CHECK-NEXT:  -      -      -      -     1.00    -     1.00    -      -      -      -      -      -      -     fmul  %st(0), %st(1)
 # CHECK-NEXT:  -      -      -      -     1.00    -     1.00    -      -      -      -      -      -      -     fmul  %st(2)
 # CHECK-NEXT:  -      -      -      -     1.00    -     1.00   1.00    -      -      -      -      -      -     fmuls (%ecx)
 # CHECK-NEXT:  -      -      -      -     1.00    -     1.00   1.00    -      -      -      -      -      -     fmull (%eax)
 # CHECK-NEXT:  -      -      -      -     1.00    -     1.00    -      -      -      -      -      -      -     fmulp %st(1)
 # CHECK-NEXT:  -      -      -      -     1.00    -     1.00    -      -      -      -      -      -      -     fmulp %st(2)
 # CHECK-NEXT:  -      -      -      -     1.00    -     1.00   1.00    -      -      -      -      -      -     fimuls  (%ecx)
 # CHECK-NEXT:  -      -      -      -     1.00    -     1.00   1.00    -      -      -      -      -      -     fimull  (%eax)

test/tools/llvm-mca/X86/Haswell/resources-x87.s

 # CHECK-NEXT:  3      4     1.00           *      *     fisttpl (%ecx)
 # CHECK-NEXT:  3      4     1.00           *      *     fisttpll  (%eax)
 # CHECK-NEXT:  1      1     0.50                  *     fld %st(0)
 # CHECK-NEXT:  1      7     0.50    *             *     flds  (%edx)
 # CHECK-NEXT:  1      7     0.50    *             *     fldl  (%ecx)
 # CHECK-NEXT:  1      7     0.50    *             *     fldt  (%eax)
 # CHECK-NEXT:  3      7     1.00    *             *     fldcw (%eax)
 # CHECK-NEXT:  64     61    14.00                 *     fldenv  (%eax)
-# CHECK-NEXT:  1      1      -                    *     fld1
+# CHECK-NEXT:  2      1     1.00                  *     fld1
 # CHECK-NEXT:  2      1     0.50                  *     fldl2e
 # CHECK-NEXT:  2      1     0.50                  *     fldl2t
 # CHECK-NEXT:  2      1     0.50                  *     fldlg2
 # CHECK-NEXT:  2      1     0.50                  *     fldln2
 # CHECK-NEXT:  2      1     0.50                  *     fldpi
 # CHECK-NEXT:  1      1     0.50                  *     fldz
 # CHECK-NEXT:  1      5     1.00                  *     fmul  %st(0), %st(1)
 # CHECK-NEXT:  1      5     1.00                  *     fmul  %st(2)

 # CHECK-NEXT: [5]   - HWPort3
 # CHECK-NEXT: [6]   - HWPort4
 # CHECK-NEXT: [7]   - HWPort5
 # CHECK-NEXT: [8]   - HWPort6
 # CHECK-NEXT: [9]   - HWPort7
 
 # CHECK:      Resource pressure per iteration:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    [8]    [9]
-# CHECK-NEXT:  -     17.00  117.92 141.92 49.00  49.00  27.00  56.92  65.25  9.00
+# CHECK-NEXT:  -     17.00  118.92 142.92 49.00  49.00  27.00  56.92  65.25  9.00
 
 # CHECK:      Resource pressure by instruction:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    [8]    [9]    Instructions:
 # CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     f2xm1
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -     fabs
 # CHECK-NEXT:  -      -      -     1.00    -      -      -      -      -      -     fadd  %st(0), %st(1)
 # CHECK-NEXT:  -      -      -     1.00    -      -      -      -      -      -     fadd  %st(2)
 # CHECK-NEXT:  -      -      -     1.00   0.50   0.50    -      -      -      -     fadds (%ecx)

