Skip to content

[flang] Optimize assignments of multidimensional arrays #146408

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 3 commits into from
Jul 4, 2025
Merged

[flang] Optimize assignments of multidimensional arrays #146408

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
c9a350f
[flang] Optimize assignments of multidimensional arrays
luporl Jun 9, 2025
cc08a23
Use hlfir.designate and hlfir.assign
luporl Jul 2, 2025
2eb7b7e
Remoeve isAllocatableArray check
luporl Jul 3, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
43 changes: 36 additions & 7 deletions flang/lib/Optimizer/HLFIR/Transforms/OptimizedBufferization.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -786,13 +786,42 @@ llvm::LogicalResult BroadcastAssignBufferization::matchAndRewrite(
mlir::Value shape = hlfir::genShape(loc, builder, lhs);
llvm::SmallVector<mlir::Value> extents =
hlfir::getIndexExtents(loc, builder, shape);
hlfir::LoopNest loopNest =
hlfir::genLoopNest(loc, builder, extents, /*isUnordered=*/true,
flangomp::shouldUseWorkshareLowering(assign));
builder.setInsertionPointToStart(loopNest.body);
auto arrayElement =
hlfir::getElementAt(loc, builder, lhs, loopNest.oneBasedIndices);
builder.create<hlfir::AssignOp>(loc, rhs, arrayElement);

if (lhs.isSimplyContiguous() && extents.size() > 1) {
// Flatten the array to use a single assign loop, that can be better
// optimized.
mlir::Value n = extents[0];
for (size_t i = 1; i < extents.size(); ++i)
n = builder.create<mlir::arith::MulIOp>(loc, n, extents[i]);
extents = {n};
shape = builder.genShape(loc, extents);
mlir::Type flatArrayType =
fir::ReferenceType::get(fir::SequenceType::get(eleTy, 1));
mlir::Value flatArray = lhs.getBase();
if (mlir::isa<fir::BoxType>(lhs.getType()))
flatArray = builder.create<fir::BoxAddrOp>(loc, flatArray);
flatArray = builder.createConvert(loc, flatArrayType, flatArray);

hlfir::LoopNest loopNest =
hlfir::genLoopNest(loc, builder, extents, /*isUnordered=*/true,
flangomp::shouldUseWorkshareLowering(assign));
builder.setInsertionPointToStart(loopNest.body);

mlir::Value coor = builder.create<fir::ArrayCoorOp>(
loc, fir::ReferenceType::get(eleTy), flatArray, shape,
/*slice=*/mlir::Value{}, loopNest.oneBasedIndices,
/*typeparams=*/mlir::ValueRange{});
builder.create<fir::StoreOp>(loc, rhs, coor);
} else {
hlfir::LoopNest loopNest =
hlfir::genLoopNest(loc, builder, extents, /*isUnordered=*/true,
flangomp::shouldUseWorkshareLowering(assign));
builder.setInsertionPointToStart(loopNest.body);
auto arrayElement =
hlfir::getElementAt(loc, builder, lhs, loopNest.oneBasedIndices);
builder.create<hlfir::AssignOp>(loc, rhs, arrayElement);
}

rewriter.eraseOp(assign);
return mlir::success();
}
Expand Down
53 changes: 41 additions & 12 deletions flang/test/HLFIR/opt-scalar-assign.fir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -12,18 +12,20 @@ func.func @_QPtest1() {
return
}
// CHECK-LABEL: func.func @_QPtest1() {
// CHECK: %[[VAL_0:.*]] = arith.constant 1 : index
// CHECK: %[[VAL_1:.*]] = arith.constant 0.000000e+00 : f32
// CHECK: %[[VAL_2:.*]] = arith.constant 11 : index
// CHECK: %[[VAL_3:.*]] = arith.constant 13 : index
// CHECK: %[[VAL_4:.*]] = fir.alloca !fir.array<11x13xf32> {bindc_name = "x", uniq_name = "_QFtest1Ex"}
// CHECK: %[[VAL_5:.*]] = fir.shape %[[VAL_2]], %[[VAL_3]] : (index, index) -> !fir.shape<2>
// CHECK: %[[VAL_6:.*]]:2 = hlfir.declare %[[VAL_4]](%[[VAL_5]]) {uniq_name = "_QFtest1Ex"} : (!fir.ref<!fir.array<11x13xf32>>, !fir.shape<2>) -> (!fir.ref<!fir.array<11x13xf32>>, !fir.ref<!fir.array<11x13xf32>>)
// CHECK: fir.do_loop %[[VAL_7:.*]] = %[[VAL_0]] to %[[VAL_3]] step %[[VAL_0]] unordered {
// CHECK: fir.do_loop %[[VAL_8:.*]] = %[[VAL_0]] to %[[VAL_2]] step %[[VAL_0]] unordered {
// CHECK: %[[VAL_9:.*]] = hlfir.designate %[[VAL_6]]#0 (%[[VAL_8]], %[[VAL_7]]) : (!fir.ref<!fir.array<11x13xf32>>, index, index) -> !fir.ref<f32>
// CHECK: hlfir.assign %[[VAL_1]] to %[[VAL_9]] : f32, !fir.ref<f32>
// CHECK: }
// CHECK: %[[VAL_0:.*]] = arith.constant 143 : index
// CHECK: %[[VAL_1:.*]] = arith.constant 1 : index
// CHECK: %[[VAL_2:.*]] = arith.constant 0.000000e+00 : f32
// CHECK: %[[VAL_3:.*]] = arith.constant 11 : index
// CHECK: %[[VAL_4:.*]] = arith.constant 13 : index
// CHECK: %[[VAL_5:.*]] = fir.alloca !fir.array<11x13xf32> {bindc_name = "x", uniq_name = "_QFtest1Ex"}
// CHECK: %[[VAL_6:.*]] = fir.shape %[[VAL_3]], %[[VAL_4]] : (index, index) -> !fir.shape<2>
// CHECK: %[[VAL_7:.*]]:2 = hlfir.declare %[[VAL_5]](%[[VAL_6]]) {uniq_name = "_QFtest1Ex"} : (!fir.ref<!fir.array<11x13xf32>>, !fir.shape<2>) -> (!fir.ref<!fir.array<11x13xf32>>, !fir.ref<!fir.array<11x13xf32>>)

