[PowerPC] Respect SDNodeFlags in lowering SELECT_CC

[qiucf]

Summary

Legalizer should respect both command-line options or SDNode-level fast-math flags.

Also, this patch propagates other flags during custom simplifying.

[qiucf]

llvm/test/CodeGen/PowerPC/scalar-min-max.ll

129

Deleting fast in this file looks okay to me, since we already used command-line flags. We can clear this in following patches.

[qiucf]

Ping..

[steven.zhang]

Can we not to remove the --enable-unsafe-fp-math in this patch ? And I think, the option --enable-unsafe-fp-math is also part of our test point.

[qiucf]

Not remove --enable-unsafe-fp-math

[qiucf]

Keep flags and options as-is, so that we can have a clear sight of what changed. Options/flags will be updated in child revision.

[steven.zhang]

LGTM now.

llvm/test/CodeGen/PowerPC/scalar_cmp.ll

129

These instructions sequence seem terrible to me, as we didn't have INF and NAN for this case. But it is not relative with your change, maybe, worth to take a look.

[qiucf]

llvm/test/CodeGen/PowerPC/scalar_cmp.ll

129

hmm.. This is expected result since we propagate these flags to SELECT_CC now. Do you mean that sub-sub here is worse than cmp-beq-fmr?

[steven.zhang]

llvm/test/CodeGen/PowerPC/scalar_cmp.ll

129

I think so, we have two subs + two fsel. Not sure why we do it that way.

llvm/lib/Target/PowerPC/PPCISelLowering.cpp

 
   ISD::CondCode CC = cast<CondCodeSDNode>(Op.getOperand(4))->get();
 
   EVT ResVT = Op.getValueType();
   EVT CmpVT = Op.getOperand(0).getValueType();
   SDValue LHS = Op.getOperand(0), RHS = Op.getOperand(1);
   SDValue TV  = Op.getOperand(2), FV  = Op.getOperand(3);
   SDLoc dl(Op);
+  SDNodeFlags Flags = Op.getNode()->getFlags();
 
   // We have xsmaxcdp/xsmincdp which are OK to emit even in the
   // presence of infinities.
   if (Subtarget.hasP9Vector() && LHS == TV && RHS == FV) {
     switch (CC) {
     default:
       break;
     case ISD::SETOGT:
     case ISD::SETGT:
       return DAG.getNode(PPCISD::XSMAXCDP, dl, Op.getValueType(), LHS, RHS);
     case ISD::SETOLT:
     case ISD::SETLT:
       return DAG.getNode(PPCISD::XSMINCDP, dl, Op.getValueType(), LHS, RHS);
     }
   }
 
   // We might be able to do better than this under some circumstances, but in
   // general, fsel-based lowering of select is a finite-math-only optimization.
   // For more information, see section F.3 of the 2.06 ISA specification.
   // With ISA 3.0
-  if (!DAG.getTarget().Options.NoInfsFPMath ||
-      !DAG.getTarget().Options.NoNaNsFPMath)
+  if ((!DAG.getTarget().Options.NoInfsFPMath && !Flags.hasNoInfs()) ||
+      (!DAG.getTarget().Options.NoNaNsFPMath && !Flags.hasNoNaNs()))
     return Op;
 
-  // TODO: Propagate flags from the select rather than global settings.
-  SDNodeFlags Flags;
-  Flags.setNoInfs(true);
-  Flags.setNoNaNs(true);
-
   // If the RHS of the comparison is a 0.0, we don't need to do the
   // subtraction at all.
   SDValue Sel1;
   if (isFloatingPointZero(RHS))
     switch (CC) {
     default: break;       // SETUO etc aren't handled by fsel.
     case ISD::SETNE:
       std::swap(TV, FV);

