Skip to content

Commit 76d5dd5

Browse files
arcbbbarcbbb
authored
[TTI][RISCV] Add cost modelling for intrinsic vp.load.ff (#169890)
This patch is a rework of #160470 (which was reverted). With getMemIntrinsicCost() now available, we can re‑land the change and reduce vp_load_ff boilerplate.
1 parent 0bd2f12 commit 76d5dd5

File tree

7 files changed

+122
-67
lines changed

7 files changed

+122
-67
lines changed

‎llvm/include/llvm/Analysis/TargetTransformInfo.h‎

Lines changed: 2 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -157,7 +157,8 @@ class MemIntrinsicCostAttributes {
157157
Alignment(Alignment) {}
158158

159159
LLVM_ABI MemIntrinsicCostAttributes(Intrinsic::ID Id, Type *DataTy,
160-
Align Alignment, unsigned AddressSpace)
160+
Align Alignment,
161+
unsigned AddressSpace = 0)
161162
: DataTy(DataTy), IID(Id), AddressSpace(AddressSpace),
162163
Alignment(Alignment) {}
163164

‎llvm/include/llvm/CodeGen/BasicTTIImpl.h‎

Lines changed: 12 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1817,7 +1817,16 @@ class BasicTTIImplBase : public TargetTransformInfoImplCRTPBase<T> {
18171817
}
18181818
}
18191819
}
1820-
1820+
if (ICA.getID() == Intrinsic::vp_load_ff) {
1821+
Type *RetTy = ICA.getReturnType();
1822+
Type *DataTy = cast<StructType>(RetTy)->getElementType(0);
1823+
Align Alignment;
1824+
if (auto *VPI = dyn_cast_or_null<VPIntrinsic>(ICA.getInst()))
1825+
Alignment = VPI->getPointerAlignment().valueOrOne();
1826+
return thisT()->getMemIntrinsicInstrCost(
1827+
MemIntrinsicCostAttributes(ICA.getID(), DataTy, Alignment),
1828+
CostKind);
1829+
}
18211830
if (ICA.getID() == Intrinsic::vp_scatter) {
18221831
if (ICA.isTypeBasedOnly()) {
18231832
IntrinsicCostAttributes MaskedScatter(
@@ -3076,6 +3085,8 @@ class BasicTTIImplBase : public TargetTransformInfoImplCRTPBase<T> {
30763085
case Intrinsic::masked_compressstore:
30773086
case Intrinsic::masked_expandload:
30783087
return thisT()->getExpandCompressMemoryOpCost(MICA, CostKind);
3088+
case Intrinsic::vp_load_ff:
3089+
return InstructionCost::getInvalid();
30793090
default:
30803091
llvm_unreachable("unexpected intrinsic");
30813092
}

‎llvm/lib/Target/RISCV/RISCVISelLowering.cpp‎

Lines changed: 16 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -25364,6 +25364,22 @@ bool RISCVTargetLowering::isLegalStridedLoadStore(EVT DataType,
2536425364
return true;
2536525365
}
2536625366

25367+
bool RISCVTargetLowering::isLegalFirstFaultLoad(EVT DataType,
25368+
Align Alignment) const {
25369+
if (!Subtarget.hasVInstructions())
25370+
return false;
25371+
25372+
EVT ScalarType = DataType.getScalarType();
25373+
if (!isLegalElementTypeForRVV(ScalarType))
25374+
return false;
25375+
25376+
if (!Subtarget.enableUnalignedVectorMem() &&
25377+
Alignment < ScalarType.getStoreSize())
25378+
return false;
25379+
25380+
return true;
25381+
}
25382+
2536725383
MachineInstr *
2536825384
RISCVTargetLowering::EmitKCFICheck(MachineBasicBlock &MBB,
2536925385
MachineBasicBlock::instr_iterator &MBBI,

‎llvm/lib/Target/RISCV/RISCVISelLowering.h‎

Lines changed: 4 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -429,6 +429,10 @@ class RISCVTargetLowering : public TargetLowering {
429429
/// alignment is legal.
430430
bool isLegalStridedLoadStore(EVT DataType, Align Alignment) const;
431431

432+
/// Return true if a fault-only-first load of the given result type and
433+
/// alignment is legal.
434+
bool isLegalFirstFaultLoad(EVT DataType, Align Alignment) const;
435+
432436
unsigned getMaxSupportedInterleaveFactor() const override { return 8; }
433437

