Skip to content

Commit

Permalink
[flang] Add fastmath attributes to complex arithmetic (#70690)
Browse files Browse the repository at this point in the history 
Propagate fast math flags through complex number lowering (when lowering
fir.*c directly to llvm floating point operations).

The lowering path through the MLIR complex dialect is unchanged.

This leads to a small improvement in spec2017 fotonik3d_r.
  • Loading branch information
@tblah
tblah committed Oct 31, 2023
1 parent 1adb898 commit 6be0e97
Show file tree
Hide file tree
Showing 7 changed files with 77 additions and 59 deletions.
18 changes: 12 additions & 6 deletions flang/include/flang/Optimizer/Dialect/FIROps.td
View file
Original file line number Diff line number Diff line change
Expand Up @@ -2538,12 +2538,18 @@ def fir_NegcOp : ComplexUnaryArithmeticOp<"negc">;

class ComplexArithmeticOp<string mnemonic, list<Trait> traits = []> :
fir_ArithmeticOp<mnemonic, traits>,
Arguments<(ins fir_ComplexType:$lhs, fir_ComplexType:$rhs)>;

def fir_AddcOp : ComplexArithmeticOp<"addc", [Commutative]>;
def fir_SubcOp : ComplexArithmeticOp<"subc">;
def fir_MulcOp : ComplexArithmeticOp<"mulc", [Commutative]>;
def fir_DivcOp : ComplexArithmeticOp<"divc">;
Arguments<(ins fir_ComplexType:$lhs, fir_ComplexType:$rhs,
DefaultValuedAttr<Arith_FastMathAttr,
"::mlir::arith::FastMathFlags::none">:$fastmath)>;

def fir_AddcOp : ComplexArithmeticOp<"addc",
[Commutative, DeclareOpInterfaceMethods<ArithFastMathInterface>]>;
def fir_SubcOp : ComplexArithmeticOp<"subc",
[DeclareOpInterfaceMethods<ArithFastMathInterface>]>;
def fir_MulcOp : ComplexArithmeticOp<"mulc",
[Commutative, DeclareOpInterfaceMethods<ArithFastMathInterface>]>;
def fir_DivcOp : ComplexArithmeticOp<"divc",
[DeclareOpInterfaceMethods<ArithFastMathInterface>]>;
// Pow is a builtin call and not a primitive

def fir_CmpcOp : fir_Op<"cmpc",
Expand Down
50 changes: 31 additions & 19 deletions flang/lib/Optimizer/CodeGen/CodeGen.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -33,6 +33,7 @@
#include "mlir/Conversion/OpenMPToLLVM/ConvertOpenMPToLLVM.h"
#include "mlir/Conversion/ReconcileUnrealizedCasts/ReconcileUnrealizedCasts.h"
#include "mlir/Conversion/VectorToLLVM/ConvertVectorToLLVM.h"
#include "mlir/Dialect/Arith/IR/Arith.h"
#include "mlir/Dialect/LLVMIR/LLVMDialect.h"
#include "mlir/Dialect/LLVMIR/Transforms/AddComdats.h"
#include "mlir/Dialect/OpenACC/OpenACC.h"
Expand All @@ -44,6 +45,7 @@
#include "mlir/Target/LLVMIR/ModuleTranslation.h"
#include "llvm/ADT/ArrayRef.h"
#include "llvm/ADT/TypeSwitch.h"
#include <mlir/Dialect/LLVMIR/LLVMAttrs.h>

namespace fir {
#define GEN_PASS_DEF_FIRTOLLVMLOWERING
Expand Down Expand Up @@ -3496,12 +3498,20 @@ struct AbsentOpConversion : public FIROpConversion<fir::AbsentOp> {
// Primitive operations on Complex types
//

template <typename OPTY>
static inline mlir::LLVM::FastmathFlagsAttr getLLVMFMFAttr(OPTY op) {
return mlir::LLVM::FastmathFlagsAttr::get(
op.getContext(),
mlir::arith::convertArithFastMathFlagsToLLVM(op.getFastmath()));
}

