Skip to content

Commit eece6ba

Browse files
arsenmarsenm
committed
IR: Add llvm.ldexp and llvm.experimental.constrained.ldexp intrinsics
AMDGPU has native instructions and target intrinsics for this, but these really should be subject to legalization and generic optimizations. This will enable legalization of f16->f32 on targets without f16 support. Implement a somewhat horrible inline expansion for targets without libcall support. This could be better if we could introduce control flow (GlobalISel version not yet implemented). Support for strictfp legalization is less complete but works for the simple cases.
1 parent 5d361ad commit eece6ba

File tree

71 files changed

+3780
-422
lines changed

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

71 files changed

+3780
-422
lines changed

‎clang/lib/CodeGen/CGBuiltin.cpp

Lines changed: 7 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -17146,8 +17146,13 @@ Value *CodeGenFunction::EmitAMDGPUBuiltinExpr(unsigned BuiltinID,
1714617146
return emitUnaryBuiltin(*this, E, Intrinsic::amdgcn_log_clamp);
1714717147
case AMDGPU::BI__builtin_amdgcn_ldexp:
1714817148
case AMDGPU::BI__builtin_amdgcn_ldexpf:
17149-
case AMDGPU::BI__builtin_amdgcn_ldexph:
17150-
return emitFPIntBuiltin(*this, E, Intrinsic::amdgcn_ldexp);
17149+
case AMDGPU::BI__builtin_amdgcn_ldexph: {
17150+
llvm::Value *Src0 = EmitScalarExpr(E->getArg(0));
17151+
llvm::Value *Src1 = EmitScalarExpr(E->getArg(1));
17152+
llvm::Function *F =
17153+
CGM.getIntrinsic(Intrinsic::ldexp, {Src0->getType(), Src1->getType()});
17154+
return Builder.CreateCall(F, {Src0, Src1});
17155+
}
1715117156
case AMDGPU::BI__builtin_amdgcn_frexp_mant:
1715217157
case AMDGPU::BI__builtin_amdgcn_frexp_mantf:
1715317158
case AMDGPU::BI__builtin_amdgcn_frexp_manth:

‎clang/test/CodeGenOpenCL/builtins-amdgcn-vi.cl

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -52,7 +52,7 @@ void test_cos_f16(global half* out, half a)
5252
}
5353

5454
// CHECK-LABEL: @test_ldexp_f16
55-
// CHECK: call half @llvm.amdgcn.ldexp.f16
55+
// CHECK: call half @llvm.ldexp.f16.i32
5656
void test_ldexp_f16(global half* out, half a, int b)
5757
{
5858
*out = __builtin_amdgcn_ldexph(a, b);

‎clang/test/CodeGenOpenCL/builtins-amdgcn.cl

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -180,14 +180,14 @@ void test_log_clamp_f32(global float* out, float a)
180180
}
181181

182182
// CHECK-LABEL: @test_ldexp_f32
183-
// CHECK: call float @llvm.amdgcn.ldexp.f32
183+
// CHECK: call float @llvm.ldexp.f32.i32
184184
void test_ldexp_f32(global float* out, float a, int b)
185185
{
186186
*out = __builtin_amdgcn_ldexpf(a, b);
187187
}
188188

189189
// CHECK-LABEL: @test_ldexp_f64
190-
// CHECK: call double @llvm.amdgcn.ldexp.f64
190+
// CHECK: call double @llvm.ldexp.f64.i32
191191
void test_ldexp_f64(global double* out, double a, int b)
192192
{
193193
*out = __builtin_amdgcn_ldexp(a, b);

‎llvm/docs/LangRef.rst

Lines changed: 82 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -14713,6 +14713,47 @@ trapping or setting ``errno``.
1471314713
When specified with the fast-math-flag 'afn', the result may be approximated
1471414714
using a less accurate calculation.
1471514715