llvm/test/CodeGen/PowerPC/scalar-equal.ll

 ; NO-FAST-P8:       # %bb.0: # %entry
 ; NO-FAST-P8-NEXT:    xscmpudp cr0, f1, f2
 ; NO-FAST-P8-NEXT:    fmr f1, f3
 ; NO-FAST-P8-NEXT:    beqlr cr0
 ; NO-FAST-P8-NEXT:  # %bb.1: # %entry
 ; NO-FAST-P8-NEXT:    fmr f1, f4
 ; NO-FAST-P8-NEXT:    blr
 entry:
+  %cmp = fcmp oeq double %a, %b
+  %cond = select i1 %cmp, double %c, double %d
+  ret double %cond
+}
+
+define double @testoeq_fast(double %a, double %b, double %c, double %d) {
+; FAST-P8-LABEL: testoeq_fast:
+; FAST-P8:       # %bb.0: # %entry
+; FAST-P8-NEXT:    xssubdp f0, f1, f2
+; FAST-P8-NEXT:    xsnegdp f1, f0
+; FAST-P8-NEXT:    fsel f0, f0, f3, f4
+; FAST-P8-NEXT:    fsel f1, f1, f0, f4
+; FAST-P8-NEXT:    blr
+;
+; FAST-P9-LABEL: testoeq_fast:
+; FAST-P9:       # %bb.0: # %entry
+; FAST-P9-NEXT:    xssubdp f0, f1, f2
+; FAST-P9-NEXT:    fsel f1, f0, f3, f4
+; FAST-P9-NEXT:    xsnegdp f0, f0
+; FAST-P9-NEXT:    fsel f1, f0, f1, f4
+; FAST-P9-NEXT:    blr
+;
+; NO-FAST-P9-LABEL: testoeq_fast:
+; NO-FAST-P9:       # %bb.0: # %entry
+; NO-FAST-P9-NEXT:    xssubdp f0, f1, f2
+; NO-FAST-P9-NEXT:    fsel f1, f0, f3, f4
+; NO-FAST-P9-NEXT:    xsnegdp f0, f0
+; NO-FAST-P9-NEXT:    fsel f1, f0, f1, f4
+; NO-FAST-P9-NEXT:    blr
+;
+; NO-FAST-P8-LABEL: testoeq_fast:
+; NO-FAST-P8:       # %bb.0: # %entry
+; NO-FAST-P8-NEXT:    xssubdp f0, f1, f2
+; NO-FAST-P8-NEXT:    xsnegdp f1, f0
+; NO-FAST-P8-NEXT:    fsel f0, f0, f3, f4
+; NO-FAST-P8-NEXT:    fsel f1, f1, f0, f4
+; NO-FAST-P8-NEXT:    blr
+entry:
   %cmp = fcmp fast oeq double %a, %b
   %cond = select fast i1 %cmp, double %c, double %d
   ret double %cond
 }

llvm/test/CodeGen/PowerPC/scalar-min-max.ll

 ;
 ; NO-FAST-P9-LABEL: testfmax_fast:
 ; NO-FAST-P9:       # %bb.0: # %entry
 ; NO-FAST-P9-NEXT:    xsmaxcdp f1, f1, f2
 ; NO-FAST-P9-NEXT:    blr
 ;
 ; NO-FAST-P8-LABEL: testfmax_fast:
 ; NO-FAST-P8:       # %bb.0: # %entry
-; NO-FAST-P8-NEXT:    fcmpu cr0, f1, f2
-; NO-FAST-P8-NEXT:    bgtlr cr0
-; NO-FAST-P8-NEXT:  # %bb.1: # %entry
-; NO-FAST-P8-NEXT:    fmr f1, f2
+; NO-FAST-P8-NEXT:    xssubsp f0, f2, f1
+; NO-FAST-P8-NEXT:    fsel f1, f0, f2, f1
 ; NO-FAST-P8-NEXT:    blr
 entry:
   %cmp = fcmp fast ogt float %a, %b
   %cond = select i1 %cmp, float %a, float %b
   ret float %cond
 }
 define dso_local double @testdmax_fast(double %a, double %b) local_unnamed_addr {
 ; CHECK-LABEL: testdmax_fast:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    xsmaxdp f1, f1, f2
 ; CHECK-NEXT:    blr
 ;
 ; NO-FAST-P9-LABEL: testdmax_fast:
 ; NO-FAST-P9:       # %bb.0: # %entry
 ; NO-FAST-P9-NEXT:    xsmaxcdp f1, f1, f2
 ; NO-FAST-P9-NEXT:    blr
 ;
 ; NO-FAST-P8-LABEL: testdmax_fast:
 ; NO-FAST-P8:       # %bb.0: # %entry
-; NO-FAST-P8-NEXT:    xscmpudp cr0, f1, f2
-; NO-FAST-P8-NEXT:    bgtlr cr0
-; NO-FAST-P8-NEXT:  # %bb.1: # %entry
-; NO-FAST-P8-NEXT:    fmr f1, f2
+; NO-FAST-P8-NEXT:    xssubdp f0, f2, f1
+; NO-FAST-P8-NEXT:    fsel f1, f0, f2, f1
 ; NO-FAST-P8-NEXT:    blr
 entry:
   %cmp = fcmp fast ogt double %a, %b
   %cond = select i1 %cmp, double %a, double %b
   ret double %cond
 }
 define dso_local float @testfmin_fast(float %a, float %b) local_unnamed_addr {
 ; CHECK-LABEL: testfmin_fast:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    xsmindp f1, f1, f2
 ; CHECK-NEXT:    blr
 ;
 ; NO-FAST-P9-LABEL: testfmin_fast:
 ; NO-FAST-P9:       # %bb.0: # %entry
 ; NO-FAST-P9-NEXT:    xsmincdp f1, f1, f2
 ; NO-FAST-P9-NEXT:    blr
 ;
 ; NO-FAST-P8-LABEL: testfmin_fast:
 ; NO-FAST-P8:       # %bb.0: # %entry
-; NO-FAST-P8-NEXT:    fcmpu cr0, f1, f2
-; NO-FAST-P8-NEXT:    bltlr cr0
-; NO-FAST-P8-NEXT:  # %bb.1: # %entry
-; NO-FAST-P8-NEXT:    fmr f1, f2
+; NO-FAST-P8-NEXT:    xssubsp f0, f1, f2
+; NO-FAST-P8-NEXT:    fsel f1, f0, f2, f1
 ; NO-FAST-P8-NEXT:    blr
 entry:
   %cmp = fcmp fast olt float %a, %b
   %cond = select i1 %cmp, float %a, float %b
   ret float %cond
 }
 define dso_local double @testdmin_fast(double %a, double %b) local_unnamed_addr {
 ; CHECK-LABEL: testdmin_fast:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    xsmindp f1, f1, f2
 ; CHECK-NEXT:    blr
 ;
 ; NO-FAST-P9-LABEL: testdmin_fast:
 ; NO-FAST-P9:       # %bb.0: # %entry
 ; NO-FAST-P9-NEXT:    xsmincdp f1, f1, f2
 ; NO-FAST-P9-NEXT:    blr
 ;
 ; NO-FAST-P8-LABEL: testdmin_fast:
 ; NO-FAST-P8:       # %bb.0: # %entry
-; NO-FAST-P8-NEXT:    xscmpudp cr0, f1, f2
-; NO-FAST-P8-NEXT:    bltlr cr0
-; NO-FAST-P8-NEXT:  # %bb.1: # %entry
-; NO-FAST-P8-NEXT:    fmr f1, f2
+; NO-FAST-P8-NEXT:    xssubdp f0, f1, f2
+; NO-FAST-P8-NEXT:    fsel f1, f0, f2, f1
 ; NO-FAST-P8-NEXT:    blr
 entry:
   %cmp = fcmp fast olt double %a, %b
   %cond = select i1 %cmp, double %a, double %b
   ret double %cond
 }