 # CHECK-NEXT:  -      -      -     1.00   0.33   0.33   1.00    -      -     0.33   fisttpl (%ecx)
 # CHECK-NEXT:  -      -      -     1.00   0.33   0.33   1.00    -      -     0.33   fisttpll  (%eax)
 # CHECK-NEXT:  -      -     0.50   0.50    -      -      -      -      -      -     fld %st(0)
 # CHECK-NEXT:  -      -      -      -     0.50   0.50    -      -      -      -     flds  (%edx)
 # CHECK-NEXT:  -      -      -      -     0.50   0.50    -      -      -      -     fldl  (%ecx)
 # CHECK-NEXT:  -      -      -      -     0.50   0.50    -      -      -      -     fldt  (%eax)
 # CHECK-NEXT:  -      -     1.50   0.50   0.50   0.50    -      -      -      -     fldcw (%eax)
 # CHECK-NEXT:  -      -     18.92  11.42  4.00   4.00    -     10.92  14.75   -     fldenv  (%eax)
-# CHECK-NEXT:  -      -      -      -      -      -      -      -      -      -     fld1
+# CHECK-NEXT:  -      -     1.00   1.00    -      -      -      -      -      -     fld1
 # CHECK-NEXT:  -      -     0.50   0.50    -      -      -      -      -      -     fldl2e
 # CHECK-NEXT:  -      -     0.50   0.50    -      -      -      -      -      -     fldl2t
 # CHECK-NEXT:  -      -     0.50   0.50    -      -      -      -      -      -     fldlg2
 # CHECK-NEXT:  -      -     0.50   0.50    -      -      -      -      -      -     fldln2
 # CHECK-NEXT:  -      -     0.50   0.50    -      -      -      -      -      -     fldpi
 # CHECK-NEXT:  -      -     0.50   0.50    -      -      -      -      -      -     fldz
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -     fmul  %st(0), %st(1)
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -     fmul  %st(2)

test/tools/llvm-mca/X86/SLM/resources-x87.s

 # CHECK-NEXT:  1      1     1.00           *      *     fisttpl (%ecx)
 # CHECK-NEXT:  1      1     1.00           *      *     fisttpll  (%eax)
 # CHECK-NEXT:  1      1     0.50                  *     fld %st(0)
 # CHECK-NEXT:  1      3     1.00    *             *     flds  (%edx)
 # CHECK-NEXT:  1      3     1.00    *             *     fldl  (%ecx)
 # CHECK-NEXT:  1      3     1.00    *             *     fldt  (%eax)
 # CHECK-NEXT:  1      3     1.00    *             *     fldcw (%eax)
 # CHECK-NEXT:  1      100   1.00                  *     fldenv  (%eax)
-# CHECK-NEXT:  1      1      -                    *     fld1
+# CHECK-NEXT:  1      1     0.50                  *     fld1
 # CHECK-NEXT:  1      100   1.00                  *     fldl2e
 # CHECK-NEXT:  1      100   1.00                  *     fldl2t
 # CHECK-NEXT:  1      100   1.00                  *     fldlg2
 # CHECK-NEXT:  1      100   1.00                  *     fldln2
 # CHECK-NEXT:  1      100   1.00                  *     fldpi
-# CHECK-NEXT:  1      1      -                    *     fldz
+# CHECK-NEXT:  1      1     0.50                  *     fldz
 # CHECK-NEXT:  1      5     2.00                  *     fmul  %st(0), %st(1)
 # CHECK-NEXT:  1      5     2.00                  *     fmul  %st(2)
 # CHECK-NEXT:  1      8     2.00    *             *     fmuls (%ecx)
 # CHECK-NEXT:  1      8     2.00    *             *     fmull (%eax)
 # CHECK-NEXT:  1      5     2.00                  *     fmulp %st(1)
 # CHECK-NEXT:  1      5     2.00                  *     fmulp %st(2)
 # CHECK-NEXT:  1      8     2.00    *             *     fimuls  (%ecx)
 # CHECK-NEXT:  1      8     2.00    *             *     fimull  (%eax)

 # CHECK-NEXT: [3]   - SLM_FPC_RSV0
 # CHECK-NEXT: [4]   - SLM_FPC_RSV1
 # CHECK-NEXT: [5]   - SLM_IEC_RSV0
 # CHECK-NEXT: [6]   - SLM_IEC_RSV1
 # CHECK-NEXT: [7]   - SLM_MEC_RSV
 
 # CHECK:      Resource pressure per iteration:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]
-# CHECK-NEXT:  -     312.00 16.00  64.00  55.00  9.50   9.50   52.00
+# CHECK-NEXT:  -     312.00 16.00  65.00  56.00  9.50   9.50   52.00
 
