Skip to content

Commit 08d14e1

Browse files
HanKuanChenHanKuanChen
authored
[SLP] Fix CommonMask will be transformed into an incorrect mask if createShuffle is called multiple times. (#124244)
We have two types of mask in SLP: a scalar mask and a vector mask. When vectorizing four i32 additions into <4 x i32>, SLP creates a mask of length 4. When vectorizing four <2 x i32> additions into <8 x i32>, SLP also creates a mask of length 4. We refer to the first case as a scalar mask (because the mask element represents a scalar, i32), and the second case as a vector mask (because the mask element represents a vector, <4 x i32>). At some point, we must convert the scalar mask into a vector mask (otherwise, calling TTI cost functions or IRBuilderBase functions may yield incorrect results). Since both ShuffleCostEstimator and ShuffleInstructionBuilder can modify the CommonMask, we have decided to perform the mask transformation only within createShuffle. However, we do not store the transformed result, as createShuffle may be called multiple times.
1 parent 79685b5 commit 08d14e1

File tree

3 files changed

+166
-24
lines changed

3 files changed

+166
-24
lines changed

‎llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp

Lines changed: 15 additions & 24 deletions
Original file line numberDiff line numberDiff line change
@@ -9374,8 +9374,15 @@ class BaseShuffleAnalysis {
93749374
/// instruction.
93759375
template <typename T, typename ShuffleBuilderTy>
93769376
static T createShuffle(Value *V1, Value *V2, ArrayRef<int> Mask,
9377-
ShuffleBuilderTy &Builder) {
9377+
ShuffleBuilderTy &Builder, Type *ScalarTy) {
93789378
assert(V1 && "Expected at least one vector value.");
9379+
unsigned ScalarTyNumElements = getNumElements(ScalarTy);
9380+
SmallVector<int> NewMask(Mask);
9381+
if (ScalarTyNumElements != 1) {
9382+
assert(SLPReVec && "FixedVectorType is not expected.");
9383+
transformScalarShuffleIndiciesToVector(ScalarTyNumElements, NewMask);
9384+
Mask = NewMask;
9385+
}
93799386
if (V2)
93809387
Builder.resizeToMatch(V1, V2);
93819388
int VF = Mask.size();
@@ -9478,7 +9485,6 @@ class BaseShuffleAnalysis {
94789485
if (isa<PoisonValue>(V1))
94799486
return Builder.createPoison(
94809487
cast<VectorType>(V1->getType())->getElementType(), Mask.size());
9481-
SmallVector<int> NewMask(Mask);
94829488
bool IsIdentity = peekThroughShuffles(V1, NewMask, /*SinglePermute=*/true);
94839489
assert(V1 && "Expected non-null value after looking through shuffles.");
94849490

@@ -10643,17 +10649,12 @@ class BoUpSLP::ShuffleCostEstimator : public BaseShuffleAnalysis {
1064310649
V2 = getAllOnesValue(*R.DL, getWidenedType(ScalarTy, CommonVF));
1064410650
}
1064510651
}
10646-
if (auto *VecTy = dyn_cast<FixedVectorType>(ScalarTy)) {
10647-
assert(SLPReVec && "FixedVectorType is not expected.");
10648-
transformScalarShuffleIndiciesToVector(VecTy->getNumElements(),
10649-
CommonMask);
10650-
}
1065110652
InVectors.front() =
1065210653
Constant::getNullValue(getWidenedType(ScalarTy, CommonMask.size()));
1065310654
if (InVectors.size() == 2)
1065410655
InVectors.pop_back();
1065510656
return ExtraCost + BaseShuffleAnalysis::createShuffle<InstructionCost>(
10656-
V1, V2, CommonMask, Builder);
10657+
V1, V2, CommonMask, Builder, ScalarTy);
1065710658
}
1065810659

1065910660
public:
@@ -14198,8 +14199,8 @@ class BoUpSLP::ShuffleInstructionBuilder final : public BaseShuffleAnalysis {
1419814199
assert(V1 && "Expected at least one vector value.");
1419914200
ShuffleIRBuilder ShuffleBuilder(Builder, R.GatherShuffleExtractSeq,
1420014201
R.CSEBlocks, *R.DL);
14201-
return BaseShuffleAnalysis::createShuffle<Value *>(V1, V2, Mask,
14202-
ShuffleBuilder);
14202+
return BaseShuffleAnalysis::createShuffle<Value *>(
14203+
V1, V2, Mask, ShuffleBuilder, ScalarTy);
1420314204
}
1420414205