/// Generate inline code for complex addition/subtraction
template <typename LLVMOP, typename OPTY>
static mlir::LLVM::InsertValueOp
complexSum(OPTY sumop, mlir::ValueRange opnds,
mlir::ConversionPatternRewriter &rewriter,
const fir::LLVMTypeConverter &lowering) {
mlir::LLVM::FastmathFlagsAttr fmf = getLLVMFMFAttr(sumop);
mlir::Value a = opnds[0];
mlir::Value b = opnds[1];
auto loc = sumop.getLoc();
Expand All @@ -3511,8 +3521,8 @@ complexSum(OPTY sumop, mlir::ValueRange opnds,
auto y0 = rewriter.create<mlir::LLVM::ExtractValueOp>(loc, a, 1);
auto x1 = rewriter.create<mlir::LLVM::ExtractValueOp>(loc, b, 0);
auto y1 = rewriter.create<mlir::LLVM::ExtractValueOp>(loc, b, 1);
auto rx = rewriter.create<LLVMOP>(loc, eleTy, x0, x1);
auto ry = rewriter.create<LLVMOP>(loc, eleTy, y0, y1);
auto rx = rewriter.create<LLVMOP>(loc, eleTy, x0, x1, fmf);
auto ry = rewriter.create<LLVMOP>(loc, eleTy, y0, y1, fmf);
auto r0 = rewriter.create<mlir::LLVM::UndefOp>(loc, ty);
auto r1 = rewriter.create<mlir::LLVM::InsertValueOp>(loc, r0, rx, 0);
return rewriter.create<mlir::LLVM::InsertValueOp>(loc, r1, ry, 1);
Expand Down Expand Up @@ -3560,6 +3570,7 @@ struct MulcOpConversion : public FIROpConversion<fir::MulcOp> {
// TODO: Can we use a call to __muldc3 ?
// given: (x + iy) * (x' + iy')
// result: (xx'-yy')+i(xy'+yx')
mlir::LLVM::FastmathFlagsAttr fmf = getLLVMFMFAttr(mulc);
mlir::Value a = adaptor.getOperands()[0];
mlir::Value b = adaptor.getOperands()[1];
auto loc = mulc.getLoc();
Expand All @@ -3569,12 +3580,12 @@ struct MulcOpConversion : public FIROpConversion<fir::MulcOp> {
auto y0 = rewriter.create<mlir::LLVM::ExtractValueOp>(loc, a, 1);
auto x1 = rewriter.create<mlir::LLVM::ExtractValueOp>(loc, b, 0);
auto y1 = rewriter.create<mlir::LLVM::ExtractValueOp>(loc, b, 1);
auto xx = rewriter.create<mlir::LLVM::FMulOp>(loc, eleTy, x0, x1);
auto yx = rewriter.create<mlir::LLVM::FMulOp>(loc, eleTy, y0, x1);
auto xy = rewriter.create<mlir::LLVM::FMulOp>(loc, eleTy, x0, y1);
auto ri = rewriter.create<mlir::LLVM::FAddOp>(loc, eleTy, xy, yx);
auto yy = rewriter.create<mlir::LLVM::FMulOp>(loc, eleTy, y0, y1);
auto rr = rewriter.create<mlir::LLVM::FSubOp>(loc, eleTy, xx, yy);
auto xx = rewriter.create<mlir::LLVM::FMulOp>(loc, eleTy, x0, x1, fmf);
auto yx = rewriter.create<mlir::LLVM::FMulOp>(loc, eleTy, y0, x1, fmf);
auto xy = rewriter.create<mlir::LLVM::FMulOp>(loc, eleTy, x0, y1, fmf);
auto ri = rewriter.create<mlir::LLVM::FAddOp>(loc, eleTy, xy, yx, fmf);
auto yy = rewriter.create<mlir::LLVM::FMulOp>(loc, eleTy, y0, y1, fmf);
auto rr = rewriter.create<mlir::LLVM::FSubOp>(loc, eleTy, xx, yy, fmf);
auto ra = rewriter.create<mlir::LLVM::UndefOp>(loc, ty);
auto r1 = rewriter.create<mlir::LLVM::InsertValueOp>(loc, ra, rr, 0);