 # CHECK:      Resource pressure by instruction:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    Instructions:
 # CHECK-NEXT:  -      -      -     1.00    -      -      -      -     f2xm1
 # CHECK-NEXT:  -      -      -     0.50   0.50    -      -      -     fabs
 # CHECK-NEXT:  -      -      -      -     1.00    -      -      -     fadd  %st(0), %st(1)
 # CHECK-NEXT:  -      -      -      -     1.00    -      -      -     fadd  %st(2)
 # CHECK-NEXT:  -      -      -      -     1.00    -      -     1.00   fadds (%ecx)

 # CHECK-NEXT:  -      -      -      -      -     0.50   0.50   1.00   fisttpl (%ecx)
 # CHECK-NEXT:  -      -      -      -      -     0.50   0.50   1.00   fisttpll  (%eax)
 # CHECK-NEXT:  -      -      -      -      -     0.50   0.50    -     fld %st(0)
 # CHECK-NEXT:  -      -      -      -      -      -      -     1.00   flds  (%edx)
 # CHECK-NEXT:  -      -      -      -      -      -      -     1.00   fldl  (%ecx)
 # CHECK-NEXT:  -      -      -      -      -      -      -     1.00   fldt  (%eax)
 # CHECK-NEXT:  -      -      -      -      -      -      -     1.00   fldcw (%eax)
 # CHECK-NEXT:  -      -      -     1.00    -      -      -      -     fldenv  (%eax)
-# CHECK-NEXT:  -      -      -      -      -      -      -      -     fld1
+# CHECK-NEXT:  -      -      -     0.50   0.50    -      -      -     fld1
 # CHECK-NEXT:  -      -      -     1.00    -      -      -      -     fldl2e
 # CHECK-NEXT:  -      -      -     1.00    -      -      -      -     fldl2t
 # CHECK-NEXT:  -      -      -     1.00    -      -      -      -     fldlg2
 # CHECK-NEXT:  -      -      -     1.00    -      -      -      -     fldln2
 # CHECK-NEXT:  -      -      -     1.00    -      -      -      -     fldpi
-# CHECK-NEXT:  -      -      -      -      -      -      -      -     fldz
+# CHECK-NEXT:  -      -      -     0.50   0.50    -      -      -     fldz
 # CHECK-NEXT:  -      -     2.00   1.00    -      -      -      -     fmul  %st(0), %st(1)
 # CHECK-NEXT:  -      -     2.00   1.00    -      -      -      -     fmul  %st(2)
 # CHECK-NEXT:  -      -     2.00   1.00    -      -      -     1.00   fmuls (%ecx)
 # CHECK-NEXT:  -      -     2.00   1.00    -      -      -     1.00   fmull (%eax)
 # CHECK-NEXT:  -      -     2.00   1.00    -      -      -      -     fmulp %st(1)
 # CHECK-NEXT:  -      -     2.00   1.00    -      -      -      -     fmulp %st(2)
 # CHECK-NEXT:  -      -     2.00   1.00    -      -      -     1.00   fimuls  (%ecx)
 # CHECK-NEXT:  -      -     2.00   1.00    -      -      -     1.00   fimull  (%eax)

test/tools/llvm-mca/X86/SandyBridge/resources-x87.s

 # CHECK-NEXT:  3      5     1.00           *      *     fisttpl (%ecx)
 # CHECK-NEXT:  3      5     1.00           *      *     fisttpll  (%eax)
 # CHECK-NEXT:  1      1     1.00                  *     fld %st(0)
 # CHECK-NEXT:  3      9     1.00    *             *     flds  (%edx)
 # CHECK-NEXT:  3      9     1.00    *             *     fldl  (%ecx)
 # CHECK-NEXT:  3      9     1.00    *             *     fldt  (%eax)
 # CHECK-NEXT:  5      8     2.00    *             *     fldcw (%eax)
 # CHECK-NEXT:  1      100   0.33                  *     fldenv  (%eax)
-# CHECK-NEXT:  1      1      -                    *     fld1
+# CHECK-NEXT:  2      1     1.00                  *     fld1
 # CHECK-NEXT:  1      100   0.33                  *     fldl2e
 # CHECK-NEXT:  1      100   0.33                  *     fldl2t
 # CHECK-NEXT:  1      100   0.33                  *     fldlg2
 # CHECK-NEXT:  1      100   0.33                  *     fldln2
 # CHECK-NEXT:  1      100   0.33                  *     fldpi
-# CHECK-NEXT:  1      1      -                    *     fldz
+# CHECK-NEXT:  1      1     1.00                  *     fldz
 # CHECK-NEXT:  1      5     1.00                  *     fmul  %st(0), %st(1)
 # CHECK-NEXT:  1      5     1.00                  *     fmul  %st(2)
 # CHECK-NEXT:  2      12    1.00    *             *     fmuls (%ecx)
 # CHECK-NEXT:  2      12    1.00    *             *     fmull (%eax)
 # CHECK-NEXT:  1      5     1.00                  *     fmulp %st(1)
 # CHECK-NEXT:  1      5     1.00                  *     fmulp %st(2)
 # CHECK-NEXT:  3      15    1.00    *             *     fimuls  (%ecx)
 # CHECK-NEXT:  3      15    1.00    *             *     fimull  (%eax)

