Skip to content

Commit ce5620b

Browse files
paulwalker-armpaulwalker-arm
authored
[LLVM][VPlan] Pick more optimal initial value for VPBlend. (#104019)
By choosing an initial value whose mask is only used by the blend we can remove the need for the mask entirely.
1 parent 87a988e commit ce5620b

File tree

9 files changed

+207
-93
lines changed

9 files changed

+207
-93
lines changed

‎llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp

Lines changed: 12 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -878,6 +878,17 @@ static void simplifyRecipe(VPRecipeBase &R, VPTypeAnalysis &TypeInfo) {
878878
// value with the others blended into it.
879879

880880
unsigned StartIndex = 0;
881+
for (unsigned I = 0; I != Blend->getNumIncomingValues(); ++I) {
882+
// If a value's mask is used only by the blend then is can be deadcoded.
883+
// TODO: Find the most expensive mask that can be deadcoded, or a mask
884+
// that's used by multiple blends where it can be removed from them all.
885+
VPValue *Mask = Blend->getMask(I);
886+
if (Mask->getNumUsers() == 1 && !match(Mask, m_False())) {
887+
StartIndex = I;
888+
break;
889+
}
890+
}
891+
881892
SmallVector<VPValue *, 4> OperandsWithMask;
882893
OperandsWithMask.push_back(Blend->getIncomingValue(StartIndex));
883894

@@ -956,6 +967,7 @@ static void simplifyRecipe(VPRecipeBase &R, VPTypeAnalysis &TypeInfo) {
956967
m_LogicalAnd(m_VPValue(X1), m_Not(m_VPValue(Y1))))) &&
957968
X == X1 && Y == Y1) {
958969
R.getVPSingleValue()->replaceAllUsesWith(X);
970+
R.eraseFromParent();
959971
return;
960972
}
961973