14716+
'``llvm.ldexp.*``' Intrinsic
14717+
^^^^^^^^^^^^^^^^^^^^^^^^^^^
14718+
14719+
Syntax:
14720+
"""""""
14721+
14722+
This is an overloaded intrinsic. You can use ``llvm.ldexp`` on any
14723+
floating point or vector of floating point type. Not all targets support
14724+
all types however.
14725+
14726+
::
14727+
14728+
declare float @llvm.ldexp.f32.i32(float %Val, i32 %Exp)
14729+
declare double @llvm.ldexp.f64.i32(double %Val, i32 %Exp)
14730+
declare x86_fp80 @llvm.ldexp.f80.i32(x86_fp80 %Val, i32 %Exp)
14731+
declare fp128 @llvm.ldexp.f128.i32(fp128 %Val, i32 %Exp)
14732+
declare ppc_fp128 @llvm.ldexp.ppcf128.i32(ppc_fp128 %Val, i32 %Exp)
14733+
declare <2 x float> @llvm.ldexp.v2f32.v2i32(<2 x float> %Val, <2 x i32> %Exp)
14734+
14735+
Overview:
14736+
"""""""""
14737+
14738+
The '``llvm.ldexp.*``' intrinsics perform the ldexp function.
14739+
14740+
Arguments:
14741+
""""""""""
14742+
14743+
The first argument and the return value are :ref:`floating-point
14744+
<t_floating>` or :ref:`vector <t_vector>` of floating-point values of
14745+
the same type. The second argument is an integer with the same number
14746+
of elements.
14747+
14748+
Semantics:
14749+
""""""""""
14750+
14751+
This function multiplies the first argument by 2 raised to the second
14752+
argument's power. If the first argument is NaN or infinite, the same
14753+
value is returned. If the result underflows a zero with the same sign
14754+
is returned. If the result overflows, the result is an infinity with
14755+
the same sign.
14756+
1471614757
'``llvm.log.*``' Intrinsic
1471714758
^^^^^^^^^^^^^^^^^^^^^^^^^^
1471814759

@@ -24306,6 +24347,47 @@ This function returns the first value raised to the second power with an
2430624347
unspecified sequence of rounding operations.
2430724348

2430824349

24350+
'``llvm.experimental.constrained.ldexp``' Intrinsic
24351+
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
24352+
24353+
Syntax:
24354+
"""""""
24355+
24356+
::
24357+
24358+
declare <type0>
24359+
@llvm.experimental.constrained.ldexp(<type0> <op1>, <type1> <op2>,
24360+
metadata <rounding mode>,
24361+
metadata <exception behavior>)
24362+
24363+
Overview:
24364+
"""""""""
24365+
24366+
The '``llvm.experimental.constrained.ldexp``' performs the ldexp function.
24367+
24368+
24369+
Arguments:
24370+
""""""""""
24371+
24372+
The first argument and the return value are :ref:`floating-point
24373+
<t_floating>` or :ref:`vector <t_vector>` of floating-point values of
24374+
the same type. The second argument is an integer with the same number
24375+
of elements.
24376+
24377+
24378+
The third and fourth arguments specify the rounding mode and exception
24379+
behavior as described above.
24380+
24381+
Semantics:
24382+
""""""""""
24383+
24384+
This function multiplies the first argument by 2 raised to the second
24385+
argument's power. If the first argument is NaN or infinite, the same
24386+
value is returned. If the result underflows a zero with the same sign
24387+
is returned. If the result overflows, the result is an infinity with
24388+
the same sign.
24389+
24390+
2430924391
'``llvm.experimental.constrained.sin``' Intrinsic
2431024392
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2431124393

‎llvm/docs/ReleaseNotes.rst

Lines changed: 2 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -61,6 +61,8 @@ Changes to the LLVM IR
6161
* The ``nofpclass`` attribute was introduced. This allows more
6262
optimizations around special floating point value comparisons.
6363

64+
* Introduced new ``llvm.ldexp`` and ``llvm.experimental.constrained.ldexp`` intrinsics.
65+
6466
* The constant expression variants of the following instructions have been
6567
removed:
6668