 # CHECK-NEXT: [3]   - SBPort1
 # CHECK-NEXT: [4]   - SBPort4
 # CHECK-NEXT: [5]   - SBPort5
 # CHECK-NEXT: [6.0] - SBPort23
 # CHECK-NEXT: [6.1] - SBPort23
 
 # CHECK:      Resource pressure per iteration:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6.0]  [6.1]
-# CHECK-NEXT:  -     136.00 48.33  87.33  17.00  54.33  34.00  34.00
+# CHECK-NEXT:  -     136.00 49.33  87.33  17.00  56.33  34.00  34.00
 
 # CHECK:      Resource pressure by instruction:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6.0]  [6.1]  Instructions:
 # CHECK-NEXT:  -      -     0.33   0.33    -     0.33    -      -     f2xm1
 # CHECK-NEXT:  -      -      -      -      -     1.00    -      -     fabs
 # CHECK-NEXT:  -      -      -     1.00    -      -      -      -     fadd  %st(0), %st(1)
 # CHECK-NEXT:  -      -      -     1.00    -      -      -      -     fadd  %st(2)
 # CHECK-NEXT:  -      -      -     1.00    -      -     0.50   0.50   fadds (%ecx)

 # CHECK-NEXT:  -      -      -     1.00   1.00    -     0.50   0.50   fisttpl (%ecx)
 # CHECK-NEXT:  -      -      -     1.00   1.00    -     0.50   0.50   fisttpll  (%eax)
 # CHECK-NEXT:  -      -      -      -      -     1.00    -      -     fld %st(0)
 # CHECK-NEXT:  -      -     0.50   0.50    -     1.00   0.50   0.50   flds  (%edx)
 # CHECK-NEXT:  -      -     0.50   0.50    -     1.00   0.50   0.50   fldl  (%ecx)
 # CHECK-NEXT:  -      -     0.50   0.50    -     1.00   0.50   0.50   fldt  (%eax)
 # CHECK-NEXT:  -      -      -      -     1.00   2.00   1.00   1.00   fldcw (%eax)
 # CHECK-NEXT:  -      -     0.33   0.33    -     0.33    -      -     fldenv  (%eax)
-# CHECK-NEXT:  -      -      -      -      -      -      -      -     fld1
+# CHECK-NEXT:  -      -     1.00    -      -     1.00    -      -     fld1
 # CHECK-NEXT:  -      -     0.33   0.33    -     0.33    -      -     fldl2e
 # CHECK-NEXT:  -      -     0.33   0.33    -     0.33    -      -     fldl2t
 # CHECK-NEXT:  -      -     0.33   0.33    -     0.33    -      -     fldlg2
 # CHECK-NEXT:  -      -     0.33   0.33    -     0.33    -      -     fldln2
 # CHECK-NEXT:  -      -     0.33   0.33    -     0.33    -      -     fldpi
-# CHECK-NEXT:  -      -      -      -      -      -      -      -     fldz
+# CHECK-NEXT:  -      -      -      -      -     1.00    -      -     fldz
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -     fmul  %st(0), %st(1)
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -     fmul  %st(2)
 # CHECK-NEXT:  -      -     1.00    -      -      -     0.50   0.50   fmuls (%ecx)
 # CHECK-NEXT:  -      -     1.00    -      -      -     0.50   0.50   fmull (%eax)
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -     fmulp %st(1)
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -     fmulp %st(2)
 # CHECK-NEXT:  -      -     1.00   1.00    -      -     0.50   0.50   fimuls  (%ecx)
 # CHECK-NEXT:  -      -     1.00   1.00    -      -     0.50   0.50   fimull  (%eax)