auto r0 = rewriter.create<mlir::LLVM::InsertValueOp>(loc, r1, ri, 1);
Expand All @@ -3594,6 +3605,7 @@ struct DivcOpConversion : public FIROpConversion<fir::DivcOp> {
// Just generate inline code for now.
// given: (x + iy) / (x' + iy')
// result: ((xx'+yy')/d) + i((yx'-xy')/d) where d = x'x' + y'y'
mlir::LLVM::FastmathFlagsAttr fmf = getLLVMFMFAttr(divc);
mlir::Value a = adaptor.getOperands()[0];
mlir::Value b = adaptor.getOperands()[1];
auto loc = divc.getLoc();
Expand All @@ -3603,17 +3615,17 @@ struct DivcOpConversion : public FIROpConversion<fir::DivcOp> {
auto y0 = rewriter.create<mlir::LLVM::ExtractValueOp>(loc, a, 1);
auto x1 = rewriter.create<mlir::LLVM::ExtractValueOp>(loc, b, 0);
auto y1 = rewriter.create<mlir::LLVM::ExtractValueOp>(loc, b, 1);
auto xx = rewriter.create<mlir::LLVM::FMulOp>(loc, eleTy, x0, x1);
auto x1x1 = rewriter.create<mlir::LLVM::FMulOp>(loc, eleTy, x1, x1);
auto yx = rewriter.create<mlir::LLVM::FMulOp>(loc, eleTy, y0, x1);
auto xy = rewriter.create<mlir::LLVM::FMulOp>(loc, eleTy, x0, y1);
auto yy = rewriter.create<mlir::LLVM::FMulOp>(loc, eleTy, y0, y1);
auto y1y1 = rewriter.create<mlir::LLVM::FMulOp>(loc, eleTy, y1, y1);
auto d = rewriter.create<mlir::LLVM::FAddOp>(loc, eleTy, x1x1, y1y1);
auto rrn = rewriter.create<mlir::LLVM::FAddOp>(loc, eleTy, xx, yy);
auto rin = rewriter.create<mlir::LLVM::FSubOp>(loc, eleTy, yx, xy);
auto rr = rewriter.create<mlir::LLVM::FDivOp>(loc, eleTy, rrn, d);
auto ri = rewriter.create<mlir::LLVM::FDivOp>(loc, eleTy, rin, d);
auto xx = rewriter.create<mlir::LLVM::FMulOp>(loc, eleTy, x0, x1, fmf);
auto x1x1 = rewriter.create<mlir::LLVM::FMulOp>(loc, eleTy, x1, x1, fmf);
auto yx = rewriter.create<mlir::LLVM::FMulOp>(loc, eleTy, y0, x1, fmf);
auto xy = rewriter.create<mlir::LLVM::FMulOp>(loc, eleTy, x0, y1, fmf);
auto yy = rewriter.create<mlir::LLVM::FMulOp>(loc, eleTy, y0, y1, fmf);
auto y1y1 = rewriter.create<mlir::LLVM::FMulOp>(loc, eleTy, y1, y1, fmf);
auto d = rewriter.create<mlir::LLVM::FAddOp>(loc, eleTy, x1x1, y1y1, fmf);
auto rrn = rewriter.create<mlir::LLVM::FAddOp>(loc, eleTy, xx, yy, fmf);
auto rin = rewriter.create<mlir::LLVM::FSubOp>(loc, eleTy, yx, xy, fmf);
auto rr = rewriter.create<mlir::LLVM::FDivOp>(loc, eleTy, rrn, d, fmf);
auto ri = rewriter.create<mlir::LLVM::FDivOp>(loc, eleTy, rin, d, fmf);
auto ra = rewriter.create<mlir::LLVM::UndefOp>(loc, ty);
auto r1 = rewriter.create<mlir::LLVM::InsertValueOp>(loc, ra, rr, 0);
auto r0 = rewriter.create<mlir::LLVM::InsertValueOp>(loc, r1, ri, 1);
Expand Down
50 changes: 25 additions & 25 deletions flang/test/Fir/convert-to-llvm.fir
View file
Original file line number Diff line number Diff line change
Expand Up @@ -507,7 +507,7 @@ func.func @test_call_return_val() -> i32 {
// result: (x + x') + i(y + y')