434438
bool fallBackToDAGISel(const Instruction &Inst) const override;

‎llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp‎

Lines changed: 19 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -1007,6 +1007,25 @@ InstructionCost RISCVTTIImpl::getScalarizationOverhead(
10071007
return Cost;
10081008
}
10091009

1010+
InstructionCost
1011+
RISCVTTIImpl::getMemIntrinsicInstrCost(const MemIntrinsicCostAttributes &MICA,
1012+
TTI::TargetCostKind CostKind) const {
1013+
Type *DataTy = MICA.getDataType();
1014+
Align Alignment = MICA.getAlignment();
1015+
switch (MICA.getID()) {
1016+
case Intrinsic::vp_load_ff: {
1017+
EVT DataTypeVT = TLI->getValueType(DL, DataTy);
1018+
if (!TLI->isLegalFirstFaultLoad(DataTypeVT, Alignment))
1019+
return BaseT::getMemIntrinsicInstrCost(MICA, CostKind);
1020+
1021+
unsigned AS = MICA.getAddressSpace();
1022+
return getMemoryOpCost(Instruction::Load, DataTy, Alignment, AS, CostKind,
1023+
{TTI::OK_AnyValue, TTI::OP_None}, nullptr);
1024+
}
1025+
}
1026+
return BaseT::getMemIntrinsicInstrCost(MICA, CostKind);
1027+
}
1028+
10101029
InstructionCost
10111030
RISCVTTIImpl::getMaskedMemoryOpCost(const MemIntrinsicCostAttributes &MICA,
10121031
TTI::TargetCostKind CostKind) const {

‎llvm/lib/Target/RISCV/RISCVTargetTransformInfo.h‎

Lines changed: 4 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -143,6 +143,10 @@ class RISCVTTIImpl final : public BasicTTIImplBase<RISCVTTIImpl> {
143143

144144
bool shouldConsiderVectorizationRegPressure() const override { return true; }
145145

146+
InstructionCost
147+
getMemIntrinsicInstrCost(const MemIntrinsicCostAttributes &MICA,
148+
TTI::TargetCostKind CostKind) const override;
149+
146150
InstructionCost
147151
getMaskedMemoryOpCost(const MemIntrinsicCostAttributes &MICA,
148152
TTI::TargetCostKind CostKind) const override;

‎llvm/test/Analysis/CostModel/RISCV/vp-intrinsics.ll‎

Lines changed: 65 additions & 65 deletions
Original file line numberDiff line numberDiff line change
@@ -836,74 +836,74 @@ define void @abs() {
836836
ret void
837837
}
838838