test/tools/llvm-mca/X86/SkylakeClient/resources-x87.s

 # CHECK-NEXT:  3      4     1.00           *      *     fisttpl (%ecx)
 # CHECK-NEXT:  3      4     1.00           *      *     fisttpll  (%eax)
 # CHECK-NEXT:  1      1     0.25                  *     fld %st(0)
 # CHECK-NEXT:  1      7     0.50    *             *     flds  (%edx)
 # CHECK-NEXT:  1      7     0.50    *             *     fldl  (%ecx)
 # CHECK-NEXT:  1      7     0.50    *             *     fldt  (%eax)
 # CHECK-NEXT:  3      7     1.00    *             *     fldcw (%eax)
 # CHECK-NEXT:  64     62    14.00                 *     fldenv  (%eax)
-# CHECK-NEXT:  1      1      -                    *     fld1
+# CHECK-NEXT:  2      1     1.00                  *     fld1
 # CHECK-NEXT:  1      100   0.25                  *     fldl2e
 # CHECK-NEXT:  1      100   0.25                  *     fldl2t
 # CHECK-NEXT:  1      100   0.25                  *     fldlg2
 # CHECK-NEXT:  1      100   0.25                  *     fldln2
 # CHECK-NEXT:  1      100   0.25                  *     fldpi
-# CHECK-NEXT:  1      1      -                    *     fldz
+# CHECK-NEXT:  1      1     0.50                  *     fldz
 # CHECK-NEXT:  1      4     1.00                  *     fmul  %st(0), %st(1)
 # CHECK-NEXT:  1      4     1.00                  *     fmul  %st(2)
 # CHECK-NEXT:  2      11    1.00    *             *     fmuls (%ecx)
 # CHECK-NEXT:  2      11    1.00    *             *     fmull (%eax)
 # CHECK-NEXT:  1      4     1.00                  *     fmulp %st(1)
 # CHECK-NEXT:  1      4     1.00                  *     fmulp %st(2)
 # CHECK-NEXT:  3      14    1.00    *             *     fimuls  (%ecx)
 # CHECK-NEXT:  3      14    1.00    *             *     fimull  (%eax)

 # CHECK-NEXT: [5]   - SKLPort3
 # CHECK-NEXT: [6]   - SKLPort4
 # CHECK-NEXT: [7]   - SKLPort5
 # CHECK-NEXT: [8]   - SKLPort6
 # CHECK-NEXT: [9]   - SKLPort7
 
 # CHECK:      Resource pressure per iteration:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    [8]    [9]
-# CHECK-NEXT:  -     7.00   121.50 53.50  49.00  49.00  27.00  144.50 70.50  9.00
+# CHECK-NEXT:  -     7.00   123.00 53.50  49.00  49.00  27.00  146.00 70.50  9.00
 
 # CHECK:      Resource pressure by instruction:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    [8]    [9]    Instructions:
 # CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     f2xm1
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -     fabs
 # CHECK-NEXT:  -      -      -      -      -      -      -     1.00    -      -     fadd  %st(0), %st(1)
 # CHECK-NEXT:  -      -      -      -      -      -      -     1.00    -      -     fadd  %st(2)
 # CHECK-NEXT:  -      -      -      -     0.50   0.50    -     1.00    -      -     fadds (%ecx)

 # CHECK-NEXT:  -      -      -      -     0.33   0.33   1.00   1.00    -     0.33   fisttpl (%ecx)
 # CHECK-NEXT:  -      -      -      -     0.33   0.33   1.00   1.00    -     0.33   fisttpll  (%eax)
 # CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     fld %st(0)
 # CHECK-NEXT:  -      -      -      -     0.50   0.50    -      -      -      -     flds  (%edx)
 # CHECK-NEXT:  -      -      -      -     0.50   0.50    -      -      -      -     fldl  (%ecx)
 # CHECK-NEXT:  -      -      -      -     0.50   0.50    -      -      -      -     fldt  (%eax)
 # CHECK-NEXT:  -      -     1.50    -     0.50   0.50    -     0.50    -      -     fldcw (%eax)
 # CHECK-NEXT:  -      -     19.25  9.75   4.00   4.00    -     12.25  14.75   -     fldenv  (%eax)
-# CHECK-NEXT:  -      -      -      -      -      -      -      -      -      -     fld1
+# CHECK-NEXT:  -      -     1.00    -      -      -      -     1.00    -      -     fld1
 # CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     fldl2e
 # CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     fldl2t
 # CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     fldlg2
 # CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     fldln2
 # CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     fldpi
-# CHECK-NEXT:  -      -      -      -      -      -      -      -      -      -     fldz
+# CHECK-NEXT:  -      -     0.50    -      -      -      -     0.50    -      -     fldz
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -     fmul  %st(0), %st(1)
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -     fmul  %st(2)
 # CHECK-NEXT:  -      -     1.00    -     0.50   0.50    -      -      -      -     fmuls (%ecx)
 # CHECK-NEXT:  -      -     1.00    -     0.50   0.50    -      -      -      -     fmull (%eax)
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -     fmulp %st(1)
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -     fmulp %st(2)
 # CHECK-NEXT:  -      -     1.00    -     0.50   0.50    -     1.00    -      -     fimuls  (%ecx)
 # CHECK-NEXT:  -      -     1.00    -     0.50   0.50    -     1.00    -      -     fimull  (%eax)