func.func @fir_complex_add(%a: !fir.complex<16>, %b: !fir.complex<16>) -> !fir.complex<16> {
%c = fir.addc %a, %b : !fir.complex<16>
%c = fir.addc %a, %b {fastmath = #arith.fastmath<fast>} : !fir.complex<16>
return %c : !fir.complex<16>
}

Expand All @@ -518,8 +518,8 @@ func.func @fir_complex_add(%a: !fir.complex<16>, %b: !fir.complex<16>) -> !fir.c
// CHECK: %[[Y0:.*]] = llvm.extractvalue %[[ARG0]][1] : !llvm.struct<(f128, f128)>
// CHECK: %[[X1:.*]] = llvm.extractvalue %[[ARG1]][0] : !llvm.struct<(f128, f128)>
// CHECK: %[[Y1:.*]] = llvm.extractvalue %[[ARG1]][1] : !llvm.struct<(f128, f128)>
// CHECK: %[[ADD_X0_X1:.*]] = llvm.fadd %[[X0]], %[[X1]] : f128
// CHECK: %[[ADD_Y0_Y1:.*]] = llvm.fadd %[[Y0]], %[[Y1]] : f128
// CHECK: %[[ADD_X0_X1:.*]] = llvm.fadd %[[X0]], %[[X1]] {fastmathFlags = #llvm.fastmath<fast>} : f128
// CHECK: %[[ADD_Y0_Y1:.*]] = llvm.fadd %[[Y0]], %[[Y1]] {fastmathFlags = #llvm.fastmath<fast>} : f128
// CHECK: %{{.*}} = llvm.mlir.undef : !llvm.struct<(f128, f128)>
// CHECK: %{{.*}} = llvm.insertvalue %[[ADD_X0_X1]], %{{.*}}[0] : !llvm.struct<(f128, f128)>
// CHECK: %{{.*}} = llvm.insertvalue %[[ADD_Y0_Y1]], %{{.*}}[1] : !llvm.struct<(f128, f128)>
Expand All @@ -532,7 +532,7 @@ func.func @fir_complex_add(%a: !fir.complex<16>, %b: !fir.complex<16>) -> !fir.c
// result: (x - x') + i(y - y')

func.func @fir_complex_sub(%a: !fir.complex<16>, %b: !fir.complex<16>) -> !fir.complex<16> {
%c = fir.subc %a, %b : !fir.complex<16>
%c = fir.subc %a, %b {fastmath = #arith.fastmath<fast>} : !fir.complex<16>
return %c : !fir.complex<16>
}

Expand All @@ -543,8 +543,8 @@ func.func @fir_complex_sub(%a: !fir.complex<16>, %b: !fir.complex<16>) -> !fir.c
// CHECK: %[[Y0:.*]] = llvm.extractvalue %[[ARG0]][1] : !llvm.struct<(f128, f128)>
// CHECK: %[[X1:.*]] = llvm.extractvalue %[[ARG1]][0] : !llvm.struct<(f128, f128)>
// CHECK: %[[Y1:.*]] = llvm.extractvalue %[[ARG1]][1] : !llvm.struct<(f128, f128)>
// CHECK: %[[SUB_X0_X1:.*]] = llvm.fsub %[[X0]], %[[X1]] : f128
// CHECK: %[[SUB_Y0_Y1:.*]] = llvm.fsub %[[Y0]], %[[Y1]] : f128
// CHECK: %[[SUB_X0_X1:.*]] = llvm.fsub %[[X0]], %[[X1]] {fastmathFlags = #llvm.fastmath<fast>} : f128
// CHECK: %[[SUB_Y0_Y1:.*]] = llvm.fsub %[[Y0]], %[[Y1]] {fastmathFlags = #llvm.fastmath<fast>} : f128
// CHECK: %{{.*}} = llvm.mlir.undef : !llvm.struct<(f128, f128)>
// CHECK: %{{.*}} = llvm.insertvalue %[[SUB_X0_X1]], %{{.*}}[0] : !llvm.struct<(f128, f128)>
// CHECK: %{{.*}} = llvm.insertvalue %[[SUB_Y0_Y1]], %{{.*}}[1] : !llvm.struct<(f128, f128)>
Expand All @@ -557,7 +557,7 @@ func.func @fir_complex_sub(%a: !fir.complex<16>, %b: !fir.complex<16>) -> !fir.c
// result: (xx'-yy')+i(xy'+yx')

