Skip to content

small fixes for QR ops lowering to custom_call #1290

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

Open
wants to merge 8 commits into
base: main
Choose a base branch
from
Open

small fixes for QR ops lowering to custom_call #1290

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
54cfe61
small fixes for QR ops lowering to `custom_call`
mofeing Aug 9, 2025
b852c83
format
mofeing Aug 9, 2025
5e853a5
Merge branch 'main' into ss/qr-fixes
mofeing Aug 9, 2025
b908cbb
Merge branch 'main' into ss/qr-fixes
wsmoses Aug 14, 2025
beb5d34
Merge branch 'main' into ss/qr-fixes
mofeing Aug 16, 2025
0bb6db1
fix tests
mofeing Aug 17, 2025
2da103c
fix tests again
mofeing Aug 17, 2025
fe265ea
fix gemqrt tall test
mofeing Aug 17, 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
19 changes: 10 additions & 9 deletions src/enzyme_ad/jax/Passes/LowerEnzymeXLALapack.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -137,7 +137,7 @@ struct GeqrfOpLowering : public OpRewritePattern<enzymexla::GeqrfOp> {
// `101` for row-major, `102` for col-major
auto layout = rewriter.create<LLVM::ConstantOp>(
op.getLoc(), type_llvm_lapack_int,
rewriter.getIntegerAttr(type_lapack_int, 101));
rewriter.getIntegerAttr(type_lapack_int, 102));
auto m = rewriter.create<LLVM::ConstantOp>(
op.getLoc(), type_llvm_lapack_int,
rewriter.getIntegerAttr(type_lapack_int, inputShape[0]));
Expand Down Expand Up @@ -183,7 +183,8 @@ struct GeqrfOpLowering : public OpRewritePattern<enzymexla::GeqrfOp> {

SmallVector<Attribute> aliases;
for (int i = 0; i < 3; ++i) {
aliases.push_back(stablehlo::OutputOperandAliasAttr::get(ctx, {}, i, {}));
aliases.push_back(
stablehlo::OutputOperandAliasAttr::get(ctx, {i}, i, {}));
}

auto jit_call_op = rewriter.create<enzymexla::JITCallOp>(
Expand Down Expand Up @@ -411,7 +412,7 @@ struct GeqrtOpLowering : public OpRewritePattern<enzymexla::GeqrtOp> {
// `101` for row-major, `102` for col-major
auto layout = rewriter.create<LLVM::ConstantOp>(
op.getLoc(), type_llvm_lapack_int,
rewriter.getIntegerAttr(type_lapack_int, 101));
rewriter.getIntegerAttr(type_lapack_int, 102));
auto m = rewriter.create<LLVM::ConstantOp>(
op.getLoc(), type_llvm_lapack_int,
rewriter.getIntegerAttr(type_lapack_int, inputShape[0]));
Expand Down Expand Up @@ -472,8 +473,8 @@ struct GeqrtOpLowering : public OpRewritePattern<enzymexla::GeqrtOp> {
op.getLoc(), type_T, cast<ElementsAttr>(makeAttr(type_T, 0)));

SmallVector<bool> isColMajorArr = {true, true, true};
SmallVector<int64_t> operandRanks = {2, 1, 0};
SmallVector<int64_t> outputRanks = {2, 1, 0};
SmallVector<int64_t> operandRanks = {2, 2, 0};
SmallVector<int64_t> outputRanks = {2, 2, 0};
auto operandLayouts =
getSHLOLayout(rewriter, operandRanks, isColMajorArr, 2);
auto resultLayouts = getSHLOLayout(rewriter, outputRanks, isColMajorArr, 2);
Expand Down Expand Up @@ -622,7 +623,7 @@ struct OrgqrOpLowering : public OpRewritePattern<enzymexla::OrgqrOp> {
// `101` for row-major, `102` for col-major
auto layout = rewriter.create<LLVM::ConstantOp>(
op.getLoc(), type_llvm_lapack_int,
rewriter.getIntegerAttr(type_lapack_int, 101));
rewriter.getIntegerAttr(type_lapack_int, 102));
auto mC = inputShape[0];
auto m = rewriter.create<LLVM::ConstantOp>(
op.getLoc(), type_llvm_lapack_int,
Expand Down Expand Up @@ -663,7 +664,7 @@ struct OrgqrOpLowering : public OpRewritePattern<enzymexla::OrgqrOp> {
auto resultLayouts = getSHLOLayout(rewriter, outputRanks, isColMajorArr, 2);