// CHECK: %[[VAL_8:.*]] = fir.shape %[[VAL_0]] : (index) -> !fir.shape<1>
// CHECK: %[[VAL_9:.*]] = fir.convert %[[VAL_7]]#0 : (!fir.ref<!fir.array<11x13xf32>>) -> !fir.ref<!fir.array<?xf32>>
// CHECK: fir.do_loop %[[VAL_10:.*]] = %[[VAL_1]] to %[[VAL_0]] step %[[VAL_1]] unordered {
// CHECK: %[[VAL_11:.*]] = fir.array_coor %[[VAL_9]](%[[VAL_8]]) %[[VAL_10]] : (!fir.ref<!fir.array<?xf32>>, !fir.shape<1>, index) -> !fir.ref<f32>
// CHECK: fir.store %[[VAL_2]] to %[[VAL_11]] : !fir.ref<f32>
// CHECK: }
// CHECK: return
// CHECK: }
Expand Down Expand Up @@ -129,3 +131,30 @@ func.func @_QPtest5(%arg0: !fir.ref<!fir.array<77xcomplex<f32>>> {fir.bindc_name
// CHECK: }
// CHECK: return
// CHECK: }

func.func @_QPtest6(%arg0: !fir.ref<!fir.box<!fir.heap<!fir.array<?x?xi32>>>> {fir.bindc_name = "x"}) {
%c0_i32 = arith.constant 0 : i32
%0:2 = hlfir.declare %arg0 {fortran_attrs = #fir.var_attrs<allocatable>, uniq_name = "_QFtest6Ex"} : (!fir.ref<!fir.box<!fir.heap<!fir.array<?x?xi32>>>>) -> (!fir.ref<!fir.box<!fir.heap<!fir.array<?x?xi32>>>>, !fir.ref<!fir.box<!fir.heap<!fir.array<?x?xi32>>>>)
hlfir.assign %c0_i32 to %0#0 realloc : i32, !fir.ref<!fir.box<!fir.heap<!fir.array<?x?xi32>>>>
return
}

// CHECK-LABEL: func.func @_QPtest6(
// CHECK-SAME: %[[VAL_0:.*]]: !fir.ref<!fir.box<!fir.heap<!fir.array<?x?xi32>>>> {fir.bindc_name = "x"}) {
// CHECK: %[[VAL_1:.*]] = arith.constant 1 : index
// CHECK: %[[VAL_2:.*]] = arith.constant 0 : index
// CHECK: %[[VAL_3:.*]] = arith.constant 0 : i32
// CHECK: %[[VAL_4:.*]]:2 = hlfir.declare %[[VAL_0]] {fortran_attrs = #fir.var_attrs<allocatable>, uniq_name = "_QFtest6Ex"} : (!fir.ref<!fir.box<!fir.heap<!fir.array<?x?xi32>>>>) -> (!fir.ref<!fir.box<!fir.heap<!fir.array<?x?xi32>>>>, !fir.ref<!fir.box<!fir.heap<!fir.array<?x?xi32>>>>)
// CHECK: %[[VAL_5:.*]] = fir.load %[[VAL_4]]#0 : !fir.ref<!fir.box<!fir.heap<!fir.array<?x?xi32>>>>
// CHECK: %[[VAL_6:.*]]:3 = fir.box_dims %[[VAL_5]], %[[VAL_2]] : (!fir.box<!fir.heap<!fir.array<?x?xi32>>>, index) -> (index, index, index)
// CHECK: %[[VAL_7:.*]]:3 = fir.box_dims %[[VAL_5]], %[[VAL_1]] : (!fir.box<!fir.heap<!fir.array<?x?xi32>>>, index) -> (index, index, index)
// CHECK: %[[VAL_8:.*]] = arith.muli %[[VAL_6]]#1, %[[VAL_7]]#1 : index
// CHECK: %[[VAL_9:.*]] = fir.shape %[[VAL_8]] : (index) -> !fir.shape<1>
// CHECK: %[[VAL_10:.*]] = fir.box_addr %[[VAL_5]] : (!fir.box<!fir.heap<!fir.array<?x?xi32>>>) -> !fir.heap<!fir.array<?x?xi32>>
// CHECK: %[[VAL_11:.*]] = fir.convert %[[VAL_10]] : (!fir.heap<!fir.array<?x?xi32>>) -> !fir.ref<!fir.array<?xi32>>
// CHECK: fir.do_loop %[[VAL_12:.*]] = %[[VAL_1]] to %[[VAL_8]] step %[[VAL_1]] unordered {
// CHECK: %[[VAL_13:.*]] = fir.array_coor %[[VAL_11]](%[[VAL_9]]) %[[VAL_12]] : (!fir.ref<!fir.array<?xi32>>, !fir.shape<1>, index) -> !fir.ref<i32>
// CHECK: fir.store %[[VAL_3]] to %[[VAL_13]] : !fir.ref<i32>
// CHECK: }
// CHECK: return
// CHECK: }
Loading