‎llvm/include/llvm/Analysis/TargetLibraryInfo.h

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -378,6 +378,7 @@ class TargetLibraryInfo {
378378
case LibFunc_trunc: case LibFunc_truncf: case LibFunc_truncl:
379379
case LibFunc_log2: case LibFunc_log2f: case LibFunc_log2l:
380380
case LibFunc_exp2: case LibFunc_exp2f: case LibFunc_exp2l:
381+
case LibFunc_ldexp: case LibFunc_ldexpf: case LibFunc_ldexpl:
381382
case LibFunc_memcpy: case LibFunc_memset: case LibFunc_memmove:
382383
case LibFunc_memcmp: case LibFunc_bcmp: case LibFunc_strcmp:
383384
case LibFunc_strcpy: case LibFunc_stpcpy: case LibFunc_strlen:

‎llvm/include/llvm/CodeGen/GlobalISel/LegalizerHelper.h

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -357,6 +357,7 @@ class LegalizerHelper {
357357
LegalizeResult narrowScalarCTLZ(MachineInstr &MI, unsigned TypeIdx, LLT Ty);
358358
LegalizeResult narrowScalarCTTZ(MachineInstr &MI, unsigned TypeIdx, LLT Ty);
359359
LegalizeResult narrowScalarCTPOP(MachineInstr &MI, unsigned TypeIdx, LLT Ty);
360+
LegalizeResult narrowScalarFLDEXP(MachineInstr &MI, unsigned TypeIdx, LLT Ty);
360361

361362
/// Perform Bitcast legalize action on G_EXTRACT_VECTOR_ELT.
362363
LegalizeResult bitcastExtractVectorElt(MachineInstr &MI, unsigned TypeIdx,

‎llvm/include/llvm/CodeGen/GlobalISel/MachineIRBuilder.h

Lines changed: 7 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -1825,6 +1825,13 @@ class MachineIRBuilder {
18251825
return buildInstr(TargetOpcode::G_FPOW, {Dst}, {Src0, Src1}, Flags);
18261826
}
18271827

1828+
/// Build and insert \p Dst = G_FLDEXP \p Src0, \p Src1
1829+
MachineInstrBuilder
1830+
buildFLdexp(const DstOp &Dst, const SrcOp &Src0, const SrcOp &Src1,
1831+
std::optional<unsigned> Flags = std::nullopt) {
1832+
return buildInstr(TargetOpcode::G_FLDEXP, {Dst}, {Src0, Src1}, Flags);
1833+
}
1834+
18281835
/// Build and insert \p Res = G_FCOPYSIGN \p Op0, \p Op1
18291836
MachineInstrBuilder buildFCopysign(const DstOp &Dst, const SrcOp &Src0,
18301837
const SrcOp &Src1) {

‎llvm/include/llvm/CodeGen/ISDOpcodes.h

Lines changed: 4 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -411,6 +411,7 @@ enum NodeType {
411411
STRICT_FSQRT,
412412
STRICT_FPOW,
413413
STRICT_FPOWI,
414+
STRICT_FLDEXP,
414415
STRICT_FSIN,
415416
STRICT_FCOS,
416417
STRICT_FEXP,
@@ -926,8 +927,10 @@ enum NodeType {
926927
FCBRT,
927928
FSIN,
928929
FCOS,
929-
FPOWI,
930930
FPOW,
931+
FPOWI,
932+
/// FLDEXP - ldexp, inspired by libm (op0 * 2**op1).
933+
FLDEXP,
931934
FLOG,
932935
FLOG2,
933936
FLOG10,

‎llvm/include/llvm/CodeGen/RuntimeLibcalls.h

Lines changed: 4 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -70,6 +70,10 @@ namespace RTLIB {
7070
/// UNKNOWN_LIBCALL if there is none.
7171
Libcall getPOWI(EVT RetVT);
7272

73+
/// getLDEXP - Return the LDEXP_* value for the given types, or
74+
/// UNKNOWN_LIBCALL if there is none.
75+
Libcall getLDEXP(EVT RetVT);
76+
7377
/// Return the SYNC_FETCH_AND_* value for the given opcode and type, or
7478
/// UNKNOWN_LIBCALL if there is none.
7579
Libcall getSYNC(unsigned Opc, MVT VT);