‎llvm/test/Transforms/LoopVectorize/AArch64/masked-call.ll

Lines changed: 21 additions & 27 deletions
Original file line numberDiff line numberDiff line change
@@ -215,16 +215,14 @@ define void @test_if_then(ptr noalias %a, ptr readnone %b) #4 {
215215
; TFCOMMON-NEXT: [[TMP6:%.*]] = icmp ugt <vscale x 2 x i64> [[WIDE_MASKED_LOAD]], shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 50, i64 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer)
216216
; TFCOMMON-NEXT: [[TMP7:%.*]] = select <vscale x 2 x i1> [[ACTIVE_LANE_MASK]], <vscale x 2 x i1> [[TMP6]], <vscale x 2 x i1> zeroinitializer
217217
; TFCOMMON-NEXT: [[TMP8:%.*]] = call <vscale x 2 x i64> @foo_vector(<vscale x 2 x i64> [[WIDE_MASKED_LOAD]], <vscale x 2 x i1> [[TMP7]])
218-
; TFCOMMON-NEXT: [[TMP9:%.*]] = xor <vscale x 2 x i1> [[TMP6]], shufflevector (<vscale x 2 x i1> insertelement (<vscale x 2 x i1> poison, i1 true, i64 0), <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer)
219-
; TFCOMMON-NEXT: [[TMP10:%.*]] = select <vscale x 2 x i1> [[ACTIVE_LANE_MASK]], <vscale x 2 x i1> [[TMP9]], <vscale x 2 x i1> zeroinitializer
220-
; TFCOMMON-NEXT: [[PREDPHI:%.*]] = select <vscale x 2 x i1> [[TMP10]], <vscale x 2 x i64> zeroinitializer, <vscale x 2 x i64> [[TMP8]]
221-
; TFCOMMON-NEXT: [[TMP11:%.*]] = getelementptr inbounds i64, ptr [[B:%.*]], i64 [[INDEX]]
222-
; TFCOMMON-NEXT: call void @llvm.masked.store.nxv2i64.p0(<vscale x 2 x i64> [[PREDPHI]], ptr [[TMP11]], i32 8, <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
218+
; TFCOMMON-NEXT: [[PREDPHI:%.*]] = select <vscale x 2 x i1> [[TMP7]], <vscale x 2 x i64> [[TMP8]], <vscale x 2 x i64> zeroinitializer
219+
; TFCOMMON-NEXT: [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[B:%.*]], i64 [[INDEX]]
220+
; TFCOMMON-NEXT: call void @llvm.masked.store.nxv2i64.p0(<vscale x 2 x i64> [[PREDPHI]], ptr [[TMP9]], i32 8, <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
223221
; TFCOMMON-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP4]]
224222
; TFCOMMON-NEXT: [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i64(i64 [[INDEX_NEXT]], i64 1025)
225-
; TFCOMMON-NEXT: [[TMP12:%.*]] = xor <vscale x 2 x i1> [[ACTIVE_LANE_MASK_NEXT]], shufflevector (<vscale x 2 x i1> insertelement (<vscale x 2 x i1> poison, i1 true, i64 0), <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer)
226-
; TFCOMMON-NEXT: [[TMP13:%.*]] = extractelement <vscale x 2 x i1> [[TMP12]], i32 0
227-
; TFCOMMON-NEXT: br i1 [[TMP13]], label [[FOR_COND_CLEANUP:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
223+
; TFCOMMON-NEXT: [[TMP10:%.*]] = xor <vscale x 2 x i1> [[ACTIVE_LANE_MASK_NEXT]], shufflevector (<vscale x 2 x i1> insertelement (<vscale x 2 x i1> poison, i1 true, i64 0), <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer)