func.func @fir_complex_mul(%a: !fir.complex<16>, %b: !fir.complex<16>) -> !fir.complex<16> {
%c = fir.mulc %a, %b : !fir.complex<16>
%c = fir.mulc %a, %b {fastmath = #arith.fastmath<fast>} : !fir.complex<16>
return %c : !fir.complex<16>
}

Expand All @@ -568,12 +568,12 @@ func.func @fir_complex_mul(%a: !fir.complex<16>, %b: !fir.complex<16>) -> !fir.c
// CHECK: %[[Y0:.*]] = llvm.extractvalue %[[ARG0]][1] : !llvm.struct<(f128, f128)>
// CHECK: %[[X1:.*]] = llvm.extractvalue %[[ARG1]][0] : !llvm.struct<(f128, f128)>
// CHECK: %[[Y1:.*]] = llvm.extractvalue %[[ARG1]][1] : !llvm.struct<(f128, f128)>
// CHECK: %[[MUL_X0_X1:.*]] = llvm.fmul %[[X0]], %[[X1]] : f128
// CHECK: %[[MUL_Y0_X1:.*]] = llvm.fmul %[[Y0]], %[[X1]] : f128
// CHECK: %[[MUL_X0_Y1:.*]] = llvm.fmul %[[X0]], %[[Y1]] : f128
// CHECK: %[[ADD:.*]] = llvm.fadd %[[MUL_X0_Y1]], %[[MUL_Y0_X1]] : f128
// CHECK: %[[MUL_Y0_Y1:.*]] = llvm.fmul %[[Y0]], %[[Y1]] : f128
// CHECK: %[[SUB:.*]] = llvm.fsub %[[MUL_X0_X1]], %[[MUL_Y0_Y1]] : f128
// CHECK: %[[MUL_X0_X1:.*]] = llvm.fmul %[[X0]], %[[X1]] {fastmathFlags = #llvm.fastmath<fast>} : f128
// CHECK: %[[MUL_Y0_X1:.*]] = llvm.fmul %[[Y0]], %[[X1]] {fastmathFlags = #llvm.fastmath<fast>} : f128
// CHECK: %[[MUL_X0_Y1:.*]] = llvm.fmul %[[X0]], %[[Y1]] {fastmathFlags = #llvm.fastmath<fast>} : f128
// CHECK: %[[ADD:.*]] = llvm.fadd %[[MUL_X0_Y1]], %[[MUL_Y0_X1]] {fastmathFlags = #llvm.fastmath<fast>} : f128
// CHECK: %[[MUL_Y0_Y1:.*]] = llvm.fmul %[[Y0]], %[[Y1]] {fastmathFlags = #llvm.fastmath<fast>} : f128
// CHECK: %[[SUB:.*]] = llvm.fsub %[[MUL_X0_X1]], %[[MUL_Y0_Y1]] {fastmathFlags = #llvm.fastmath<fast>} : f128
// CHECK: %{{.*}} = llvm.mlir.undef : !llvm.struct<(f128, f128)>
// CHECK: %{{.*}} = llvm.insertvalue %[[SUB]], %{{.*}}[0] : !llvm.struct<(f128, f128)>
// CHECK: %{{.*}} = llvm.insertvalue %[[ADD]], %{{.*}}[1] : !llvm.struct<(f128, f128)>
Expand All @@ -586,7 +586,7 @@ func.func @fir_complex_mul(%a: !fir.complex<16>, %b: !fir.complex<16>) -> !fir.c
// result: ((xx'+yy')/d) + i((yx'-xy')/d) where d = x'x' + y'y'

func.func @fir_complex_div(%a: !fir.complex<16>, %b: !fir.complex<16>) -> !fir.complex<16> {
%c = fir.divc %a, %b : !fir.complex<16>
%c = fir.divc %a, %b {fastmath = #arith.fastmath<fast>} : !fir.complex<16>
return %c : !fir.complex<16>
}