llvm/test/CodeGen/PowerPC/scalar_cmp.ll

 ; FAST-P9-NEXT:    xssubsp f0, f2, f1
 ; FAST-P9-NEXT:    xssubsp f1, f1, f2
 ; FAST-P9-NEXT:    fsel f1, f1, f3, f4
 ; FAST-P9-NEXT:    fsel f1, f0, f1, f4
 ; FAST-P9-NEXT:    blr
 ;
 ; NO-FAST-P8-LABEL: select_fast_oeq_float:
 ; NO-FAST-P8:       # %bb.0: # %entry
-; NO-FAST-P8-NEXT:    fcmpu cr0, f1, f2
-; NO-FAST-P8-NEXT:    fmr f1, f3
-; NO-FAST-P8-NEXT:    beqlr cr0
-; NO-FAST-P8-NEXT:  # %bb.1: # %entry
-; NO-FAST-P8-NEXT:    fmr f1, f4
+; NO-FAST-P8-NEXT:    xssubsp f0, f1, f2
+; NO-FAST-P8-NEXT:    xssubsp f1, f2, f1
+; NO-FAST-P8-NEXT:    fsel f0, f0, f3, f4
+; NO-FAST-P8-NEXT:    fsel f1, f1, f0, f4
 ; NO-FAST-P8-NEXT:    blr
 ;
 ; NO-FAST-P9-LABEL: select_fast_oeq_float:
 ; NO-FAST-P9:       # %bb.0: # %entry
-; NO-FAST-P9-NEXT:    fcmpu cr0, f1, f2
-; NO-FAST-P9-NEXT:    beq cr0, .LBB2_2
-; NO-FAST-P9-NEXT:  # %bb.1: # %entry
-; NO-FAST-P9-NEXT:    fmr f3, f4
-; NO-FAST-P9-NEXT:  .LBB2_2: # %entry
-; NO-FAST-P9-NEXT:    fmr f1, f3
+; NO-FAST-P9-NEXT:    xssubsp f0, f2, f1
+; NO-FAST-P9-NEXT:    xssubsp f1, f1, f2
+; NO-FAST-P9-NEXT:    fsel f1, f1, f3, f4
+; NO-FAST-P9-NEXT:    fsel f1, f0, f1, f4
 ; NO-FAST-P9-NEXT:    blr
 entry:
   %cmp = fcmp fast oeq float %a, %b
   %cond = select fast i1 %cmp, float %c, float %d
   ret float %cond
 }
 
 define double @select_fast_oeq_double(double %a, double %b, double %c, double %d) {