SmallVector<Attribute> aliases;
aliases.push_back(stablehlo::OutputOperandAliasAttr::get(ctx, {0}, 0, {}));
aliases.push_back(stablehlo::OutputOperandAliasAttr::get(ctx, {}, 0, {}));

auto jit_call_op = rewriter.create<enzymexla::JITCallOp>(
op.getLoc(), TypeRange{inputType},
Expand Down Expand Up @@ -933,7 +934,7 @@ struct OrmqrOpLowering : public OpRewritePattern<enzymexla::OrmqrOp> {
// `101` for row-major, `102` for col-major
auto layout = rewriter.create<LLVM::ConstantOp>(
op.getLoc(), type_llvm_lapack_int,
rewriter.getIntegerAttr(type_lapack_int, 101));
rewriter.getIntegerAttr(type_lapack_int, 102));

auto side = rewriter.create<LLVM::ConstantOp>(
op.getLoc(), type_llvm_char,
Expand Down Expand Up @@ -1200,7 +1201,7 @@ struct GemqrtOpLowering : public OpRewritePattern<enzymexla::GemqrtOp> {
// `101` for row-major, `102` for col-major
auto layout = rewriter.create<LLVM::ConstantOp>(
op.getLoc(), type_llvm_lapack_int,
rewriter.getIntegerAttr(type_lapack_int, 101));
rewriter.getIntegerAttr(type_lapack_int, 102));

auto side = rewriter.create<LLVM::ConstantOp>(
op.getLoc(), type_llvm_char,
Expand Down
8 changes: 4 additions & 4 deletions test/lit_tests/linalg/gemqrt_square.mlir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,7 +8,7 @@ module {
}

// CPU: llvm.func @enzymexla_wrapper_lapacke_sgemqrt_[[WRAPPER_ID:[0-9]+]](%arg0: !llvm.ptr, %arg1: !llvm.ptr, %arg2: !llvm.ptr) {
// CPU-NEXT: %0 = llvm.mlir.constant(101 : i64) : i64
// CPU-NEXT: %0 = llvm.mlir.constant(102 : i64) : i64
// CPU-NEXT: %1 = llvm.mlir.constant(76 : i8) : i8
// CPU-NEXT: %2 = llvm.mlir.constant(78 : i8) : i8
// CPU-NEXT: %3 = llvm.mlir.constant(64 : i64) : i64
Expand All @@ -30,7 +30,7 @@ module {
}

// CPU: llvm.func @enzymexla_wrapper_lapacke_sgemqrt_[[WRAPPER_ID:[0-9]+]](%arg0: !llvm.ptr, %arg1: !llvm.ptr, %arg2: !llvm.ptr) {
// CPU-NEXT: %0 = llvm.mlir.constant(101 : i64) : i64
// CPU-NEXT: %0 = llvm.mlir.constant(102 : i64) : i64
// CPU-NEXT: %1 = llvm.mlir.constant(76 : i8) : i8
// CPU-NEXT: %2 = llvm.mlir.constant(84 : i8) : i8
// CPU-NEXT: %3 = llvm.mlir.constant(64 : i64) : i64
Expand All @@ -52,7 +52,7 @@ module {
}

// CPU: llvm.func @enzymexla_wrapper_lapacke_sgemqrt_[[WRAPPER_ID:[0-9]+]](%arg0: !llvm.ptr, %arg1: !llvm.ptr, %arg2: !llvm.ptr) {
// CPU-NEXT: %0 = llvm.mlir.constant(101 : i64) : i64
// CPU-NEXT: %0 = llvm.mlir.constant(102 : i64) : i64
// CPU-NEXT: %1 = llvm.mlir.constant(82 : i8) : i8
// CPU-NEXT: %2 = llvm.mlir.constant(78 : i8) : i8
// CPU-NEXT: %3 = llvm.mlir.constant(48 : i64) : i64
Expand All @@ -74,7 +74,7 @@ module {
}