839-
define void @load() {
839+
define void @load(ptr %src) {
840840
; CHECK-LABEL: 'load'
841-
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %t0 = call <2 x i8> @llvm.vp.load.v2i8.p0(ptr undef, <2 x i1> undef, i32 undef)
842-
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %t1 = load <2 x i8>, ptr undef, align 2
843-
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %t2 = call <4 x i8> @llvm.vp.load.v4i8.p0(ptr undef, <4 x i1> undef, i32 undef)
844-
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %t3 = load <4 x i8>, ptr undef, align 4
845-
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %t4 = call <8 x i8> @llvm.vp.load.v8i8.p0(ptr undef, <8 x i1> undef, i32 undef)
846-
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %t5 = load <8 x i8>, ptr undef, align 8
847-
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %t6 = call <16 x i8> @llvm.vp.load.v16i8.p0(ptr undef, <16 x i1> undef, i32 undef)
848-
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %t7 = load <16 x i8>, ptr undef, align 16
849-
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %t8 = call <2 x i64> @llvm.vp.load.v2i64.p0(ptr undef, <2 x i1> undef, i32 undef)
850-
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %t9 = load <2 x i64>, ptr undef, align 16
851-
; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %t10 = call <4 x i64> @llvm.vp.load.v4i64.p0(ptr undef, <4 x i1> undef, i32 undef)
852-
; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %t12 = load <4 x i64>, ptr undef, align 32
853-
; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %t13 = call <8 x i64> @llvm.vp.load.v8i64.p0(ptr undef, <8 x i1> undef, i32 undef)
854-
; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %t14 = load <8 x i64>, ptr undef, align 64
855-
; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %t15 = call <16 x i64> @llvm.vp.load.v16i64.p0(ptr undef, <16 x i1> undef, i32 undef)
856-
; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %t16 = load <16 x i64>, ptr undef, align 128
857-
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %t17 = call <vscale x 2 x i8> @llvm.vp.load.nxv2i8.p0(ptr undef, <vscale x 2 x i1> undef, i32 undef)
858-
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %t18 = load <vscale x 2 x i8>, ptr undef, align 2
859-
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %t19 = call <vscale x 4 x i8> @llvm.vp.load.nxv4i8.p0(ptr undef, <vscale x 4 x i1> undef, i32 undef)
860-
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %t20 = load <vscale x 4 x i8>, ptr undef, align 4
861-
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %t21 = call <vscale x 8 x i8> @llvm.vp.load.nxv8i8.p0(ptr undef, <vscale x 8 x i1> undef, i32 undef)
862-
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %t22 = load <vscale x 8 x i8>, ptr undef, align 8
863-
; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %t23 = call <vscale x 16 x i8> @llvm.vp.load.nxv16i8.p0(ptr undef, <vscale x 16 x i1> undef, i32 undef)
864-
; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %t24 = load <vscale x 16 x i8>, ptr undef, align 16
865-
; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %t25 = call <vscale x 2 x i64> @llvm.vp.load.nxv2i64.p0(ptr undef, <vscale x 2 x i1> undef, i32 undef)
866-
; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %t26 = load <vscale x 2 x i64>, ptr undef, align 16
867-
; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %t27 = call <vscale x 4 x i64> @llvm.vp.load.nxv4i64.p0(ptr undef, <vscale x 4 x i1> undef, i32 undef)
868-
; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %t28 = load <vscale x 4 x i64>, ptr undef, align 32
869-
; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %t29 = call <vscale x 8 x i64> @llvm.vp.load.nxv8i64.p0(ptr undef, <vscale x 8 x i1> undef, i32 undef)