Expand All @@ -597,17 +597,17 @@ func.func @fir_complex_div(%a: !fir.complex<16>, %b: !fir.complex<16>) -> !fir.c
// CHECK: %[[Y0:.*]] = llvm.extractvalue %[[ARG0]][1] : !llvm.struct<(f128, f128)>
// CHECK: %[[X1:.*]] = llvm.extractvalue %[[ARG1]][0] : !llvm.struct<(f128, f128)>
// CHECK: %[[Y1:.*]] = llvm.extractvalue %[[ARG1]][1] : !llvm.struct<(f128, f128)>
// CHECK: %[[MUL_X0_X1:.*]] = llvm.fmul %[[X0]], %[[X1]] : f128
// CHECK: %[[MUL_X1_X1:.*]] = llvm.fmul %[[X1]], %[[X1]] : f128
// CHECK: %[[MUL_Y0_X1:.*]] = llvm.fmul %[[Y0]], %[[X1]] : f128
// CHECK: %[[MUL_X0_Y1:.*]] = llvm.fmul %[[X0]], %[[Y1]] : f128
// CHECK: %[[MUL_Y0_Y1:.*]] = llvm.fmul %[[Y0]], %[[Y1]] : f128
// CHECK: %[[MUL_Y1_Y1:.*]] = llvm.fmul %[[Y1]], %[[Y1]] : f128
// CHECK: %[[ADD_X1X1_Y1Y1:.*]] = llvm.fadd %[[MUL_X1_X1]], %[[MUL_Y1_Y1]] : f128
// CHECK: %[[ADD_X0X1_Y0Y1:.*]] = llvm.fadd %[[MUL_X0_X1]], %[[MUL_Y0_Y1]] : f128
// CHECK: %[[SUB_Y0X1_X0Y1:.*]] = llvm.fsub %[[MUL_Y0_X1]], %[[MUL_X0_Y1]] : f128
// CHECK: %[[DIV0:.*]] = llvm.fdiv %[[ADD_X0X1_Y0Y1]], %[[ADD_X1X1_Y1Y1]] : f128
// CHECK: %[[DIV1:.*]] = llvm.fdiv %[[SUB_Y0X1_X0Y1]], %[[ADD_X1X1_Y1Y1]] : f128
// CHECK: %[[MUL_X0_X1:.*]] = llvm.fmul %[[X0]], %[[X1]] {fastmathFlags = #llvm.fastmath<fast>} : f128
// CHECK: %[[MUL_X1_X1:.*]] = llvm.fmul %[[X1]], %[[X1]] {fastmathFlags = #llvm.fastmath<fast>} : f128
// CHECK: %[[MUL_Y0_X1:.*]] = llvm.fmul %[[Y0]], %[[X1]] {fastmathFlags = #llvm.fastmath<fast>} : f128
// CHECK: %[[MUL_X0_Y1:.*]] = llvm.fmul %[[X0]], %[[Y1]] {fastmathFlags = #llvm.fastmath<fast>} : f128
// CHECK: %[[MUL_Y0_Y1:.*]] = llvm.fmul %[[Y0]], %[[Y1]] {fastmathFlags = #llvm.fastmath<fast>} : f128
// CHECK: %[[MUL_Y1_Y1:.*]] = llvm.fmul %[[Y1]], %[[Y1]] {fastmathFlags = #llvm.fastmath<fast>} : f128
// CHECK: %[[ADD_X1X1_Y1Y1:.*]] = llvm.fadd %[[MUL_X1_X1]], %[[MUL_Y1_Y1]] {fastmathFlags = #llvm.fastmath<fast>} : f128
// CHECK: %[[ADD_X0X1_Y0Y1:.*]] = llvm.fadd %[[MUL_X0_X1]], %[[MUL_Y0_Y1]] {fastmathFlags = #llvm.fastmath<fast>} : f128
// CHECK: %[[SUB_Y0X1_X0Y1:.*]] = llvm.fsub %[[MUL_Y0_X1]], %[[MUL_X0_Y1]] {fastmathFlags = #llvm.fastmath<fast>} : f128
// CHECK: %[[DIV0:.*]] = llvm.fdiv %[[ADD_X0X1_Y0Y1]], %[[ADD_X1X1_Y1Y1]] {fastmathFlags = #llvm.fastmath<fast>} : f128
// CHECK: %[[DIV1:.*]] = llvm.fdiv %[[SUB_Y0X1_X0Y1]], %[[ADD_X1X1_Y1Y1]] {fastmathFlags = #llvm.fastmath<fast>} : f128
// CHECK: %{{.*}} = llvm.mlir.undef : !llvm.struct<(f128, f128)>
// CHECK: %{{.*}} = llvm.insertvalue %[[DIV0]], %{{.*}}[0] : !llvm.struct<(f128, f128)>
// CHECK: %{{.*}} = llvm.insertvalue %[[DIV1]], %{{.*}}[1] : !llvm.struct<(f128, f128)>
Expand Down
6 changes: 3 additions & 3 deletions flang/test/Lower/HLFIR/binary-ops.f90
View file
Original file line number Diff line number Diff line change
Expand Up @@ -32,7 +32,7 @@ subroutine complex_add(x, y, z)
! CHECK: %[[VAL_5:.*]]:2 = hlfir.declare %{{.*}}z"} : (!fir.ref<!fir.complex<4>>) -> (!fir.ref<!fir.complex<4>>, !fir.ref<!fir.complex<4>>)
! CHECK: %[[VAL_6:.*]] = fir.load %[[VAL_4]]#0 : !fir.ref<!fir.complex<4>>
! CHECK: %[[VAL_7:.*]] = fir.load %[[VAL_5]]#0 : !fir.ref<!fir.complex<4>>
! CHECK: %[[VAL_8:.*]] = fir.addc %[[VAL_6]], %[[VAL_7]] : !fir.complex<4>
! CHECK: %[[VAL_8:.*]] = fir.addc %[[VAL_6]], %[[VAL_7]] {fastmath = #arith.fastmath<contract>} : !fir.complex<4>