224+
; TFCOMMON-NEXT: [[TMP11:%.*]] = extractelement <vscale x 2 x i1> [[TMP10]], i32 0
225+
; TFCOMMON-NEXT: br i1 [[TMP11]], label [[FOR_COND_CLEANUP:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
228226
; TFCOMMON: for.cond.cleanup:
229227
; TFCOMMON-NEXT: ret void
230228
;
@@ -259,27 +257,23 @@ define void @test_if_then(ptr noalias %a, ptr readnone %b) #4 {
259257
; TFA_INTERLEAVE-NEXT: [[TMP14:%.*]] = select <vscale x 2 x i1> [[ACTIVE_LANE_MASK2]], <vscale x 2 x i1> [[TMP12]], <vscale x 2 x i1> zeroinitializer
260258
; TFA_INTERLEAVE-NEXT: [[TMP15:%.*]] = call <vscale x 2 x i64> @foo_vector(<vscale x 2 x i64> [[WIDE_MASKED_LOAD]], <vscale x 2 x i1> [[TMP13]])
261259
; TFA_INTERLEAVE-NEXT: [[TMP16:%.*]] = call <vscale x 2 x i64> @foo_vector(<vscale x 2 x i64> [[WIDE_MASKED_LOAD3]], <vscale x 2 x i1> [[TMP14]])
262-
; TFA_INTERLEAVE-NEXT: [[TMP17:%.*]] = xor <vscale x 2 x i1> [[TMP11]], shufflevector (<vscale x 2 x i1> insertelement (<vscale x 2 x i1> poison, i1 true, i64 0), <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer)
263-
; TFA_INTERLEAVE-NEXT: [[TMP18:%.*]] = xor <vscale x 2 x i1> [[TMP12]], shufflevector (<vscale x 2 x i1> insertelement (<vscale x 2 x i1> poison, i1 true, i64 0), <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer)
264-
; TFA_INTERLEAVE-NEXT: [[TMP19:%.*]] = select <vscale x 2 x i1> [[ACTIVE_LANE_MASK]], <vscale x 2 x i1> [[TMP17]], <vscale x 2 x i1> zeroinitializer
265-
; TFA_INTERLEAVE-NEXT: [[TMP20:%.*]] = select <vscale x 2 x i1> [[ACTIVE_LANE_MASK2]], <vscale x 2 x i1> [[TMP18]], <vscale x 2 x i1> zeroinitializer
266-
; TFA_INTERLEAVE-NEXT: [[PREDPHI:%.*]] = select <vscale x 2 x i1> [[TMP19]], <vscale x 2 x i64> zeroinitializer, <vscale x 2 x i64> [[TMP15]]
267-
; TFA_INTERLEAVE-NEXT: [[PREDPHI4:%.*]] = select <vscale x 2 x i1> [[TMP20]], <vscale x 2 x i64> zeroinitializer, <vscale x 2 x i64> [[TMP16]]
268-
; TFA_INTERLEAVE-NEXT: [[TMP21:%.*]] = getelementptr inbounds i64, ptr [[B:%.*]], i64 [[INDEX]]
269-
; TFA_INTERLEAVE-NEXT: [[TMP22:%.*]] = call i64 @llvm.vscale.i64()
270-
; TFA_INTERLEAVE-NEXT: [[TMP23:%.*]] = mul i64 [[TMP22]], 2
271-
; TFA_INTERLEAVE-NEXT: [[TMP24:%.*]] = getelementptr inbounds i64, ptr [[TMP21]], i64 [[TMP23]]
272-
; TFA_INTERLEAVE-NEXT: call void @llvm.masked.store.nxv2i64.p0(<vscale x 2 x i64> [[PREDPHI]], ptr [[TMP21]], i32 8, <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
273-
; TFA_INTERLEAVE-NEXT: call void @llvm.masked.store.nxv2i64.p0(<vscale x 2 x i64> [[PREDPHI4]], ptr [[TMP24]], i32 8, <vscale x 2 x i1> [[ACTIVE_LANE_MASK2]])
260+
; TFA_INTERLEAVE-NEXT: [[PREDPHI:%.*]] = select <vscale x 2 x i1> [[TMP13]], <vscale x 2 x i64> [[TMP15]], <vscale x 2 x i64> zeroinitializer