 ; FAST-P9-NEXT:    xssubdp f0, f1, f2
 ; FAST-P9-NEXT:    fsel f1, f0, f3, f4
 ; FAST-P9-NEXT:    xsnegdp f0, f0
 ; FAST-P9-NEXT:    fsel f1, f0, f1, f4
 ; FAST-P9-NEXT:    blr
 ;
 ; NO-FAST-P8-LABEL: select_fast_oeq_double:
 ; NO-FAST-P8:       # %bb.0: # %entry
-; NO-FAST-P8-NEXT:    xscmpudp cr0, f1, f2
-; NO-FAST-P8-NEXT:    fmr f1, f3
-; NO-FAST-P8-NEXT:    beqlr cr0
-; NO-FAST-P8-NEXT:  # %bb.1: # %entry
-; NO-FAST-P8-NEXT:    fmr f1, f4
+; NO-FAST-P8-NEXT:    xssubdp f0, f1, f2
+; NO-FAST-P8-NEXT:    xsnegdp f1, f0
+; NO-FAST-P8-NEXT:    fsel f0, f0, f3, f4
+; NO-FAST-P8-NEXT:    fsel f1, f1, f0, f4
 ; NO-FAST-P8-NEXT:    blr
 ;
 ; NO-FAST-P9-LABEL: select_fast_oeq_double:
 ; NO-FAST-P9:       # %bb.0: # %entry
-; NO-FAST-P9-NEXT:    xscmpudp cr0, f1, f2
-; NO-FAST-P9-NEXT:    beq cr0, .LBB3_2
-; NO-FAST-P9-NEXT:  # %bb.1: # %entry
-; NO-FAST-P9-NEXT:    fmr f3, f4
-; NO-FAST-P9-NEXT:  .LBB3_2: # %entry
-; NO-FAST-P9-NEXT:    fmr f1, f3
+; NO-FAST-P9-NEXT:    xssubdp f0, f1, f2
+; NO-FAST-P9-NEXT:    fsel f1, f0, f3, f4
+; NO-FAST-P9-NEXT:    xsnegdp f0, f0
+; NO-FAST-P9-NEXT:    fsel f1, f0, f1, f4
 ; NO-FAST-P9-NEXT:    blr
 entry:
   %cmp = fcmp fast oeq double %a, %b
   %cond = select fast i1 %cmp, double %c, double %d
   ret double %cond
 }
 
 

 ; FAST-P9-NEXT:    xssubsp f0, f2, f1
 ; FAST-P9-NEXT:    xssubsp f1, f1, f2
 ; FAST-P9-NEXT:    fsel f1, f1, f4, f3
 ; FAST-P9-NEXT:    fsel f1, f0, f1, f3
 ; FAST-P9-NEXT:    blr
 ;
 ; NO-FAST-P8-LABEL: select_fast_one_float:
 ; NO-FAST-P8:       # %bb.0: # %entry
-; NO-FAST-P8-NEXT:    fcmpu cr0, f1, f2
-; NO-FAST-P8-NEXT:    fmr f1, f3
-; NO-FAST-P8-NEXT:    bnelr cr0
-; NO-FAST-P8-NEXT:  # %bb.1: # %entry
-; NO-FAST-P8-NEXT:    fmr f1, f4
+; NO-FAST-P8-NEXT:    xssubsp f0, f1, f2
+; NO-FAST-P8-NEXT:    xssubsp f1, f2, f1
+; NO-FAST-P8-NEXT:    fsel f0, f0, f4, f3
+; NO-FAST-P8-NEXT:    fsel f1, f1, f0, f3
 ; NO-FAST-P8-NEXT:    blr
 ;
 ; NO-FAST-P9-LABEL: select_fast_one_float:
 ; NO-FAST-P9:       # %bb.0: # %entry
-; NO-FAST-P9-NEXT:    fcmpu cr0, f1, f2
-; NO-FAST-P9-NEXT:    bne cr0, .LBB6_2
-; NO-FAST-P9-NEXT:  # %bb.1: # %entry
-; NO-FAST-P9-NEXT:    fmr f3, f4
-; NO-FAST-P9-NEXT:  .LBB6_2: # %entry
-; NO-FAST-P9-NEXT:    fmr f1, f3
+; NO-FAST-P9-NEXT:    xssubsp f0, f2, f1
+; NO-FAST-P9-NEXT:    xssubsp f1, f1, f2
+; NO-FAST-P9-NEXT:    fsel f1, f1, f4, f3
+; NO-FAST-P9-NEXT:    fsel f1, f0, f1, f3
 ; NO-FAST-P9-NEXT:    blr
 entry:
   %cmp = fcmp fast one float %a, %b
   %cond = select fast i1 %cmp, float %c, float %d
   ret float %cond
 }
 
 define double @select_fast_one_double(double %a, double %b, double %c, double %d) {