870-
; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %t30 = load <vscale x 8 x i64>, ptr undef, align 64
871-
; CHECK-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %t31 = call <vscale x 16 x i64> @llvm.vp.load.nxv16i64.p0(ptr undef, <vscale x 16 x i1> undef, i32 undef)
872-
; CHECK-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %t32 = load <vscale x 16 x i64>, ptr undef, align 128
841+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %t0 = call <2 x i8> @llvm.vp.load.v2i8.p0(ptr %src, <2 x i1> undef, i32 undef)
842+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %t1 = call { <2 x i8>, i32 } @llvm.vp.load.ff.v2i8.p0(ptr align 1 %src, <2 x i1> undef, i32 undef)
843+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %t2 = call <4 x i8> @llvm.vp.load.v4i8.p0(ptr %src, <4 x i1> undef, i32 undef)
844+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %t3 = call { <4 x i8>, i32 } @llvm.vp.load.ff.v4i8.p0(ptr align 1 %src, <4 x i1> undef, i32 undef)
845+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %t4 = call <8 x i8> @llvm.vp.load.v8i8.p0(ptr %src, <8 x i1> undef, i32 undef)
846+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %t5 = call { <8 x i8>, i32 } @llvm.vp.load.ff.v8i8.p0(ptr align 1 %src, <8 x i1> undef, i32 undef)
847+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %t6 = call <16 x i8> @llvm.vp.load.v16i8.p0(ptr %src, <16 x i1> undef, i32 undef)
848+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %t7 = call { <16 x i8>, i32 } @llvm.vp.load.ff.v16i8.p0(ptr align 1 %src, <16 x i1> undef, i32 undef)
849+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %t8 = call <2 x i64> @llvm.vp.load.v2i64.p0(ptr %src, <2 x i1> undef, i32 undef)
850+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %t9 = call { <2 x i64>, i32 } @llvm.vp.load.ff.v2i64.p0(ptr align 8 %src, <2 x i1> undef, i32 undef)
851+
; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %t10 = call <4 x i64> @llvm.vp.load.v4i64.p0(ptr %src, <4 x i1> undef, i32 undef)
852+
; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %t11 = call { <4 x i64>, i32 } @llvm.vp.load.ff.v4i64.p0(ptr align 8 %src, <4 x i1> undef, i32 undef)
853+
; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %t12 = call <8 x i64> @llvm.vp.load.v8i64.p0(ptr %src, <8 x i1> undef, i32 undef)
854+
; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %t13 = call { <8 x i64>, i32 } @llvm.vp.load.ff.v8i64.p0(ptr align 8 %src, <8 x i1> undef, i32 undef)
855+
; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %t14 = call <16 x i64> @llvm.vp.load.v16i64.p0(ptr %src, <16 x i1> undef, i32 undef)
856+
; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %t15 = call { <16 x i64>, i32 } @llvm.vp.load.ff.v16i64.p0(ptr align 8 %src, <16 x i1> undef, i32 undef)
857+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %t16 = call <vscale x 2 x i8> @llvm.vp.load.nxv2i8.p0(ptr %src, <vscale x 2 x i1> undef, i32 undef)
858+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %t17 = call { <vscale x 2 x i8>, i32 } @llvm.vp.load.ff.nxv2i8.p0(ptr align 1 %src, <vscale x 2 x i1> undef, i32 undef)
859+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %t18 = call <vscale x 4 x i8> @llvm.vp.load.nxv4i8.p0(ptr %src, <vscale x 4 x i1> undef, i32 undef)
860+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %t19 = call { <vscale x 4 x i8>, i32 } @llvm.vp.load.ff.nxv4i8.p0(ptr align 1 %src, <vscale x 4 x i1> undef, i32 undef)
861+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %t20 = call <vscale x 8 x i8> @llvm.vp.load.nxv8i8.p0(ptr %src, <vscale x 8 x i1> undef, i32 undef)
862+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %t21 = call { <vscale x 8 x i8>, i32 } @llvm.vp.load.ff.nxv8i8.p0(ptr align 1 %src, <vscale x 8 x i1> undef, i32 undef)