subroutine int_sub(x, y, z)
integer :: x, y, z
Expand Down Expand Up @@ -65,7 +65,7 @@ subroutine complex_sub(x, y, z)
! CHECK: %[[VAL_5:.*]]:2 = hlfir.declare %{{.*}}z"} : (!fir.ref<!fir.complex<4>>) -> (!fir.ref<!fir.complex<4>>, !fir.ref<!fir.complex<4>>)
! CHECK: %[[VAL_6:.*]] = fir.load %[[VAL_4]]#0 : !fir.ref<!fir.complex<4>>
! CHECK: %[[VAL_7:.*]] = fir.load %[[VAL_5]]#0 : !fir.ref<!fir.complex<4>>
! CHECK: %[[VAL_8:.*]] = fir.subc %[[VAL_6]], %[[VAL_7]] : !fir.complex<4>
! CHECK: %[[VAL_8:.*]] = fir.subc %[[VAL_6]], %[[VAL_7]] {fastmath = #arith.fastmath<contract>} : !fir.complex<4>

subroutine int_mul(x, y, z)
integer :: x, y, z
Expand Down Expand Up @@ -98,7 +98,7 @@ subroutine complex_mul(x, y, z)
! CHECK: %[[VAL_5:.*]]:2 = hlfir.declare %{{.*}}z"} : (!fir.ref<!fir.complex<4>>) -> (!fir.ref<!fir.complex<4>>, !fir.ref<!fir.complex<4>>)
! CHECK: %[[VAL_6:.*]] = fir.load %[[VAL_4]]#0 : !fir.ref<!fir.complex<4>>
! CHECK: %[[VAL_7:.*]] = fir.load %[[VAL_5]]#0 : !fir.ref<!fir.complex<4>>
! CHECK: %[[VAL_8:.*]] = fir.mulc %[[VAL_6]], %[[VAL_7]] : !fir.complex<4>
! CHECK: %[[VAL_8:.*]] = fir.mulc %[[VAL_6]], %[[VAL_7]] {fastmath = #arith.fastmath<contract>} : !fir.complex<4>

