Index: lib/Target/AMDGPU/SIISelLowering.cpp
===================================================================
--- lib/Target/AMDGPU/SIISelLowering.cpp
+++ lib/Target/AMDGPU/SIISelLowering.cpp
@@ -1840,35 +1840,38 @@
     return SDValue();
 
   SDLoc SL(Op);
-  SDValue LHS = Op.getOperand(0);
-  SDValue RHS = Op.getOperand(1);
 
-  SDValue r1 = DAG.getNode(ISD::FABS, SL, MVT::f32, RHS);
+  SDValue X = Op.getOperand(0);
+  SDValue Y = Op.getOperand(1);
+
+  const SDValue One = DAG.getConstantFP(1.0, SL, MVT::f32);
 
-  const APFloat K0Val(BitsToFloat(0x6f800000));
-  const SDValue K0 = DAG.getConstantFP(K0Val, SL, MVT::f32);
+  SDVTList ScaleVT = DAG.getVTList(MVT::f32, MVT::i1);
 
-  const APFloat K1Val(BitsToFloat(0x2f800000));
-  const SDValue K1 = DAG.getConstantFP(K1Val, SL, MVT::f32);
+  SDValue d_scaled = DAG.getNode(AMDGPUISD::DIV_SCALE, SL, ScaleVT, Y, Y, X);
+  SDValue n_scaled = DAG.getNode(AMDGPUISD::DIV_SCALE, SL, ScaleVT, X, Y, X);
 
-  const SDValue One = DAG.getConstantFP(1.0, SL, MVT::f32);
+  SDValue approx_rcp = DAG.getNode(AMDGPUISD::RCP, SL, MVT::f32, d_scaled);
 
-  EVT SetCCVT =
-      getSetCCResultType(DAG.getDataLayout(), *DAG.getContext(), MVT::f32);
+  SDValue NegDivScale0 = DAG.getNode(ISD::FNEG, SL, MVT::f32, d_scaled);
 
-  SDValue r2 = DAG.getSetCC(SL, SetCCVT, r1, K0, ISD::SETOGT);
+  SDValue Fma0 = DAG.getNode(ISD::FMA, SL, MVT::f32, NegDivScale0, approx_rcp, One);
 
-  SDValue r3 = DAG.getNode(ISD::SELECT, SL, MVT::f32, r2, K1, One);
+  SDValue Fma1 = DAG.getNode(ISD::FMA, SL, MVT::f32, Fma0, approx_rcp, approx_rcp);
 
-  // TODO: Should this propagate fast-math-flags?
+  SDValue Mul = DAG.getNode(ISD::FMUL, SL, MVT::f32, n_scaled, Fma1);
+
+  SDValue Fma2 = DAG.getNode(ISD::FMA, SL, MVT::f32, NegDivScale0, Mul, n_scaled);
+
+  SDValue Fma3 = DAG.getNode(ISD::FMA, SL, MVT::f32, Fma2, Fma1, Mul);
 
-  r1 = DAG.getNode(ISD::FMUL, SL, MVT::f32, RHS, r3);
+  SDValue Fma4 = DAG.getNode(ISD::FMA, SL, MVT::f32, NegDivScale0, Fma3, n_scaled);
 
-  SDValue r0 = DAG.getNode(AMDGPUISD::RCP, SL, MVT::f32, r1);
+  SDValue Scale = n_scaled.getValue(1);
 
-  SDValue Mul = DAG.getNode(ISD::FMUL, SL, MVT::f32, LHS, r0);
+  SDValue Fmas = DAG.getNode(AMDGPUISD::DIV_FMAS, SL, MVT::f32, Fma4, Fma1, Fma3, Scale);
 
-  return DAG.getNode(ISD::FMUL, SL, MVT::f32, r3, Mul);
+  return DAG.getNode(AMDGPUISD::DIV_FIXUP, SL, MVT::f32, Fmas, Y, X);
 }
 
 SDValue SITargetLowering::LowerFDIV64(SDValue Op, SelectionDAG &DAG) const {
Index: test/CodeGen/AMDGPU/frem.ll
===================================================================
--- test/CodeGen/AMDGPU/frem.ll
+++ test/CodeGen/AMDGPU/frem.ll
@@ -5,11 +5,12 @@
 ; FUNC-LABEL: {{^}}frem_f32:
 ; GCN-DAG: buffer_load_dword [[X:v[0-9]+]], {{.*$}}
 ; GCN-DAG: buffer_load_dword [[Y:v[0-9]+]], {{.*}} offset:16
-; GCN-DAG: v_cmp
-; GCN-DAG: v_mul_f32
+; GCN: v_div_scale_f32
 ; GCN: v_rcp_f32_e32
+; GCN: v_fma_f32
 ; GCN: v_mul_f32_e32
-; GCN: v_mul_f32_e32
+; GCN: v_div_fmas_f32
+; GCN: v_div_fixup_f32
 ; GCN: v_trunc_f32_e32
 ; GCN: v_mad_f32
 ; GCN: s_endpgm