1420514206
/// Cast value \p V to the vector type with the same number of elements, but
@@ -14518,14 +14519,6 @@ class BoUpSLP::ShuffleInstructionBuilder final : public BaseShuffleAnalysis {
1451814519
ArrayRef<int> SubVectorsMask, unsigned VF = 0,
1451914520
function_ref<void(Value *&, SmallVectorImpl<int> &)> Action = {}) {
1452014521
IsFinalized = true;
14521-
unsigned ScalarTyNumElements = getNumElements(ScalarTy);
14522-
SmallVector<int> NewExtMask(ExtMask);
14523-
if (ScalarTyNumElements != 1) {
14524-
assert(SLPReVec && "FixedVectorType is not expected.");
14525-
transformScalarShuffleIndiciesToVector(ScalarTyNumElements, CommonMask);
14526-
transformScalarShuffleIndiciesToVector(ScalarTyNumElements, NewExtMask);
14527-
ExtMask = NewExtMask;
14528-
}
1452914522
if (Action) {
1453014523
Value *Vec = InVectors.front();
1453114524
if (InVectors.size() == 2) {
@@ -14566,17 +14559,15 @@ class BoUpSLP::ShuffleInstructionBuilder final : public BaseShuffleAnalysis {
1456614559
return !isKnownNonNegative(
1456714560
V, SimplifyQuery(*R.DL));
1456814561
}));
14569-
unsigned InsertionIndex = Idx * ScalarTyNumElements;
14562+
unsigned InsertionIndex = Idx * getNumElements(ScalarTy);
1457014563
Vec = createInsertVector(
1457114564
Builder, Vec, V, InsertionIndex,
1457214565
std::bind(&ShuffleInstructionBuilder::createShuffle, this, _1, _2,
1457314566
_3));
1457414567
if (!CommonMask.empty()) {
14575-
std::iota(
14576-
std::next(CommonMask.begin(), InsertionIndex),
14577-
std::next(CommonMask.begin(),
14578-
(Idx + E->getVectorFactor()) * ScalarTyNumElements),
14579-
InsertionIndex);
14568+
std::iota(std::next(CommonMask.begin(), Idx),
14569+
std::next(CommonMask.begin(), Idx + E->getVectorFactor()),
14570+
Idx);
1458014571
}
1458114572
}
1458214573
return Vec;