// CPU: llvm.func @enzymexla_wrapper_lapacke_sgemqrt_[[WRAPPER_ID:[0-9]+]](%arg0: !llvm.ptr, %arg1: !llvm.ptr, %arg2: !llvm.ptr) {
// CPU-NEXT: %0 = llvm.mlir.constant(101 : i64) : i64
// CPU-NEXT: %0 = llvm.mlir.constant(102 : i64) : i64
// CPU-NEXT: %1 = llvm.mlir.constant(82 : i8) : i8
// CPU-NEXT: %2 = llvm.mlir.constant(84 : i8) : i8
// CPU-NEXT: %3 = llvm.mlir.constant(48 : i64) : i64
Expand Down
8 changes: 4 additions & 4 deletions test/lit_tests/linalg/gemqrt_tall.mlir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,7 +8,7 @@ module {
}

// CPU: llvm.func @enzymexla_wrapper_lapacke_sgemqrt_[[WRAPPER_ID:[0-9]+]](%arg0: !llvm.ptr, %arg1: !llvm.ptr, %arg2: !llvm.ptr) {
// CPU-NEXT: %0 = llvm.mlir.constant(101 : i64) : i64
// CPU-NEXT: %0 = llvm.mlir.constant(102 : i64) : i64
// CPU-NEXT: %1 = llvm.mlir.constant(76 : i8) : i8
// CPU-NEXT: %2 = llvm.mlir.constant(78 : i8) : i8
// CPU-NEXT: %3 = llvm.mlir.constant(64 : i64) : i64
Expand All @@ -31,7 +31,7 @@ module {
}

// CPU: llvm.func @enzymexla_wrapper_lapacke_sgemqrt_[[WRAPPER_ID:[0-9]+]](%arg0: !llvm.ptr, %arg1: !llvm.ptr, %arg2: !llvm.ptr) {
// CPU-NEXT: %0 = llvm.mlir.constant(101 : i64) : i64
// CPU-NEXT: %0 = llvm.mlir.constant(102 : i64) : i64
// CPU-NEXT: %1 = llvm.mlir.constant(76 : i8) : i8
// CPU-NEXT: %2 = llvm.mlir.constant(84 : i8) : i8
// CPU-NEXT: %3 = llvm.mlir.constant(64 : i64) : i64
Expand All @@ -54,7 +54,7 @@ module {
}

// CPU: llvm.func @enzymexla_wrapper_lapacke_sgemqrt_[[WRAPPER_ID:[0-9]+]](%arg0: !llvm.ptr, %arg1: !llvm.ptr, %arg2: !llvm.ptr) {
// CPU-NEXT: %0 = llvm.mlir.constant(101 : i64) : i64
// CPU-NEXT: %0 = llvm.mlir.constant(102 : i64) : i64
// CPU-NEXT: %1 = llvm.mlir.constant(82 : i8) : i8
// CPU-NEXT: %2 = llvm.mlir.constant(78 : i8) : i8
// CPU-NEXT: %3 = llvm.mlir.constant(48 : i64) : i64
Expand All @@ -77,7 +77,7 @@ module {
}

// CPU: llvm.func @enzymexla_wrapper_lapacke_sgemqrt_[[WRAPPER_ID:[0-9]+]](%arg0: !llvm.ptr, %arg1: !llvm.ptr, %arg2: !llvm.ptr) {
// CPU-NEXT: %0 = llvm.mlir.constant(101 : i64) : i64
// CPU-NEXT: %0 = llvm.mlir.constant(102 : i64) : i64
// CPU-NEXT: %1 = llvm.mlir.constant(82 : i8) : i8
// CPU-NEXT: %2 = llvm.mlir.constant(84 : i8) : i8
// CPU-NEXT: %3 = llvm.mlir.constant(48 : i64) : i64
Expand Down
4 changes: 2 additions & 2 deletions test/lit_tests/linalg/geqrf_square.mlir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,7 +10,7 @@ module {
}