 ; FAST-P9-NEXT:    xssubdp f0, f1, f2
 ; FAST-P9-NEXT:    fsel f1, f0, f4, f3
 ; FAST-P9-NEXT:    xsnegdp f0, f0
 ; FAST-P9-NEXT:    fsel f1, f0, f1, f3
 ; FAST-P9-NEXT:    blr
 ;
 ; NO-FAST-P8-LABEL: select_fast_one_double:
 ; NO-FAST-P8:       # %bb.0: # %entry
-; NO-FAST-P8-NEXT:    xscmpudp cr0, f1, f2
-; NO-FAST-P8-NEXT:    fmr f1, f3
-; NO-FAST-P8-NEXT:    bnelr cr0
-; NO-FAST-P8-NEXT:  # %bb.1: # %entry
-; NO-FAST-P8-NEXT:    fmr f1, f4
+; NO-FAST-P8-NEXT:    xssubdp f0, f1, f2
+; NO-FAST-P8-NEXT:    xsnegdp f1, f0
+; NO-FAST-P8-NEXT:    fsel f0, f0, f4, f3
+; NO-FAST-P8-NEXT:    fsel f1, f1, f0, f3
 ; NO-FAST-P8-NEXT:    blr
 ;
 ; NO-FAST-P9-LABEL: select_fast_one_double:
 ; NO-FAST-P9:       # %bb.0: # %entry
-; NO-FAST-P9-NEXT:    xscmpudp cr0, f1, f2
-; NO-FAST-P9-NEXT:    bne cr0, .LBB7_2
-; NO-FAST-P9-NEXT:  # %bb.1: # %entry
-; NO-FAST-P9-NEXT:    fmr f3, f4
-; NO-FAST-P9-NEXT:  .LBB7_2: # %entry
-; NO-FAST-P9-NEXT:    fmr f1, f3
+; NO-FAST-P9-NEXT:    xssubdp f0, f1, f2
+; NO-FAST-P9-NEXT:    fsel f1, f0, f4, f3
+; NO-FAST-P9-NEXT:    xsnegdp f0, f0
+; NO-FAST-P9-NEXT:    fsel f1, f0, f1, f3
 ; NO-FAST-P9-NEXT:    blr
 entry:
   %cmp = fcmp fast one double %a, %b
   %cond = select fast i1 %cmp, double %c, double %d
   ret double %cond
 }
 
 

 ; FAST-P9-LABEL: select_fast_oge_float:
 ; FAST-P9:       # %bb.0: # %entry
 ; FAST-P9-NEXT:    xssubsp f0, f1, f2
 ; FAST-P9-NEXT:    fsel f1, f0, f3, f4
 ; FAST-P9-NEXT:    blr
 ;
 ; NO-FAST-P8-LABEL: select_fast_oge_float:
 ; NO-FAST-P8:       # %bb.0: # %entry
-; NO-FAST-P8-NEXT:    fcmpu cr0, f1, f2
-; NO-FAST-P8-NEXT:    fmr f1, f3
-; NO-FAST-P8-NEXT:    bgelr cr0
-; NO-FAST-P8-NEXT:  # %bb.1: # %entry
-; NO-FAST-P8-NEXT:    fmr f1, f4
+; NO-FAST-P8-NEXT:    xssubsp f0, f1, f2
+; NO-FAST-P8-NEXT:    fsel f1, f0, f3, f4
 ; NO-FAST-P8-NEXT:    blr
 ;
 ; NO-FAST-P9-LABEL: select_fast_oge_float:
 ; NO-FAST-P9:       # %bb.0: # %entry
-; NO-FAST-P9-NEXT:    fcmpu cr0, f1, f2
-; NO-FAST-P9-NEXT:    bge cr0, .LBB10_2
-; NO-FAST-P9-NEXT:  # %bb.1: # %entry
-; NO-FAST-P9-NEXT:    fmr f3, f4
-; NO-FAST-P9-NEXT:  .LBB10_2: # %entry
-; NO-FAST-P9-NEXT:    fmr f1, f3
+; NO-FAST-P9-NEXT:    xssubsp f0, f1, f2
+; NO-FAST-P9-NEXT:    fsel f1, f0, f3, f4
 ; NO-FAST-P9-NEXT:    blr
 entry:
   %cmp = fcmp fast oge float %a, %b
   %cond = select fast i1 %cmp, float %c, float %d
   ret float %cond
 }
 