‎llvm/test/Transforms/SLPVectorizer/RISCV/revec.ll

Lines changed: 37 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -231,3 +231,40 @@ define ptr @test4() {
231231
%28 = tail call float @llvm.sqrt.f32(float %26)
232232
ret ptr null
233233
}
234+
235+
define i32 @test5() {
236+
; CHECK-LABEL: @test5(
237+
; CHECK-NEXT: entry:
238+
; CHECK-NEXT: [[TMP0:%.*]] = call <4 x double> @llvm.vector.insert.v4f64.v2f64(<4 x double> poison, <2 x double> zeroinitializer, i64 0)
239+
; CHECK-NEXT: [[TMP1:%.*]] = call <4 x double> @llvm.vector.insert.v4f64.v2f64(<4 x double> [[TMP0]], <2 x double> zeroinitializer, i64 2)
240+
; CHECK-NEXT: [[TMP2:%.*]] = fdiv <4 x double> [[TMP1]], [[TMP1]]
241+
; CHECK-NEXT: [[TMP3:%.*]] = call <8 x double> @llvm.vector.insert.v8f64.v2f64(<8 x double> poison, <2 x double> zeroinitializer, i64 0)
242+
; CHECK-NEXT: [[TMP4:%.*]] = call <8 x double> @llvm.vector.insert.v8f64.v2f64(<8 x double> [[TMP3]], <2 x double> zeroinitializer, i64 2)
243+
; CHECK-NEXT: [[TMP5:%.*]] = call <8 x double> @llvm.vector.insert.v8f64.v2f64(<8 x double> [[TMP4]], <2 x double> zeroinitializer, i64 4)
244+
; CHECK-NEXT: [[TMP6:%.*]] = call <8 x double> @llvm.vector.insert.v8f64.v2f64(<8 x double> [[TMP5]], <2 x double> zeroinitializer, i64 6)
245+
; CHECK-NEXT: [[TMP7:%.*]] = call <8 x double> @llvm.vector.insert.v8f64.v2f64(<8 x double> poison, <2 x double> zeroinitializer, i64 2)
246+
; CHECK-NEXT: [[TMP8:%.*]] = call <8 x double> @llvm.vector.insert.v8f64.v2f64(<8 x double> [[TMP7]], <2 x double> zeroinitializer, i64 6)
247+
; CHECK-NEXT: [[TMP9:%.*]] = call <8 x double> @llvm.vector.insert.v8f64.v4f64(<8 x double> poison, <4 x double> [[TMP2]], i64 0)
248+
; CHECK-NEXT: [[TMP10:%.*]] = shufflevector <8 x double> [[TMP9]], <8 x double> [[TMP8]], <8 x i32> <i32 0, i32 1, i32 10, i32 11, i32 2, i32 3, i32 14, i32 15>
249+
; CHECK-NEXT: [[TMP11:%.*]] = fadd <8 x double> [[TMP6]], [[TMP10]]
250+
; CHECK-NEXT: br label [[FOR_END47:%.*]]
251+
; CHECK: for.end47:
252+
; CHECK-NEXT: [[TMP12:%.*]] = phi <8 x double> [ [[TMP11]], [[ENTRY:%.*]] ]
253+
; CHECK-NEXT: ret i32 0
254+
;
255+
entry:
256+
%div0 = fdiv <2 x double> zeroinitializer, zeroinitializer
257+
%div1 = fdiv <2 x double> zeroinitializer, zeroinitializer
258+
%add0 = fadd <2 x double> zeroinitializer, %div0
259+
%add1 = fadd <2 x double> zeroinitializer, zeroinitializer
260+
%add2 = fadd <2 x double> %div1, zeroinitializer
261+
%add3 = fadd <2 x double> zeroinitializer, zeroinitializer
262+
br label %for.end47
263+
264+
for.end47: ; preds = %entry
265+
%add0.lcssa = phi <2 x double> [ %add0, %entry ]
266+
%add1.lcssa = phi <2 x double> [ %add1, %entry ]
267+
%add2.lcssa = phi <2 x double> [ %add2, %entry ]
268+
%add3.lcssa = phi <2 x double> [ %add3, %entry ]
269+
ret i32 0
270+
}