// CPU: llvm.func @enzymexla_wrapper_lapacke_sgeqrf_[[WRAPPER_ID:[0-9]+]](%arg0: !llvm.ptr, %arg1: !llvm.ptr, %arg2: !llvm.ptr) {
// CPU-NEXT: %0 = llvm.mlir.constant(101 : i64) : i64
// CPU-NEXT: %0 = llvm.mlir.constant(102 : i64) : i64
// CPU-NEXT: %1 = llvm.mlir.constant(64 : i64) : i64
// CPU-NEXT: %2 = llvm.call @enzymexla_lapacke_sgeqrf_(%0, %1, %1, %arg0, %1, %arg1) : (i64, i64, i64, !llvm.ptr, i64, !llvm.ptr) -> i64
// CPU-NEXT: llvm.store %2, %arg2 : i64, !llvm.ptr
Expand All @@ -20,7 +20,7 @@ module {
// CPU-NEXT: func.func @main(%arg0: tensor<64x64xf32>) -> (tensor<64x64xf32>, tensor<64xf32>, tensor<i64>) {
// CPU-NEXT: %c = stablehlo.constant dense<-1> : tensor<i64>
// CPU-NEXT: %cst = stablehlo.constant dense<0.000000e+00> : tensor<64xf32>
// CPU-NEXT: %0:3 = enzymexla.jit_call @enzymexla_wrapper_lapacke_sgeqrf_[[WRAPPER_ID]] (%arg0, %cst, %c) {operand_layouts = [dense<[0, 1]> : tensor<2xindex>, dense<0> : tensor<1xindex>, dense<> : tensor<0xindex>], output_operand_aliases = [#stablehlo.output_operand_alias<output_tuple_indices = [], operand_index = 0, operand_tuple_indices = []>, #stablehlo.output_operand_alias<output_tuple_indices = [], operand_index = 1, operand_tuple_indices = []>, #stablehlo.output_operand_alias<output_tuple_indices = [], operand_index = 2, operand_tuple_indices = []>], result_layouts = [dense<[0, 1]> : tensor<2xindex>, dense<0> : tensor<1xindex>, dense<> : tensor<0xindex>], xla_side_effect_free} : (tensor<64x64xf32>, tensor<64xf32>, tensor<i64>) -> (tensor<64x64xf32>, tensor<64xf32>, tensor<i64>)
// CPU-NEXT: %0:3 = enzymexla.jit_call @enzymexla_wrapper_lapacke_sgeqrf_[[WRAPPER_ID]] (%arg0, %cst, %c) {operand_layouts = [dense<[0, 1]> : tensor<2xindex>, dense<0> : tensor<1xindex>, dense<> : tensor<0xindex>], output_operand_aliases = [#stablehlo.output_operand_alias<output_tuple_indices = [0], operand_index = 0, operand_tuple_indices = []>, #stablehlo.output_operand_alias<output_tuple_indices = [1], operand_index = 1, operand_tuple_indices = []>, #stablehlo.output_operand_alias<output_tuple_indices = [2], operand_index = 2, operand_tuple_indices = []>], result_layouts = [dense<[0, 1]> : tensor<2xindex>, dense<0> : tensor<1xindex>, dense<> : tensor<0xindex>], xla_side_effect_free} : (tensor<64x64xf32>, tensor<64xf32>, tensor<i64>) -> (tensor<64x64xf32>, tensor<64xf32>, tensor<i64>)
// CPU-NEXT: return %0#0, %0#1, %0#2 : tensor<64x64xf32>, tensor<64xf32>, tensor<i64>
// CPU-NEXT: }

Expand Down
4 changes: 2 additions & 2 deletions test/lit_tests/linalg/geqrf_tall_thin.mlir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,7 +10,7 @@ module {
}