‎llvm/include/llvm/IR/ConstrainedOps.def

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -89,6 +89,7 @@ DAG_FUNCTION(minimum, 2, 0, experimental_constrained_minimum, FMINIMU
8989
DAG_FUNCTION(nearbyint, 1, 1, experimental_constrained_nearbyint, FNEARBYINT)
9090
DAG_FUNCTION(pow, 2, 1, experimental_constrained_pow, FPOW)
9191
DAG_FUNCTION(powi, 2, 1, experimental_constrained_powi, FPOWI)
92+
DAG_FUNCTION(ldexp, 2, 1, experimental_constrained_ldexp, FLDEXP)
9293
DAG_FUNCTION(rint, 1, 1, experimental_constrained_rint, FRINT)
9394
DAG_FUNCTION(round, 1, 0, experimental_constrained_round, FROUND)
9495
DAG_FUNCTION(roundeven, 1, 0, experimental_constrained_roundeven, FROUNDEVEN)

‎llvm/include/llvm/IR/Intrinsics.td

Lines changed: 9 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -1037,6 +1037,10 @@ let IntrProperties = [IntrNoMem, IntrSpeculatable, IntrWillReturn] in {
10371037
def int_llround : DefaultAttrsIntrinsic<[llvm_anyint_ty], [llvm_anyfloat_ty]>;
10381038
def int_lrint : DefaultAttrsIntrinsic<[llvm_anyint_ty], [llvm_anyfloat_ty]>;
10391039
def int_llrint : DefaultAttrsIntrinsic<[llvm_anyint_ty], [llvm_anyfloat_ty]>;
1040+
1041+
// TODO: int operand should be constrained to same number of elements as the result.
1042+
def int_ldexp : DefaultAttrsIntrinsic<[llvm_anyfloat_ty], [LLVMMatchType<0>,
1043+
llvm_anyint_ty]>;
10401044
}
10411045

10421046
def int_minnum : DefaultAttrsIntrinsic<[llvm_anyfloat_ty],
@@ -1168,6 +1172,11 @@ let IntrProperties = [IntrInaccessibleMemOnly, IntrWillReturn] in {
11681172
llvm_i32_ty,
11691173
llvm_metadata_ty,
11701174
llvm_metadata_ty ]>;
1175+
def int_experimental_constrained_ldexp : DefaultAttrsIntrinsic<[ llvm_anyfloat_ty ],
1176+
[ LLVMMatchType<0>,
1177+
llvm_anyint_ty,
1178+
llvm_metadata_ty,
1179+
llvm_metadata_ty ]>;
11711180
def int_experimental_constrained_sin : DefaultAttrsIntrinsic<[ llvm_anyfloat_ty ],
11721181
[ LLVMMatchType<0>,
11731182
llvm_metadata_ty,

‎llvm/include/llvm/IR/RuntimeLibcalls.def

Lines changed: 5 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -279,6 +279,11 @@ HANDLE_LIBCALL(LLRINT_F64, "llrint")
279279
HANDLE_LIBCALL(LLRINT_F80, "llrintl")
280280
HANDLE_LIBCALL(LLRINT_F128, "llrintl")
281281
HANDLE_LIBCALL(LLRINT_PPCF128, "llrintl")
282+
HANDLE_LIBCALL(LDEXP_F32, "ldexpf")
283+
HANDLE_LIBCALL(LDEXP_F64, "ldexp")
284+
HANDLE_LIBCALL(LDEXP_F80, "ldexpl")
285+
HANDLE_LIBCALL(LDEXP_F128, "ldexpl")
286+
HANDLE_LIBCALL(LDEXP_PPCF128, "ldexpl")
282287

283288
// Floating point environment
284289
HANDLE_LIBCALL(FEGETENV, "fegetenv")

‎llvm/include/llvm/Support/TargetOpcodes.def

Lines changed: 4 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -613,6 +613,9 @@ HANDLE_TARGET_OPCODE(G_FLOG2)
613613
/// Floating point base-10 logarithm of a value.
614614
HANDLE_TARGET_OPCODE(G_FLOG10)
615615