 define double @select_fast_oge_double(double %a, double %b, double %c, double %d) {
 ; FAST-P8-LABEL: select_fast_oge_double:
 ; FAST-P8:       # %bb.0: # %entry
 ; FAST-P8-NEXT:    xssubdp f0, f1, f2
 ; FAST-P8-NEXT:    fsel f1, f0, f3, f4
 ; FAST-P8-NEXT:    blr
 ;
 ; FAST-P9-LABEL: select_fast_oge_double:
 ; FAST-P9:       # %bb.0: # %entry
 ; FAST-P9-NEXT:    xssubdp f0, f1, f2
 ; FAST-P9-NEXT:    fsel f1, f0, f3, f4
 ; FAST-P9-NEXT:    blr
 ;
 ; NO-FAST-P8-LABEL: select_fast_oge_double:
 ; NO-FAST-P8:       # %bb.0: # %entry
-; NO-FAST-P8-NEXT:    xscmpudp cr0, f1, f2
-; NO-FAST-P8-NEXT:    fmr f1, f3
-; NO-FAST-P8-NEXT:    bgelr cr0
-; NO-FAST-P8-NEXT:  # %bb.1: # %entry
-; NO-FAST-P8-NEXT:    fmr f1, f4
+; NO-FAST-P8-NEXT:    xssubdp f0, f1, f2
+; NO-FAST-P8-NEXT:    fsel f1, f0, f3, f4
 ; NO-FAST-P8-NEXT:    blr
 ;
 ; NO-FAST-P9-LABEL: select_fast_oge_double:
 ; NO-FAST-P9:       # %bb.0: # %entry
-; NO-FAST-P9-NEXT:    xscmpudp cr0, f1, f2
-; NO-FAST-P9-NEXT:    bge cr0, .LBB11_2
-; NO-FAST-P9-NEXT:  # %bb.1: # %entry
-; NO-FAST-P9-NEXT:    fmr f3, f4
-; NO-FAST-P9-NEXT:  .LBB11_2: # %entry
-; NO-FAST-P9-NEXT:    fmr f1, f3
+; NO-FAST-P9-NEXT:    xssubdp f0, f1, f2
+; NO-FAST-P9-NEXT:    fsel f1, f0, f3, f4
 ; NO-FAST-P9-NEXT:    blr
 entry:
   %cmp = fcmp fast oge double %a, %b
   %cond = select fast i1 %cmp, double %c, double %d
   ret double %cond
 }
 
 

 ; FAST-P9-LABEL: select_fast_olt_float:
 ; FAST-P9:       # %bb.0: # %entry
 ; FAST-P9-NEXT:    xssubsp f0, f1, f2
 ; FAST-P9-NEXT:    fsel f1, f0, f4, f3
 ; FAST-P9-NEXT:    blr
 ;
 ; NO-FAST-P8-LABEL: select_fast_olt_float:
 ; NO-FAST-P8:       # %bb.0: # %entry
-; NO-FAST-P8-NEXT:    fcmpu cr0, f1, f2
-; NO-FAST-P8-NEXT:    fmr f1, f3
-; NO-FAST-P8-NEXT:    bltlr cr0
-; NO-FAST-P8-NEXT:  # %bb.1: # %entry
-; NO-FAST-P8-NEXT:    fmr f1, f4
+; NO-FAST-P8-NEXT:    xssubsp f0, f1, f2
+; NO-FAST-P8-NEXT:    fsel f1, f0, f4, f3
 ; NO-FAST-P8-NEXT:    blr
 ;
 ; NO-FAST-P9-LABEL: select_fast_olt_float:
 ; NO-FAST-P9:       # %bb.0: # %entry
-; NO-FAST-P9-NEXT:    fcmpu cr0, f1, f2
-; NO-FAST-P9-NEXT:    blt cr0, .LBB14_2
-; NO-FAST-P9-NEXT:  # %bb.1: # %entry
-; NO-FAST-P9-NEXT:    fmr f3, f4
-; NO-FAST-P9-NEXT:  .LBB14_2: # %entry
-; NO-FAST-P9-NEXT:    fmr f1, f3
+; NO-FAST-P9-NEXT:    xssubsp f0, f1, f2
+; NO-FAST-P9-NEXT:    fsel f1, f0, f4, f3
 ; NO-FAST-P9-NEXT:    blr
 entry:
   %cmp = fcmp fast olt float %a, %b
   %cond = select fast i1 %cmp, float %c, float %d
   ret float %cond
 }
 
 define double @select_fast_olt_double(double %a, double %b, double %c, double %d) {
 ; FAST-P8-LABEL: select_fast_olt_double:
 ; FAST-P8:       # %bb.0: # %entry
 ; FAST-P8-NEXT:    xssubdp f0, f1, f2
 ; FAST-P8-NEXT:    fsel f1, f0, f4, f3
 ; FAST-P8-NEXT:    blr
 ;
 ; FAST-P9-LABEL: select_fast_olt_double:
 ; FAST-P9:       # %bb.0: # %entry
 ; FAST-P9-NEXT:    xssubdp f0, f1, f2
 ; FAST-P9-NEXT:    fsel f1, f0, f4, f3
 ; FAST-P9-NEXT:    blr
 ;
 ; NO-FAST-P8-LABEL: select_fast_olt_double:
 ; NO-FAST-P8:       # %bb.0: # %entry
-; NO-FAST-P8-NEXT:    xscmpudp cr0, f1, f2
-; NO-FAST-P8-NEXT:    fmr f1, f3
-; NO-FAST-P8-NEXT:    bltlr cr0
-; NO-FAST-P8-NEXT:  # %bb.1: # %entry
-; NO-FAST-P8-NEXT:    fmr f1, f4
+; NO-FAST-P8-NEXT:    xssubdp f0, f1, f2
+; NO-FAST-P8-NEXT:    fsel f1, f0, f4, f3
 ; NO-FAST-P8-NEXT:    blr
 ;
 ; NO-FAST-P9-LABEL: select_fast_olt_double:
 ; NO-FAST-P9:       # %bb.0: # %entry
-; NO-FAST-P9-NEXT:    xscmpudp cr0, f1, f2
-; NO-FAST-P9-NEXT:    blt cr0, .LBB15_2
-; NO-FAST-P9-NEXT:  # %bb.1: # %entry
-; NO-FAST-P9-NEXT:    fmr f3, f4
-; NO-FAST-P9-NEXT:  .LBB15_2: # %entry
-; NO-FAST-P9-NEXT:    fmr f1, f3
+; NO-FAST-P9-NEXT:    xssubdp f0, f1, f2
+; NO-FAST-P9-NEXT:    fsel f1, f0, f4, f3
 ; NO-FAST-P9-NEXT:    blr
 entry:
   %cmp = fcmp fast olt double %a, %b
   %cond = select fast i1 %cmp, double %c, double %d
   ret double %cond
 }
 