261+
; TFA_INTERLEAVE-NEXT: [[PREDPHI4:%.*]] = select <vscale x 2 x i1> [[TMP14]], <vscale x 2 x i64> [[TMP16]], <vscale x 2 x i64> zeroinitializer
262+
; TFA_INTERLEAVE-NEXT: [[TMP17:%.*]] = getelementptr inbounds i64, ptr [[B:%.*]], i64 [[INDEX]]
263+
; TFA_INTERLEAVE-NEXT: [[TMP18:%.*]] = call i64 @llvm.vscale.i64()
264+
; TFA_INTERLEAVE-NEXT: [[TMP19:%.*]] = mul i64 [[TMP18]], 2
265+
; TFA_INTERLEAVE-NEXT: [[TMP20:%.*]] = getelementptr inbounds i64, ptr [[TMP17]], i64 [[TMP19]]
266+
; TFA_INTERLEAVE-NEXT: call void @llvm.masked.store.nxv2i64.p0(<vscale x 2 x i64> [[PREDPHI]], ptr [[TMP17]], i32 8, <vscale x 2 x i1> [[ACTIVE_LANE_MASK]])
267+
; TFA_INTERLEAVE-NEXT: call void @llvm.masked.store.nxv2i64.p0(<vscale x 2 x i64> [[PREDPHI4]], ptr [[TMP20]], i32 8, <vscale x 2 x i1> [[ACTIVE_LANE_MASK2]])
274268
; TFA_INTERLEAVE-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP4]]
275-
; TFA_INTERLEAVE-NEXT: [[TMP25:%.*]] = call i64 @llvm.vscale.i64()
276-
; TFA_INTERLEAVE-NEXT: [[TMP26:%.*]] = mul i64 [[TMP25]], 2
277-
; TFA_INTERLEAVE-NEXT: [[TMP27:%.*]] = add i64 [[INDEX_NEXT]], [[TMP26]]
269+
; TFA_INTERLEAVE-NEXT: [[TMP21:%.*]] = call i64 @llvm.vscale.i64()
270+
; TFA_INTERLEAVE-NEXT: [[TMP22:%.*]] = mul i64 [[TMP21]], 2
271+
; TFA_INTERLEAVE-NEXT: [[TMP23:%.*]] = add i64 [[INDEX_NEXT]], [[TMP22]]
278272
; TFA_INTERLEAVE-NEXT: [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i64(i64 [[INDEX_NEXT]], i64 1025)
279-
; TFA_INTERLEAVE-NEXT: [[ACTIVE_LANE_MASK_NEXT5]] = call <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i64(i64 [[TMP27]], i64 1025)
280-
; TFA_INTERLEAVE-NEXT: [[TMP28:%.*]] = xor <vscale x 2 x i1> [[ACTIVE_LANE_MASK_NEXT]], shufflevector (<vscale x 2 x i1> insertelement (<vscale x 2 x i1> poison, i1 true, i64 0), <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer)
281-
; TFA_INTERLEAVE-NEXT: [[TMP29:%.*]] = extractelement <vscale x 2 x i1> [[TMP28]], i32 0
282-
; TFA_INTERLEAVE-NEXT: br i1 [[TMP29]], label [[FOR_COND_CLEANUP:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
273+
; TFA_INTERLEAVE-NEXT: [[ACTIVE_LANE_MASK_NEXT5]] = call <vscale x 2 x i1> @llvm.get.active.lane.mask.nxv2i1.i64(i64 [[TMP23]], i64 1025)
274+
; TFA_INTERLEAVE-NEXT: [[TMP24:%.*]] = xor <vscale x 2 x i1> [[ACTIVE_LANE_MASK_NEXT]], shufflevector (<vscale x 2 x i1> insertelement (<vscale x 2 x i1> poison, i1 true, i64 0), <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer)
275+
; TFA_INTERLEAVE-NEXT: [[TMP25:%.*]] = extractelement <vscale x 2 x i1> [[TMP24]], i32 0
276+
; TFA_INTERLEAVE-NEXT: br i1 [[TMP25]], label [[FOR_COND_CLEANUP:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
283277
; TFA_INTERLEAVE: for.cond.cleanup:
284278
; TFA_INTERLEAVE-NEXT: ret void
285279
;