// CPU: llvm.func @enzymexla_wrapper_lapacke_sgeqrf_[[WRAPPER_ID:[0-9]+]](%arg0: !llvm.ptr, %arg1: !llvm.ptr, %arg2: !llvm.ptr) {
// CPU-NEXT: %0 = llvm.mlir.constant(101 : i64) : i64
// CPU-NEXT: %0 = llvm.mlir.constant(102 : i64) : i64
// CPU-NEXT: %1 = llvm.mlir.constant(64 : i64) : i64
// CPU-NEXT: %2 = llvm.mlir.constant(32 : i64) : i64
// CPU-NEXT: %3 = llvm.call @enzymexla_lapacke_sgeqrf_(%0, %1, %2, %arg0, %1, %arg1) : (i64, i64, i64, !llvm.ptr, i64, !llvm.ptr) -> i64
Expand All @@ -21,7 +21,7 @@ module {
// CPU-NEXT: func.func @main(%arg0: tensor<64x32xf32>) -> (tensor<64x32xf32>, tensor<32xf32>, tensor<i64>) {
// CPU-NEXT: %c = stablehlo.constant dense<-1> : tensor<i64>
// CPU-NEXT: %cst = stablehlo.constant dense<0.000000e+00> : tensor<32xf32>
// CPU-NEXT: %0:3 = enzymexla.jit_call @enzymexla_wrapper_lapacke_sgeqrf_[[WRAPPER_ID]] (%arg0, %cst, %c) {operand_layouts = [dense<[0, 1]> : tensor<2xindex>, dense<0> : tensor<1xindex>, dense<> : tensor<0xindex>], output_operand_aliases = [#stablehlo.output_operand_alias<output_tuple_indices = [], operand_index = 0, operand_tuple_indices = []>, #stablehlo.output_operand_alias<output_tuple_indices = [], operand_index = 1, operand_tuple_indices = []>, #stablehlo.output_operand_alias<output_tuple_indices = [], operand_index = 2, operand_tuple_indices = []>], result_layouts = [dense<[0, 1]> : tensor<2xindex>, dense<0> : tensor<1xindex>, dense<> : tensor<0xindex>], xla_side_effect_free} : (tensor<64x32xf32>, tensor<32xf32>, tensor<i64>) -> (tensor<64x32xf32>, tensor<32xf32>, tensor<i64>)
// CPU-NEXT: %0:3 = enzymexla.jit_call @enzymexla_wrapper_lapacke_sgeqrf_[[WRAPPER_ID]] (%arg0, %cst, %c) {operand_layouts = [dense<[0, 1]> : tensor<2xindex>, dense<0> : tensor<1xindex>, dense<> : tensor<0xindex>], output_operand_aliases = [#stablehlo.output_operand_alias<output_tuple_indices = [0], operand_index = 0, operand_tuple_indices = []>, #stablehlo.output_operand_alias<output_tuple_indices = [1], operand_index = 1, operand_tuple_indices = []>, #stablehlo.output_operand_alias<output_tuple_indices = [2], operand_index = 2, operand_tuple_indices = []>], result_layouts = [dense<[0, 1]> : tensor<2xindex>, dense<0> : tensor<1xindex>, dense<> : tensor<0xindex>], xla_side_effect_free} : (tensor<64x32xf32>, tensor<32xf32>, tensor<i64>) -> (tensor<64x32xf32>, tensor<32xf32>, tensor<i64>)
// CPU-NEXT: return %0#0, %0#1, %0#2 : tensor<64x32xf32>, tensor<32xf32>, tensor<i64>
// CPU-NEXT: }

Expand Down
4 changes: 2 additions & 2 deletions test/lit_tests/linalg/geqrf_wide_thin.mlir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,7 +10,7 @@ module {
}