616+
/// Floating point x * 2^n
617+
HANDLE_TARGET_OPCODE(G_FLDEXP)
618+
616619
/// Generic FP negation.
617620
HANDLE_TARGET_OPCODE(G_FNEG)
618621

@@ -762,6 +765,7 @@ HANDLE_TARGET_OPCODE(G_STRICT_FDIV)
762765
HANDLE_TARGET_OPCODE(G_STRICT_FREM)
763766
HANDLE_TARGET_OPCODE(G_STRICT_FMA)
764767
HANDLE_TARGET_OPCODE(G_STRICT_FSQRT)
768+
HANDLE_TARGET_OPCODE(G_STRICT_FLDEXP)
765769

766770
/// read_register intrinsic
767771
HANDLE_TARGET_OPCODE(G_READ_REGISTER)

‎llvm/include/llvm/Target/GenericOpcodes.td

Lines changed: 8 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -923,6 +923,13 @@ def G_FLOG10 : GenericInstruction {
923923
let hasSideEffects = false;
924924
}
925925

926+
// Floating point x * 2^n
927+
def G_FLDEXP : GenericInstruction {
928+
let OutOperandList = (outs type0:$dst);
929+
let InOperandList = (ins type0:$src0, type1:$src1);
930+
let hasSideEffects = false;
931+
}
932+
926933
// Floating point ceiling of a value.
927934
def G_FCEIL : GenericInstruction {
928935
let OutOperandList = (outs type0:$dst);
@@ -1384,6 +1391,7 @@ def G_STRICT_FDIV : ConstrainedInstruction<G_FDIV>;
13841391
def G_STRICT_FREM : ConstrainedInstruction<G_FREM>;
13851392
def G_STRICT_FMA : ConstrainedInstruction<G_FMA>;
13861393
def G_STRICT_FSQRT : ConstrainedInstruction<G_FSQRT>;
1394+
def G_STRICT_FLDEXP : ConstrainedInstruction<G_FLDEXP>;
13871395

13881396
//------------------------------------------------------------------------------
13891397
// Memory intrinsics

‎llvm/include/llvm/Target/GlobalISel/SelectionDAGCompat.td

Lines changed: 2 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -103,6 +103,7 @@ def : GINodeEquiv<G_FREM, frem>;
103103
def : GINodeEquiv<G_FPOW, fpow>;
104104
def : GINodeEquiv<G_FEXP2, fexp2>;
105105
def : GINodeEquiv<G_FLOG2, flog2>;
106+
def : GINodeEquiv<G_FLDEXP, fldexp>;
106107
def : GINodeEquiv<G_FCANONICALIZE, fcanonicalize>;
107108
def : GINodeEquiv<G_IS_FPCLASS, is_fpclass>;
108109
def : GINodeEquiv<G_INTRINSIC, intrinsic_wo_chain>;
@@ -158,6 +159,7 @@ def : GINodeEquiv<G_STRICT_FDIV, strict_fdiv>;
158159
def : GINodeEquiv<G_STRICT_FREM, strict_frem>;
159160
def : GINodeEquiv<G_STRICT_FMA, strict_fma>;
160161
def : GINodeEquiv<G_STRICT_FSQRT, strict_fsqrt>;
162+
def : GINodeEquiv<G_STRICT_FLDEXP, strict_fldexp>;
161163

162164
// Broadly speaking G_LOAD is equivalent to ISD::LOAD but there are some
163165
// complications that tablegen must take care of. For example, Predicates such