‎llvm/test/Transforms/LoopVectorize/AArch64/scalable-strict-fadd.ll

Lines changed: 13 additions & 15 deletions
Original file line numberDiff line numberDiff line change
@@ -1216,7 +1216,7 @@ define float @fadd_conditional(ptr noalias nocapture readonly %a, ptr noalias no
12161216
; CHECK-ORDERED-TF: vector.body:
12171217
; CHECK-ORDERED-TF-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
12181218
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK:%.*]] = phi <vscale x 4 x i1> [ [[ACTIVE_LANE_MASK_ENTRY]], [[VECTOR_PH]] ], [ [[ACTIVE_LANE_MASK_NEXT:%.*]], [[VECTOR_BODY]] ]
1219-
; CHECK-ORDERED-TF-NEXT: [[VEC_PHI:%.*]] = phi float [ 1.000000e+00, [[VECTOR_PH]] ], [ [[TMP20:%.*]], [[VECTOR_BODY]] ]
1219+
; CHECK-ORDERED-TF-NEXT: [[VEC_PHI:%.*]] = phi float [ 1.000000e+00, [[VECTOR_PH]] ], [ [[TMP18:%.*]], [[VECTOR_BODY]] ]
12201220
; CHECK-ORDERED-TF-NEXT: [[TMP10:%.*]] = add i64 [[INDEX]], 0
12211221
; CHECK-ORDERED-TF-NEXT: [[TMP11:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[TMP10]]
12221222
; CHECK-ORDERED-TF-NEXT: [[TMP12:%.*]] = getelementptr inbounds float, ptr [[TMP11]], i32 0
@@ -1226,41 +1226,39 @@ define float @fadd_conditional(ptr noalias nocapture readonly %a, ptr noalias no
12261226
; CHECK-ORDERED-TF-NEXT: [[TMP15:%.*]] = getelementptr float, ptr [[A]], i64 [[TMP10]]
12271227
; CHECK-ORDERED-TF-NEXT: [[TMP16:%.*]] = getelementptr float, ptr [[TMP15]], i32 0
12281228
; CHECK-ORDERED-TF-NEXT: [[WIDE_MASKED_LOAD1:%.*]] = call <vscale x 4 x float> @llvm.masked.load.nxv4f32.p0(ptr [[TMP16]], i32 4, <vscale x 4 x i1> [[TMP14]], <vscale x 4 x float> poison)
1229-
; CHECK-ORDERED-TF-NEXT: [[TMP17:%.*]] = xor <vscale x 4 x i1> [[TMP13]], shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer)
1230-
; CHECK-ORDERED-TF-NEXT: [[TMP18:%.*]] = select <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x i1> [[TMP17]], <vscale x 4 x i1> zeroinitializer
1231-
; CHECK-ORDERED-TF-NEXT: [[PREDPHI:%.*]] = select <vscale x 4 x i1> [[TMP18]], <vscale x 4 x float> shufflevector (<vscale x 4 x float> insertelement (<vscale x 4 x float> poison, float 3.000000e+00, i64 0), <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer), <vscale x 4 x float> [[WIDE_MASKED_LOAD1]]
1232-
; CHECK-ORDERED-TF-NEXT: [[TMP19:%.*]] = select <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x float> [[PREDPHI]], <vscale x 4 x float> shufflevector (<vscale x 4 x float> insertelement (<vscale x 4 x float> poison, float -0.000000e+00, i64 0), <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer)
1233-
; CHECK-ORDERED-TF-NEXT: [[TMP20]] = call float @llvm.vector.reduce.fadd.nxv4f32(float [[VEC_PHI]], <vscale x 4 x float> [[TMP19]])
1229+
; CHECK-ORDERED-TF-NEXT: [[PREDPHI:%.*]] = select <vscale x 4 x i1> [[TMP14]], <vscale x 4 x float> [[WIDE_MASKED_LOAD1]], <vscale x 4 x float> shufflevector (<vscale x 4 x float> insertelement (<vscale x 4 x float> poison, float 3.000000e+00, i64 0), <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer)
1230+
; CHECK-ORDERED-TF-NEXT: [[TMP17:%.*]] = select <vscale x 4 x i1> [[ACTIVE_LANE_MASK]], <vscale x 4 x float> [[PREDPHI]], <vscale x 4 x float> shufflevector (<vscale x 4 x float> insertelement (<vscale x 4 x float> poison, float -0.000000e+00, i64 0), <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer)
1231+
; CHECK-ORDERED-TF-NEXT: [[TMP18]] = call float @llvm.vector.reduce.fadd.nxv4f32(float [[VEC_PHI]], <vscale x 4 x float> [[TMP17]])
12341232
; CHECK-ORDERED-TF-NEXT: [[INDEX_NEXT]] = add i64 [[INDEX]], [[TMP4]]
12351233