subroutine int_div(x, y, z)
integer :: x, y, z
Expand Down
4 changes: 2 additions & 2 deletions flang/test/Lower/OpenACC/acc-reduction.f90
View file
Original file line number Diff line number Diff line change
Expand Up @@ -163,7 +163,7 @@
! CHECK: ^bb0(%[[ARG0:.*]]: !fir.ref<!fir.complex<4>>, %[[ARG1:.*]]: !fir.ref<!fir.complex<4>>):
! CHECK: %[[LOAD0:.*]] = fir.load %[[ARG0]] : !fir.ref<!fir.complex<4>>
! CHECK: %[[LOAD1:.*]] = fir.load %[[ARG1]] : !fir.ref<!fir.complex<4>>
! CHECK: %[[COMBINED:.*]] = fir.mulc %[[LOAD0]], %[[LOAD1]] : !fir.complex<4>
! CHECK: %[[COMBINED:.*]] = fir.mulc %[[LOAD0]], %[[LOAD1]] {fastmath = #arith.fastmath<contract>} : !fir.complex<4>
! CHECK: fir.store %[[COMBINED]] to %[[ARG0]] : !fir.ref<!fir.complex<4>>
! CHECK: acc.yield %[[ARG0]] : !fir.ref<!fir.complex<4>>
! CHECK: }
Expand All @@ -183,7 +183,7 @@
! CHECK: ^bb0(%[[ARG0:.*]]: !fir.ref<!fir.complex<4>>, %[[ARG1:.*]]: !fir.ref<!fir.complex<4>>):
! CHECK: %[[LOAD0:.*]] = fir.load %[[ARG0]] : !fir.ref<!fir.complex<4>>
! CHECK: %[[LOAD1:.*]] = fir.load %[[ARG1]] : !fir.ref<!fir.complex<4>>
! CHECK: %[[COMBINED:.*]] = fir.addc %[[LOAD0]], %[[LOAD1]] : !fir.complex<4>
! CHECK: %[[COMBINED:.*]] = fir.addc %[[LOAD0]], %[[LOAD1]] {fastmath = #arith.fastmath<contract>} : !fir.complex<4>
! CHECK: fir.store %[[COMBINED]] to %[[ARG0]] : !fir.ref<!fir.complex<4>>
! CHECK: acc.yield %[[ARG0]] : !fir.ref<!fir.complex<4>>
! CHECK: }
Expand Down
2 changes: 1 addition & 1 deletion flang/test/Lower/array-elemental-calls-2.f90
View file
Original file line number Diff line number Diff line change
Expand Up @@ -144,7 +144,7 @@ subroutine check_cmplx_part()
! CHECK: %[[VAL_13:.*]] = fir.load %{{.*}} : !fir.ref<!fir.complex<8>>
! CHECK: fir.do_loop
! CHECK: %[[VAL_23:.*]] = fir.array_fetch %{{.*}}, %{{.*}} : (!fir.array<10x!fir.complex<8>>, index) -> !fir.complex<8>
! CHECK: %[[VAL_24:.*]] = fir.addc %[[VAL_23]], %[[VAL_13]] : !fir.complex<8>
! CHECK: %[[VAL_24:.*]] = fir.addc %[[VAL_23]], %[[VAL_13]] {fastmath = #arith.fastmath<contract>} : !fir.complex<8>
! CHECK: %[[VAL_25:.*]] = fir.extract_value %[[VAL_24]], [1 : index] : (!fir.complex<8>) -> f64
! CHECK: fir.call @_QPelem_func_real(%[[VAL_25]]) {{.*}}: (f64) -> i32
end subroutine
Expand Down
Loading

0 comments on commit 6be0e97

Please sign in to comment.