‎llvm/test/Transforms/SLPVectorizer/revec-shufflevector.ll

Lines changed: 114 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -121,3 +121,117 @@ entry:
121121
store <4 x i32> %1, ptr %3, align 4
122122
ret void
123123
}
124+
125+
define void @test6(ptr %in0, ptr %in1, ptr %in2) {
126+
; CHECK-LABEL: @test6(
127+
; CHECK-NEXT: entry:
128+
; CHECK-NEXT: [[GEP1:%.*]] = getelementptr inbounds nuw i8, ptr [[IN0:%.*]], i64 32
129+
; CHECK-NEXT: [[LOAD2:%.*]] = load <4 x float>, ptr [[GEP1]], align 16
130+
; CHECK-NEXT: [[TMP0:%.*]] = load <8 x float>, ptr [[IN0]], align 16
131+
; CHECK-NEXT: [[TMP1:%.*]] = load <32 x i8>, ptr [[IN1:%.*]], align 1
132+
; CHECK-NEXT: [[TMP2:%.*]] = uitofp <32 x i8> [[TMP1]] to <32 x float>
133+
; CHECK-NEXT: [[TMP14:%.*]] = shufflevector <8 x float> [[TMP0]], <8 x float> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
134+
; CHECK-NEXT: [[TMP15:%.*]] = shufflevector <4 x float> [[LOAD2]], <4 x float> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
135+
; CHECK-NEXT: [[TMP16:%.*]] = shufflevector <16 x float> [[TMP14]], <16 x float> [[TMP15]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 19, i32 poison, i32 poison, i32 poison, i32 poison>
136+
; CHECK-NEXT: [[TMP3:%.*]] = shufflevector <16 x float> [[TMP16]], <16 x float> poison, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
137+
; CHECK-NEXT: [[TMP4:%.*]] = fmul <32 x float> [[TMP3]], [[TMP2]]
138+
; CHECK-NEXT: store <32 x float> [[TMP4]], ptr [[IN2:%.*]], align 16
139+
; CHECK-NEXT: [[GEP10:%.*]] = getelementptr inbounds nuw i8, ptr [[IN1]], i64 32
140+
; CHECK-NEXT: [[LOAD5:%.*]] = load <16 x i8>, ptr [[GEP10]], align 1
141+
; CHECK-NEXT: [[TMP5:%.*]] = shufflevector <8 x float> [[TMP0]], <8 x float> poison, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
142+
; CHECK-NEXT: [[GEP11:%.*]] = getelementptr inbounds nuw i8, ptr [[IN2]], i64 128
143+
; CHECK-NEXT: [[TMP6:%.*]] = uitofp <16 x i8> [[LOAD5]] to <16 x float>
144+
; CHECK-NEXT: [[TMP7:%.*]] = shufflevector <4 x float> [[LOAD2]], <4 x float> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
145+
; CHECK-NEXT: [[TMP8:%.*]] = shufflevector <8 x float> [[TMP0]], <8 x float> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
146+
; CHECK-NEXT: [[TMP9:%.*]] = shufflevector <16 x float> [[TMP7]], <16 x float> [[TMP8]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
147+
; CHECK-NEXT: [[TMP10:%.*]] = shufflevector <4 x float> [[TMP5]], <4 x float> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
148+
; CHECK-NEXT: [[TMP11:%.*]] = shufflevector <16 x float> [[TMP9]], <16 x float> [[TMP10]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 19, i32 poison, i32 poison, i32 poison, i32 poison>
149+
; CHECK-NEXT: [[TMP12:%.*]] = shufflevector <16 x float> [[TMP11]], <16 x float> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 0, i32 1, i32 2, i32 3>
150+
; CHECK-NEXT: [[TMP13:%.*]] = fmul <16 x float> [[TMP12]], [[TMP6]]
151+
; CHECK-NEXT: store <16 x float> [[TMP13]], ptr [[GEP11]], align 16
152+
; CHECK-NEXT: ret void
153+
;
154+
entry:
155+
%gep0 = getelementptr inbounds i8, ptr %in0, i64 16
156+
%gep1 = getelementptr inbounds i8, ptr %in0, i64 32
157+
%load0 = load <4 x float>, ptr %in0, align 16
158+
%load1 = load <4 x float>, ptr %gep0, align 16
159+
%load2 = load <4 x float>, ptr %gep1, align 16
160+
%gep2 = getelementptr inbounds i8, ptr %in1, i64 16
161+
%load3 = load <16 x i8>, ptr %in1, align 1
162+
%load4 = load <16 x i8>, ptr %gep2, align 1
163+
%shufflevector0 = shufflevector <16 x i8> %load3, <16 x i8> poison, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
164+
%shufflevector1 = shufflevector <16 x i8> %load3, <16 x i8> poison, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
165+
%shufflevector2 = shufflevector <16 x i8> %load4, <16 x i8> poison, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
166+
%shufflevector3 = shufflevector <16 x i8> %load4, <16 x i8> poison, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
167+
%zext0 = zext <8 x i8> %shufflevector0 to <8 x i16>
168+
%zext1 = zext <8 x i8> %shufflevector1 to <8 x i16>
169+
%zext2 = zext <8 x i8> %shufflevector2 to <8 x i16>
170+
%zext3 = zext <8 x i8> %shufflevector3 to <8 x i16>
171+
%shufflevector4 = shufflevector <8 x i16> %zext0, <8 x i16> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
172+
%shufflevector5 = shufflevector <8 x i16> %zext0, <8 x i16> poison, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