; CHECK-ORDERED-TF-NEXT: [[ACTIVE_LANE_MASK_NEXT]] = call <vscale x 4 x i1> @llvm.get.active.lane.mask.nxv4i1.i64(i64 [[INDEX]], i64 [[TMP9]])
1236-
; CHECK-ORDERED-TF-NEXT: [[TMP21:%.*]] = xor <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer)
1237-
; CHECK-ORDERED-TF-NEXT: [[TMP22:%.*]] = extractelement <vscale x 4 x i1> [[TMP21]], i32 0
1238-
; CHECK-ORDERED-TF-NEXT: br i1 [[TMP22]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
1234+
; CHECK-ORDERED-TF-NEXT: [[TMP19:%.*]] = xor <vscale x 4 x i1> [[ACTIVE_LANE_MASK_NEXT]], shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer)
1235+
; CHECK-ORDERED-TF-NEXT: [[TMP20:%.*]] = extractelement <vscale x 4 x i1> [[TMP19]], i32 0
1236+
; CHECK-ORDERED-TF-NEXT: br i1 [[TMP20]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
12391237
; CHECK-ORDERED-TF: middle.block:
12401238
; CHECK-ORDERED-TF-NEXT: br i1 true, label [[FOR_END:%.*]], label [[SCALAR_PH]]
12411239
; CHECK-ORDERED-TF: scalar.ph:
12421240
; CHECK-ORDERED-TF-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
1243-
; CHECK-ORDERED-TF-NEXT: [[BC_MERGE_RDX:%.*]] = phi float [ [[TMP20]], [[MIDDLE_BLOCK]] ], [ 1.000000e+00, [[ENTRY]] ]
1241+
; CHECK-ORDERED-TF-NEXT: [[BC_MERGE_RDX:%.*]] = phi float [ [[TMP18]], [[MIDDLE_BLOCK]] ], [ 1.000000e+00, [[ENTRY]] ]
12441242
; CHECK-ORDERED-TF-NEXT: br label [[FOR_BODY:%.*]]
12451243
; CHECK-ORDERED-TF: for.body:
12461244
; CHECK-ORDERED-TF-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_INC:%.*]] ]
12471245
; CHECK-ORDERED-TF-NEXT: [[RES:%.*]] = phi float [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[FADD:%.*]], [[FOR_INC]] ]
12481246
; CHECK-ORDERED-TF-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds float, ptr [[B]], i64 [[IV]]
1249-
; CHECK-ORDERED-TF-NEXT: [[TMP23:%.*]] = load float, ptr [[ARRAYIDX]], align 4
1250-
; CHECK-ORDERED-TF-NEXT: [[TOBOOL:%.*]] = fcmp une float [[TMP23]], 0.000000e+00
1247+
; CHECK-ORDERED-TF-NEXT: [[TMP21:%.*]] = load float, ptr [[ARRAYIDX]], align 4
1248+
; CHECK-ORDERED-TF-NEXT: [[TOBOOL:%.*]] = fcmp une float [[TMP21]], 0.000000e+00
12511249
; CHECK-ORDERED-TF-NEXT: br i1 [[TOBOOL]], label [[IF_THEN:%.*]], label [[FOR_INC]]
12521250
; CHECK-ORDERED-TF: if.then:
12531251
; CHECK-ORDERED-TF-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds float, ptr [[A]], i64 [[IV]]
1254-
; CHECK-ORDERED-TF-NEXT: [[TMP24:%.*]] = load float, ptr [[ARRAYIDX2]], align 4
1252+
; CHECK-ORDERED-TF-NEXT: [[TMP22:%.*]] = load float, ptr [[ARRAYIDX2]], align 4
12551253
; CHECK-ORDERED-TF-NEXT: br label [[FOR_INC]]
12561254
; CHECK-ORDERED-TF: for.inc:
1257-
; CHECK-ORDERED-TF-NEXT: [[PHI:%.*]] = phi float [ [[TMP24]], [[IF_THEN]] ], [ 3.000000e+00, [[FOR_BODY]] ]
1255+
; CHECK-ORDERED-TF-NEXT: [[PHI:%.*]] = phi float [ [[TMP22]], [[IF_THEN]] ], [ 3.000000e+00, [[FOR_BODY]] ]
12581256
; CHECK-ORDERED-TF-NEXT: [[FADD]] = fadd float [[RES]], [[PHI]]
12591257
; CHECK-ORDERED-TF-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
12601258
; CHECK-ORDERED-TF-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
12611259
; CHECK-ORDERED-TF-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP11:![0-9]+]]
12621260
; CHECK-ORDERED-TF: for.end:
1263-
; CHECK-ORDERED-TF-NEXT: [[RDX:%.*]] = phi float [ [[FADD]], [[FOR_INC]] ], [ [[TMP20]], [[MIDDLE_BLOCK]] ]
1261+
; CHECK-ORDERED-TF-NEXT: [[RDX:%.*]] = phi float [ [[FADD]], [[FOR_INC]] ], [ [[TMP18]], [[MIDDLE_BLOCK]] ]
12641262
; CHECK-ORDERED-TF-NEXT: ret float [[RDX]]
12651263
;
12661264

0 commit comments

Comments
 (0)