863+
; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %t22 = call <vscale x 16 x i8> @llvm.vp.load.nxv16i8.p0(ptr %src, <vscale x 16 x i1> undef, i32 undef)
864+
; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %t23 = call { <vscale x 16 x i8>, i32 } @llvm.vp.load.ff.nxv16i8.p0(ptr align 1 %src, <vscale x 16 x i1> undef, i32 undef)
865+
; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %t24 = call <vscale x 2 x i64> @llvm.vp.load.nxv2i64.p0(ptr %src, <vscale x 2 x i1> undef, i32 undef)
866+
; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %t25 = call { <vscale x 2 x i64>, i32 } @llvm.vp.load.ff.nxv2i64.p0(ptr align 8 %src, <vscale x 2 x i1> undef, i32 undef)
867+
; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %t26 = call <vscale x 4 x i64> @llvm.vp.load.nxv4i64.p0(ptr %src, <vscale x 4 x i1> undef, i32 undef)
868+
; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %t27 = call { <vscale x 4 x i64>, i32 } @llvm.vp.load.ff.nxv4i64.p0(ptr align 8 %src, <vscale x 4 x i1> undef, i32 undef)
869+
; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %t28 = call <vscale x 8 x i64> @llvm.vp.load.nxv8i64.p0(ptr %src, <vscale x 8 x i1> undef, i32 undef)
870+
; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %t29 = call { <vscale x 8 x i64>, i32 } @llvm.vp.load.ff.nxv8i64.p0(ptr align 8 %src, <vscale x 8 x i1> undef, i32 undef)
871+
; CHECK-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %t30 = call <vscale x 16 x i64> @llvm.vp.load.nxv16i64.p0(ptr %src, <vscale x 16 x i1> undef, i32 undef)
872+
; CHECK-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %t31 = call { <vscale x 16 x i64>, i32 } @llvm.vp.load.ff.nxv16i64.p0(ptr align 8 %src, <vscale x 16 x i1> undef, i32 undef)
873873
; CHECK-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
874874
;
875-
%t0 = call <2 x i8> @llvm.vp.load.v2i8(ptr undef, <2 x i1> undef, i32 undef)
876-
%t1 = load <2 x i8>, ptr undef
877-
%t2 = call <4 x i8> @llvm.vp.load.v4i8(ptr undef, <4 x i1> undef, i32 undef)
878-
%t3 = load <4 x i8>, ptr undef
879-
%t4 = call <8 x i8> @llvm.vp.load.v8i8(ptr undef, <8 x i1> undef, i32 undef)
880-
%t5 = load <8 x i8>, ptr undef
881-
%t6 = call <16 x i8> @llvm.vp.load.v16i8(ptr undef, <16 x i1> undef, i32 undef)
882-
%t7 = load <16 x i8>, ptr undef
883-
%t8 = call <2 x i64> @llvm.vp.load.v2i64(ptr undef, <2 x i1> undef, i32 undef)
884-
%t9 = load <2 x i64>, ptr undef
885-
%t10 = call <4 x i64> @llvm.vp.load.v4i64(ptr undef, <4 x i1> undef, i32 undef)
886-
%t12 = load <4 x i64>, ptr undef
887-
%t13 = call <8 x i64> @llvm.vp.load.v8i64(ptr undef, <8 x i1> undef, i32 undef)
888-
%t14 = load <8 x i64>, ptr undef
889-
%t15 = call <16 x i64> @llvm.vp.load.v16i64(ptr undef, <16 x i1> undef, i32 undef)
890-
%t16 = load <16 x i64>, ptr undef
891-
%t17 = call <vscale x 2 x i8> @llvm.vp.load.nxv2i8(ptr undef, <vscale x 2 x i1> undef, i32 undef)
892-
%t18 = load <vscale x 2 x i8>, ptr undef
893-
%t19 = call <vscale x 4 x i8> @llvm.vp.load.nxv4i8(ptr undef, <vscale x 4 x i1> undef, i32 undef)
894-
%t20 = load <vscale x 4 x i8>, ptr undef
895-
%t21 = call <vscale x 8 x i8> @llvm.vp.load.nxv8i8(ptr undef, <vscale x 8 x i1> undef, i32 undef)
896-
%t22 = load <vscale x 8 x i8>, ptr undef
897-
%t23 = call <vscale x 16 x i8> @llvm.vp.load.nxv16i8(ptr undef, <vscale x 16 x i1> undef, i32 undef)
898-
%t24 = load <vscale x 16 x i8>, ptr undef
899-
%t25 = call <vscale x 2 x i64> @llvm.vp.load.nxv2i64(ptr undef, <vscale x 2 x i1> undef, i32 undef)
900-
%t26 = load <vscale x 2 x i64>, ptr undef
901-
%t27 = call <vscale x 4 x i64> @llvm.vp.load.nxv4i64(ptr undef, <vscale x 4 x i1> undef, i32 undef)
902-
%t28 = load <vscale x 4 x i64>, ptr undef