173+
%shufflevector6 = shufflevector <8 x i16> %zext1, <8 x i16> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
174+
%shufflevector7 = shufflevector <8 x i16> %zext1, <8 x i16> poison, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
175+
%shufflevector8 = shufflevector <8 x i16> %zext2, <8 x i16> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
176+
%shufflevector9 = shufflevector <8 x i16> %zext2, <8 x i16> poison, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
177+
%shufflevector10 = shufflevector <8 x i16> %zext3, <8 x i16> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
178+
%shufflevector11 = shufflevector <8 x i16> %zext3, <8 x i16> poison, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
179+
%uitofp0 = uitofp nneg <4 x i16> %shufflevector4 to <4 x float>
180+
%uitofp1 = uitofp nneg <4 x i16> %shufflevector5 to <4 x float>
181+
%uitofp2 = uitofp nneg <4 x i16> %shufflevector6 to <4 x float>
182+
%uitofp3 = uitofp nneg <4 x i16> %shufflevector7 to <4 x float>
183+
%uitofp4 = uitofp nneg <4 x i16> %shufflevector8 to <4 x float>
184+
%uitofp5 = uitofp nneg <4 x i16> %shufflevector9 to <4 x float>
185+
%uitofp6 = uitofp nneg <4 x i16> %shufflevector10 to <4 x float>
186+
%uitofp7 = uitofp nneg <4 x i16> %shufflevector11 to <4 x float>
187+
%fmul0 = fmul <4 x float> %load0, %uitofp0
188+
%fmul1 = fmul <4 x float> %load1, %uitofp1
189+
%fmul2 = fmul <4 x float> %load2, %uitofp2
190+
%fmul3 = fmul <4 x float> %load0, %uitofp3
191+
%fmul4 = fmul <4 x float> %load1, %uitofp4
192+
%fmul5 = fmul <4 x float> %load2, %uitofp5
193+
%fmul6 = fmul <4 x float> %load0, %uitofp6
194+
%fmul7 = fmul <4 x float> %load1, %uitofp7
195+
%gep3 = getelementptr inbounds i8, ptr %in2, i64 16
196+
%gep4 = getelementptr inbounds i8, ptr %in2, i64 32
197+
%gep5 = getelementptr inbounds i8, ptr %in2, i64 48
198+
%gep6 = getelementptr inbounds i8, ptr %in2, i64 64
199+
%gep7 = getelementptr inbounds i8, ptr %in2, i64 80
200+
%gep8 = getelementptr inbounds i8, ptr %in2, i64 96
201+
%gep9 = getelementptr inbounds i8, ptr %in2, i64 112
202+
store <4 x float> %fmul0, ptr %in2, align 16
203+
store <4 x float> %fmul1, ptr %gep3, align 16
204+
store <4 x float> %fmul2, ptr %gep4, align 16
205+
store <4 x float> %fmul3, ptr %gep5, align 16
206+
store <4 x float> %fmul4, ptr %gep6, align 16
207+
store <4 x float> %fmul5, ptr %gep7, align 16
208+
store <4 x float> %fmul6, ptr %gep8, align 16
209+
store <4 x float> %fmul7, ptr %gep9, align 16
210+
%gep10 = getelementptr inbounds i8, ptr %in1, i64 32
211+
%load5 = load <16 x i8>, ptr %gep10, align 1
212+
%shufflevector12 = shufflevector <16 x i8> %load5, <16 x i8> poison, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
213+
%shufflevector13 = shufflevector <16 x i8> %load5, <16 x i8> poison, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
214+
%zext4 = zext <8 x i8> %shufflevector12 to <8 x i16>
215+
%zext5 = zext <8 x i8> %shufflevector13 to <8 x i16>
216+
%shufflevector14 = shufflevector <8 x i16> %zext4, <8 x i16> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
217+
%shufflevector15 = shufflevector <8 x i16> %zext4, <8 x i16> poison, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
218+
%shufflevector16 = shufflevector <8 x i16> %zext5, <8 x i16> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
219+
%shufflevector17 = shufflevector <8 x i16> %zext5, <8 x i16> poison, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
220+
%uitofp8 = uitofp nneg <4 x i16> %shufflevector14 to <4 x float>
221+
%uitofp9 = uitofp nneg <4 x i16> %shufflevector15 to <4 x float>
222+
%uitofp10 = uitofp nneg <4 x i16> %shufflevector16 to <4 x float>
223+
%uitofp11 = uitofp nneg <4 x i16> %shufflevector17 to <4 x float>
224+
%fmul8 = fmul <4 x float> %load2, %uitofp8
225+
%fmul9 = fmul <4 x float> %load0, %uitofp9
226+
%fmul10 = fmul <4 x float> %load1, %uitofp10
227+
%fmul11 = fmul <4 x float> %load2, %uitofp11
228+
%gep11 = getelementptr inbounds i8, ptr %in2, i64 128
229+
%gep12 = getelementptr inbounds i8, ptr %in2, i64 144
230+
%gep13 = getelementptr inbounds i8, ptr %in2, i64 160
231+
%gep14 = getelementptr inbounds i8, ptr %in2, i64 176
232+
store <4 x float> %fmul8, ptr %gep11, align 16
233+
store <4 x float> %fmul9, ptr %gep12, align 16
234+
store <4 x float> %fmul10, ptr %gep13, align 16
235+
store <4 x float> %fmul11, ptr %gep14, align 16
236+
ret void
237+
}

0 commit comments

Comments
 (0)