test/tools/llvm-mca/X86/SkylakeServer/resources-x87.s

 # CHECK-NEXT:  3      4     1.00           *      *     fisttpl (%ecx)
 # CHECK-NEXT:  3      4     1.00           *      *     fisttpll  (%eax)
 # CHECK-NEXT:  1      1     0.25                  *     fld %st(0)
 # CHECK-NEXT:  1      7     0.50    *             *     flds  (%edx)
 # CHECK-NEXT:  1      7     0.50    *             *     fldl  (%ecx)
 # CHECK-NEXT:  1      7     0.50    *             *     fldt  (%eax)
 # CHECK-NEXT:  3      7     1.00    *             *     fldcw (%eax)
 # CHECK-NEXT:  64     62    14.00                 *     fldenv  (%eax)
-# CHECK-NEXT:  1      1      -                    *     fld1
+# CHECK-NEXT:  2      1     1.00                  *     fld1
 # CHECK-NEXT:  1      100   0.25                  *     fldl2e
 # CHECK-NEXT:  1      100   0.25                  *     fldl2t
 # CHECK-NEXT:  1      100   0.25                  *     fldlg2
 # CHECK-NEXT:  1      100   0.25                  *     fldln2
 # CHECK-NEXT:  1      100   0.25                  *     fldpi
-# CHECK-NEXT:  1      1      -                    *     fldz
+# CHECK-NEXT:  1      1     0.50                  *     fldz
 # CHECK-NEXT:  1      4     1.00                  *     fmul  %st(0), %st(1)
 # CHECK-NEXT:  1      4     1.00                  *     fmul  %st(2)
 # CHECK-NEXT:  2      11    1.00    *             *     fmuls (%ecx)
 # CHECK-NEXT:  2      11    1.00    *             *     fmull (%eax)
 # CHECK-NEXT:  1      4     1.00                  *     fmulp %st(1)
 # CHECK-NEXT:  1      4     1.00                  *     fmulp %st(2)
 # CHECK-NEXT:  3      14    1.00    *             *     fimuls  (%ecx)
 # CHECK-NEXT:  3      14    1.00    *             *     fimull  (%eax)

 # CHECK-NEXT: [5]   - SKXPort3
 # CHECK-NEXT: [6]   - SKXPort4
 # CHECK-NEXT: [7]   - SKXPort5
 # CHECK-NEXT: [8]   - SKXPort6
 # CHECK-NEXT: [9]   - SKXPort7
 
 # CHECK:      Resource pressure per iteration:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    [8]    [9]
-# CHECK-NEXT:  -     7.00   121.50 53.50  49.00  49.00  27.00  144.50 70.50  9.00
+# CHECK-NEXT:  -     7.00   123.00 53.50  49.00  49.00  27.00  146.00 70.50  9.00
 
 # CHECK:      Resource pressure by instruction:
 # CHECK-NEXT: [0]    [1]    [2]    [3]    [4]    [5]    [6]    [7]    [8]    [9]    Instructions:
 # CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     f2xm1
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -     fabs
 # CHECK-NEXT:  -      -      -      -      -      -      -     1.00    -      -     fadd  %st(0), %st(1)
 # CHECK-NEXT:  -      -      -      -      -      -      -     1.00    -      -     fadd  %st(2)
 # CHECK-NEXT:  -      -      -      -     0.50   0.50    -     1.00    -      -     fadds (%ecx)