903-
%t29 = call <vscale x 8 x i64> @llvm.vp.load.nxv8i64(ptr undef, <vscale x 8 x i1> undef, i32 undef)
904-
%t30 = load <vscale x 8 x i64>, ptr undef
905-
%t31 = call <vscale x 16 x i64> @llvm.vp.load.nxv16i64(ptr undef, <vscale x 16 x i1> undef, i32 undef)
906-
%t32 = load <vscale x 16 x i64>, ptr undef
875+
%t0 = call <2 x i8> @llvm.vp.load.v2i8(ptr %src, <2 x i1> undef, i32 undef)
876+
%t1 = call { <2 x i8>, i32 } @llvm.vp.load.ff.v2i8.p0(ptr align 1 %src, <2 x i1> undef, i32 undef)
877+
%t2 = call <4 x i8> @llvm.vp.load.v4i8(ptr %src, <4 x i1> undef, i32 undef)
878+
%t3 = call { <4 x i8>, i32 } @llvm.vp.load.ff.v4i8.p0(ptr align 1 %src, <4 x i1> undef, i32 undef)
879+
%t4 = call <8 x i8> @llvm.vp.load.v8i8(ptr %src, <8 x i1> undef, i32 undef)
880+
%t5 = call { <8 x i8>, i32 } @llvm.vp.load.ff.v8i8.p0(ptr align 1 %src, <8 x i1> undef, i32 undef)
881+
%t6 = call <16 x i8> @llvm.vp.load.v16i8(ptr %src, <16 x i1> undef, i32 undef)
882+
%t7 = call { <16 x i8>, i32 } @llvm.vp.load.ff.v16i8.p0(ptr align 1 %src, <16 x i1> undef, i32 undef)
883+
%t8 = call <2 x i64> @llvm.vp.load.v2i64(ptr %src, <2 x i1> undef, i32 undef)
884+
%t9 = call { <2 x i64>, i32 } @llvm.vp.load.ff.v2i64.p0(ptr align 8 %src, <2 x i1> undef, i32 undef)
885+
%t10 = call <4 x i64> @llvm.vp.load.v4i64(ptr %src, <4 x i1> undef, i32 undef)
886+
%t11 = call { <4 x i64>, i32 } @llvm.vp.load.ff.v4i64.p0(ptr align 8 %src, <4 x i1> undef, i32 undef)
887+
%t12 = call <8 x i64> @llvm.vp.load.v8i64(ptr %src, <8 x i1> undef, i32 undef)
888+
%t13 = call { <8 x i64>, i32 } @llvm.vp.load.ff.v8i64.p0(ptr align 8 %src, <8 x i1> undef, i32 undef)
889+
%t14 = call <16 x i64> @llvm.vp.load.v16i64(ptr %src, <16 x i1> undef, i32 undef)
890+
%t15 = call { <16 x i64>, i32 } @llvm.vp.load.ff.v16i64.p0(ptr align 8 %src, <16 x i1> undef, i32 undef)
891+
%t16 = call <vscale x 2 x i8> @llvm.vp.load.nxv2i8(ptr %src, <vscale x 2 x i1> undef, i32 undef)
892+
%t17 = call { <vscale x 2 x i8>, i32 } @llvm.vp.load.ff.nxv2i8.p0(ptr align 1 %src, <vscale x 2 x i1> undef, i32 undef)
893+
%t18 = call <vscale x 4 x i8> @llvm.vp.load.nxv4i8(ptr %src, <vscale x 4 x i1> undef, i32 undef)
894+
%t19 = call { <vscale x 4 x i8>, i32 } @llvm.vp.load.ff.nxv4i8.p0(ptr align 1 %src, <vscale x 4 x i1> undef, i32 undef)
895+
%t20 = call <vscale x 8 x i8> @llvm.vp.load.nxv8i8(ptr %src, <vscale x 8 x i1> undef, i32 undef)
896+
%t21 = call { <vscale x 8 x i8>, i32 } @llvm.vp.load.ff.nxv8i8.p0(ptr align 1 %src, <vscale x 8 x i1> undef, i32 undef)
897+
%t22 = call <vscale x 16 x i8> @llvm.vp.load.nxv16i8(ptr %src, <vscale x 16 x i1> undef, i32 undef)
898+
%t23 = call { <vscale x 16 x i8>, i32 } @llvm.vp.load.ff.nxv16i8.p0(ptr align 1 %src, <vscale x 16 x i1> undef, i32 undef)
899+
%t24 = call <vscale x 2 x i64> @llvm.vp.load.nxv2i64(ptr %src, <vscale x 2 x i1> undef, i32 undef)
900+
%t25 = call { <vscale x 2 x i64>, i32 } @llvm.vp.load.ff.nxv2i64.p0(ptr align 8 %src, <vscale x 2 x i1> undef, i32 undef)
901+
%t26 = call <vscale x 4 x i64> @llvm.vp.load.nxv4i64(ptr %src, <vscale x 4 x i1> undef, i32 undef)
902+
%t27 = call { <vscale x 4 x i64>, i32 } @llvm.vp.load.ff.nxv4i64.p0(ptr align 8 %src, <vscale x 4 x i1> undef, i32 undef)
903+
%t28 = call <vscale x 8 x i64> @llvm.vp.load.nxv8i64(ptr %src, <vscale x 8 x i1> undef, i32 undef)
904+
%t29 = call { <vscale x 8 x i64>, i32 } @llvm.vp.load.ff.nxv8i64.p0(ptr align 8 %src, <vscale x 8 x i1> undef, i32 undef)
905+
%t30 = call <vscale x 16 x i64> @llvm.vp.load.nxv16i64(ptr %src, <vscale x 16 x i1> undef, i32 undef)
906+
%t31 = call { <vscale x 16 x i64>, i32 } @llvm.vp.load.ff.nxv16i64.p0(ptr align 8 %src, <vscale x 16 x i1> undef, i32 undef)
907907
ret void
908908
}
909909

0 commit comments

Comments
 (0)