‎llvm/include/llvm/Target/TargetSelectionDAG.td

Lines changed: 9 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -173,6 +173,9 @@ def SDTFPToIntOp : SDTypeProfile<1, 1, [ // fp_to_[su]int
173173
def SDTFPToIntSatOp : SDTypeProfile<1, 2, [ // fp_to_[su]int_sat
174174
SDTCisInt<0>, SDTCisFP<1>, SDTCisSameNumEltsAs<0, 1>, SDTCisVT<2, OtherVT>
175175
]>;
176+
def SDTFPExpOp : SDTypeProfile<1, 2, [ // ldexp
177+
SDTCisSameAs<0, 1>, SDTCisFP<0>, SDTCisInt<2>
178+
]>;
176179
def SDTExtInreg : SDTypeProfile<1, 2, [ // sext_inreg
177180
SDTCisSameAs<0, 1>, SDTCisInt<0>, SDTCisVT<2, OtherVT>,
178181
SDTCisVTSmallerThanOp<2, 1>
@@ -499,6 +502,7 @@ def fcos : SDNode<"ISD::FCOS" , SDTFPUnaryOp>;
499502
def fexp2 : SDNode<"ISD::FEXP2" , SDTFPUnaryOp>;
500503
def fpow : SDNode<"ISD::FPOW" , SDTFPBinOp>;
501504
def flog2 : SDNode<"ISD::FLOG2" , SDTFPUnaryOp>;
505+
def fldexp : SDNode<"ISD::FLDEXP" , SDTFPExpOp>;
502506
def frint : SDNode<"ISD::FRINT" , SDTFPUnaryOp>;
503507
def ftrunc : SDNode<"ISD::FTRUNC" , SDTFPUnaryOp>;
504508
def fceil : SDNode<"ISD::FCEIL" , SDTFPUnaryOp>;
@@ -549,6 +553,8 @@ def strict_fexp2 : SDNode<"ISD::STRICT_FEXP2",
549553
SDTFPUnaryOp, [SDNPHasChain]>;
550554
def strict_fpow : SDNode<"ISD::STRICT_FPOW",
551555
SDTFPBinOp, [SDNPHasChain]>;
556+
def strict_fldexp : SDNode<"ISD::STRICT_FLDEXP",
557+
SDTFPExpOp, [SDNPHasChain]>;
552558
def strict_flog2 : SDNode<"ISD::STRICT_FLOG2",
553559
SDTFPUnaryOp, [SDNPHasChain]>;
554560
def strict_frint : SDNode<"ISD::STRICT_FRINT",
@@ -1449,6 +1455,9 @@ def any_fexp2 : PatFrags<(ops node:$src),
14491455
def any_fpow : PatFrags<(ops node:$lhs, node:$rhs),
14501456
[(strict_fpow node:$lhs, node:$rhs),
14511457
(fpow node:$lhs, node:$rhs)]>;
1458+
def any_fldexp : PatFrags<(ops node:$lhs, node:$rhs),
1459+
[(strict_fldexp node:$lhs, node:$rhs),
1460+
(fldexp node:$lhs, node:$rhs)]>;
14521461
def any_flog2 : PatFrags<(ops node:$src),
14531462
[(strict_flog2 node:$src),
14541463
(flog2 node:$src)]>;

‎llvm/lib/CodeGen/GlobalISel/IRTranslator.cpp

Lines changed: 4 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -1759,6 +1759,8 @@ unsigned IRTranslator::getSimpleIntrinsicOpcode(Intrinsic::ID ID) {
17591759
return TargetOpcode::G_FLOG2;
17601760
case Intrinsic::log10:
17611761
return TargetOpcode::G_FLOG10;
1762+
case Intrinsic::ldexp:
1763+
return TargetOpcode::G_FLDEXP;
17621764
case Intrinsic::nearbyint:
17631765
return TargetOpcode::G_FNEARBYINT;
17641766
case Intrinsic::pow:
@@ -1851,6 +1853,8 @@ static unsigned getConstrainedOpcode(Intrinsic::ID ID) {
18511853
return TargetOpcode::G_STRICT_FMA;
18521854
case Intrinsic::experimental_constrained_sqrt:
18531855
return TargetOpcode::G_STRICT_FSQRT;
1856+
case Intrinsic::experimental_constrained_ldexp:
1857+
return TargetOpcode::G_STRICT_FLDEXP;
18541858
default:
18551859
return 0;
18561860
}

0 commit comments

Comments
 (0)