 # CHECK-NEXT:  -      -      -      -     0.33   0.33   1.00   1.00    -     0.33   fisttpl (%ecx)
 # CHECK-NEXT:  -      -      -      -     0.33   0.33   1.00   1.00    -     0.33   fisttpll  (%eax)
 # CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     fld %st(0)
 # CHECK-NEXT:  -      -      -      -     0.50   0.50    -      -      -      -     flds  (%edx)
 # CHECK-NEXT:  -      -      -      -     0.50   0.50    -      -      -      -     fldl  (%ecx)
 # CHECK-NEXT:  -      -      -      -     0.50   0.50    -      -      -      -     fldt  (%eax)
 # CHECK-NEXT:  -      -     1.50    -     0.50   0.50    -     0.50    -      -     fldcw (%eax)
 # CHECK-NEXT:  -      -     19.25  9.75   4.00   4.00    -     12.25  14.75   -     fldenv  (%eax)
-# CHECK-NEXT:  -      -      -      -      -      -      -      -      -      -     fld1
+# CHECK-NEXT:  -      -     1.00    -      -      -      -     1.00    -      -     fld1
 # CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     fldl2e
 # CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     fldl2t
 # CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     fldlg2
 # CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     fldln2
 # CHECK-NEXT:  -      -     0.25   0.25    -      -      -     0.25   0.25    -     fldpi
-# CHECK-NEXT:  -      -      -      -      -      -      -      -      -      -     fldz
+# CHECK-NEXT:  -      -     0.50    -      -      -      -     0.50    -      -     fldz
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -     fmul  %st(0), %st(1)
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -     fmul  %st(2)
 # CHECK-NEXT:  -      -     1.00    -     0.50   0.50    -      -      -      -     fmuls (%ecx)
 # CHECK-NEXT:  -      -     1.00    -     0.50   0.50    -      -      -      -     fmull (%eax)
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -     fmulp %st(1)
 # CHECK-NEXT:  -      -     1.00    -      -      -      -      -      -      -     fmulp %st(2)
 # CHECK-NEXT:  -      -     1.00    -     0.50   0.50    -     1.00    -      -     fimuls  (%ecx)
 # CHECK-NEXT:  -      -     1.00    -     0.50   0.50    -     1.00    -      -     fimull  (%eax)

test/tools/llvm-mca/X86/Znver1/resources-x87.s

 # CHECK-NEXT:  1      12    0.50           *      *     fisttpl (%ecx)
 # CHECK-NEXT:  1      12    0.50           *      *     fisttpll  (%eax)
 # CHECK-NEXT:  1      1     0.50                  *     fld %st(0)
 # CHECK-NEXT:  1      8     0.50    *             *     flds  (%edx)
 # CHECK-NEXT:  1      8     0.50    *             *     fldl  (%ecx)
 # CHECK-NEXT:  2      1     0.50    *             *     fldt  (%eax)
 # CHECK-NEXT:  1      100    -      *             *     fldcw (%eax)
 # CHECK-NEXT:  1      100    -                    *     fldenv  (%eax)
-# CHECK-NEXT:  1      11    1.00                  *     fld1
+# CHECK-NEXT:  2      11    1.00                  *     fld1
 # CHECK-NEXT:  1      11    1.00                  *     fldl2e
 # CHECK-NEXT:  1      11    1.00                  *     fldl2t
 # CHECK-NEXT:  1      11    1.00                  *     fldlg2
 # CHECK-NEXT:  1      11    1.00                  *     fldln2
 # CHECK-NEXT:  1      11    1.00                  *     fldpi
-# CHECK-NEXT:  1      8     0.50                  *     fldz
+# CHECK-NEXT:  2      8     0.50                  *     fldz
 # CHECK-NEXT:  1      3     0.50                  *     fmul  %st(0), %st(1)
 # CHECK-NEXT:  1      3     0.50                  *     fmul  %st(2)
 # CHECK-NEXT:  2      10    0.50    *             *     fmuls (%ecx)
 # CHECK-NEXT:  2      10    0.50    *             *     fmull (%eax)
 # CHECK-NEXT:  1      3     0.50                  *     fmulp %st(1)
 # CHECK-NEXT:  1      3     0.50                  *     fmulp %st(2)
 # CHECK-NEXT:  2      10    0.50    *             *     fimuls  (%ecx)
 # CHECK-NEXT:  2      10    0.50    *             *     fimull  (%eax)