 

 ; FAST-P9-LABEL: select_fast_ogt_float:
 ; FAST-P9:       # %bb.0: # %entry
 ; FAST-P9-NEXT:    xssubsp f0, f2, f1
 ; FAST-P9-NEXT:    fsel f1, f0, f4, f3
 ; FAST-P9-NEXT:    blr
 ;
 ; NO-FAST-P8-LABEL: select_fast_ogt_float:
 ; NO-FAST-P8:       # %bb.0: # %entry
-; NO-FAST-P8-NEXT:    fcmpu cr0, f1, f2
-; NO-FAST-P8-NEXT:    fmr f1, f3
-; NO-FAST-P8-NEXT:    bgtlr cr0
-; NO-FAST-P8-NEXT:  # %bb.1: # %entry
-; NO-FAST-P8-NEXT:    fmr f1, f4
+; NO-FAST-P8-NEXT:    xssubsp f0, f2, f1
+; NO-FAST-P8-NEXT:    fsel f1, f0, f4, f3
 ; NO-FAST-P8-NEXT:    blr
 ;
 ; NO-FAST-P9-LABEL: select_fast_ogt_float:
 ; NO-FAST-P9:       # %bb.0: # %entry
-; NO-FAST-P9-NEXT:    fcmpu cr0, f1, f2
-; NO-FAST-P9-NEXT:    bgt cr0, .LBB18_2
-; NO-FAST-P9-NEXT:  # %bb.1: # %entry
-; NO-FAST-P9-NEXT:    fmr f3, f4
-; NO-FAST-P9-NEXT:  .LBB18_2: # %entry
-; NO-FAST-P9-NEXT:    fmr f1, f3
+; NO-FAST-P9-NEXT:    xssubsp f0, f2, f1
+; NO-FAST-P9-NEXT:    fsel f1, f0, f4, f3
 ; NO-FAST-P9-NEXT:    blr
 entry:
   %cmp = fcmp fast ogt float %a, %b
   %cond = select fast i1 %cmp, float %c, float %d
   ret float %cond
 }
 
 define double @select_fast_ogt_double(double %a, double %b, double %c, double %d) {
 ; FAST-P8-LABEL: select_fast_ogt_double:
 ; FAST-P8:       # %bb.0: # %entry
 ; FAST-P8-NEXT:    xssubdp f0, f2, f1
 ; FAST-P8-NEXT:    fsel f1, f0, f4, f3
 ; FAST-P8-NEXT:    blr
 ;
 ; FAST-P9-LABEL: select_fast_ogt_double:
 ; FAST-P9:       # %bb.0: # %entry
 ; FAST-P9-NEXT:    xssubdp f0, f2, f1
 ; FAST-P9-NEXT:    fsel f1, f0, f4, f3
 ; FAST-P9-NEXT:    blr
 ;
 ; NO-FAST-P8-LABEL: select_fast_ogt_double:
 ; NO-FAST-P8:       # %bb.0: # %entry
-; NO-FAST-P8-NEXT:    xscmpudp cr0, f1, f2
-; NO-FAST-P8-NEXT:    fmr f1, f3
-; NO-FAST-P8-NEXT:    bgtlr cr0
-; NO-FAST-P8-NEXT:  # %bb.1: # %entry
-; NO-FAST-P8-NEXT:    fmr f1, f4
+; NO-FAST-P8-NEXT:    xssubdp f0, f2, f1
+; NO-FAST-P8-NEXT:    fsel f1, f0, f4, f3
 ; NO-FAST-P8-NEXT:    blr
 ;
 ; NO-FAST-P9-LABEL: select_fast_ogt_double:
 ; NO-FAST-P9:       # %bb.0: # %entry
-; NO-FAST-P9-NEXT:    xscmpudp cr0, f1, f2
-; NO-FAST-P9-NEXT:    bgt cr0, .LBB19_2
-; NO-FAST-P9-NEXT:  # %bb.1: # %entry
-; NO-FAST-P9-NEXT:    fmr f3, f4
-; NO-FAST-P9-NEXT:  .LBB19_2: # %entry
-; NO-FAST-P9-NEXT:    fmr f1, f3
+; NO-FAST-P9-NEXT:    xssubdp f0, f2, f1
+; NO-FAST-P9-NEXT:    fsel f1, f0, f4, f3
 ; NO-FAST-P9-NEXT:    blr
 entry:
   %cmp = fcmp fast ogt double %a, %b
   %cond = select fast i1 %cmp, double %c, double %d
   ret double %cond
 }
 
 

 ; FAST-P9-LABEL: select_fast_ole_float:
 ; FAST-P9:       # %bb.0: # %entry
 ; FAST-P9-NEXT:    xssubsp f0, f2, f1
 ; FAST-P9-NEXT:    fsel f1, f0, f3, f4
 ; FAST-P9-NEXT:    blr
 ;
 ; NO-FAST-P8-LABEL: select_fast_ole_float:
 ; NO-FAST-P8:       # %bb.0: # %entry
-; NO-FAST-P8-NEXT:    fcmpu cr0, f1, f2
-; NO-FAST-P8-NEXT:    fmr f1, f3
-; NO-FAST-P8-NEXT:    blelr cr0
-; NO-FAST-P8-NEXT:  # %bb.1: # %entry
-; NO-FAST-P8-NEXT:    fmr f1, f4
+; NO-FAST-P8-NEXT:    xssubsp f0, f2, f1
+; NO-FAST-P8-NEXT:    fsel f1, f0, f3, f4
 ; NO-FAST-P8-NEXT:    blr
 ;
 ; NO-FAST-P9-LABEL: select_fast_ole_float:
 ; NO-FAST-P9:       # %bb.0: # %entry
-; NO-FAST-P9-NEXT:    fcmpu cr0, f1, f2
-; NO-FAST-P9-NEXT:    ble cr0, .LBB22_2
-; NO-FAST-P9-NEXT:  # %bb.1: # %entry
-; NO-FAST-P9-NEXT:    fmr f3, f4
-; NO-FAST-P9-NEXT:  .LBB22_2: # %entry
-; NO-FAST-P9-NEXT:    fmr f1, f3
+; NO-FAST-P9-NEXT:    xssubsp f0, f2, f1
+; NO-FAST-P9-NEXT:    fsel f1, f0, f3, f4
 ; NO-FAST-P9-NEXT:    blr
 entry:
   %cmp = fcmp fast ole float %a, %b
   %cond = select fast i1 %cmp, float %c, float %d
   ret float %cond
 }
 
 define double @select_fast_ole_double(double %a, double %b, double %c, double %d) {
 ; FAST-P8-LABEL: select_fast_ole_double:
 ; FAST-P8:       # %bb.0: # %entry
 ; FAST-P8-NEXT:    xssubdp f0, f2, f1
 ; FAST-P8-NEXT:    fsel f1, f0, f3, f4
 ; FAST-P8-NEXT:    blr
 ;
 ; FAST-P9-LABEL: select_fast_ole_double:
 ; FAST-P9:       # %bb.0: # %entry
 ; FAST-P9-NEXT:    xssubdp f0, f2, f1
 ; FAST-P9-NEXT:    fsel f1, f0, f3, f4
 ; FAST-P9-NEXT:    blr
 ;
 ; NO-FAST-P8-LABEL: select_fast_ole_double:
 ; NO-FAST-P8:       # %bb.0: # %entry
-; NO-FAST-P8-NEXT:    xscmpudp cr0, f1, f2
-; NO-FAST-P8-NEXT:    fmr f1, f3
-; NO-FAST-P8-NEXT:    blelr cr0
-; NO-FAST-P8-NEXT:  # %bb.1: # %entry
-; NO-FAST-P8-NEXT:    fmr f1, f4
+; NO-FAST-P8-NEXT:    xssubdp f0, f2, f1
+; NO-FAST-P8-NEXT:    fsel f1, f0, f3, f4
 ; NO-FAST-P8-NEXT:    blr
 ;
 ; NO-FAST-P9-LABEL: select_fast_ole_double:
 ; NO-FAST-P9:       # %bb.0: # %entry
-; NO-FAST-P9-NEXT:    xscmpudp cr0, f1, f2
-; NO-FAST-P9-NEXT:    ble cr0, .LBB23_2
-; NO-FAST-P9-NEXT:  # %bb.1: # %entry
-; NO-FAST-P9-NEXT:    fmr f3, f4
-; NO-FAST-P9-NEXT:  .LBB23_2: # %entry
-; NO-FAST-P9-NEXT:    fmr f1, f3
+; NO-FAST-P9-NEXT:    xssubdp f0, f2, f1
+; NO-FAST-P9-NEXT:    fsel f1, f0, f3, f4
 ; NO-FAST-P9-NEXT:    blr
 entry:
   %cmp = fcmp fast ole double %a, %b
   %cond = select fast i1 %cmp, double %c, double %d
   ret double %cond
 }
 
 ; Test RHS is 1.000000e+00