// CPU: llvm.func @enzymexla_wrapper_lapacke_sgeqrf_[[WRAPPER_ID:[0-9]+]](%arg0: !llvm.ptr, %arg1: !llvm.ptr, %arg2: !llvm.ptr) {
// CPU-NEXT: %0 = llvm.mlir.constant(101 : i64) : i64
// CPU-NEXT: %0 = llvm.mlir.constant(102 : i64) : i64
// CPU-NEXT: %1 = llvm.mlir.constant(32 : i64) : i64
// CPU-NEXT: %2 = llvm.mlir.constant(64 : i64) : i64
// CPU-NEXT: %3 = llvm.call @enzymexla_lapacke_sgeqrf_(%0, %1, %2, %arg0, %1, %arg1) : (i64, i64, i64, !llvm.ptr, i64, !llvm.ptr) -> i64
Expand All @@ -21,7 +21,7 @@ module {
// CPU-NEXT: func.func @main(%arg0: tensor<32x64xf32>) -> (tensor<32x64xf32>, tensor<32xf32>, tensor<i64>) {
// CPU-NEXT: %c = stablehlo.constant dense<-1> : tensor<i64>
// CPU-NEXT: %cst = stablehlo.constant dense<0.000000e+00> : tensor<32xf32>
// CPU-NEXT: %0:3 = enzymexla.jit_call @enzymexla_wrapper_lapacke_sgeqrf_[[WRAPPER_ID]] (%arg0, %cst, %c) {operand_layouts = [dense<[0, 1]> : tensor<2xindex>, dense<0> : tensor<1xindex>, dense<> : tensor<0xindex>], output_operand_aliases = [#stablehlo.output_operand_alias<output_tuple_indices = [], operand_index = 0, operand_tuple_indices = []>, #stablehlo.output_operand_alias<output_tuple_indices = [], operand_index = 1, operand_tuple_indices = []>, #stablehlo.output_operand_alias<output_tuple_indices = [], operand_index = 2, operand_tuple_indices = []>], result_layouts = [dense<[0, 1]> : tensor<2xindex>, dense<0> : tensor<1xindex>, dense<> : tensor<0xindex>], xla_side_effect_free} : (tensor<32x64xf32>, tensor<32xf32>, tensor<i64>) -> (tensor<32x64xf32>, tensor<32xf32>, tensor<i64>)
// CPU-NEXT: %0:3 = enzymexla.jit_call @enzymexla_wrapper_lapacke_sgeqrf_[[WRAPPER_ID]] (%arg0, %cst, %c) {operand_layouts = [dense<[0, 1]> : tensor<2xindex>, dense<0> : tensor<1xindex>, dense<> : tensor<0xindex>], output_operand_aliases = [#stablehlo.output_operand_alias<output_tuple_indices = [0], operand_index = 0, operand_tuple_indices = []>, #stablehlo.output_operand_alias<output_tuple_indices = [1], operand_index = 1, operand_tuple_indices = []>, #stablehlo.output_operand_alias<output_tuple_indices = [2], operand_index = 2, operand_tuple_indices = []>], result_layouts = [dense<[0, 1]> : tensor<2xindex>, dense<0> : tensor<1xindex>, dense<> : tensor<0xindex>], xla_side_effect_free} : (tensor<32x64xf32>, tensor<32xf32>, tensor<i64>) -> (tensor<32x64xf32>, tensor<32xf32>, tensor<i64>)
// CPU-NEXT: return %0#0, %0#1, %0#2 : tensor<32x64xf32>, tensor<32xf32>, tensor<i64>
// CPU-NEXT: }

Expand Down
4 changes: 2 additions & 2 deletions test/lit_tests/linalg/geqrt_square.mlir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,7 +8,7 @@ module {
}

// CPU: llvm.func @enzymexla_wrapper_lapacke_sgeqrt_[[WRAPPER_ID:[0-9]+]](%arg0: !llvm.ptr, %arg1: !llvm.ptr, %arg2: !llvm.ptr) {
// CPU-NEXT: %0 = llvm.mlir.constant(101 : i64) : i64
// CPU-NEXT: %0 = llvm.mlir.constant(102 : i64) : i64
// CPU-NEXT: %1 = llvm.mlir.constant(64 : i64) : i64
// CPU-NEXT: %2 = llvm.call @enzymexla_lapacke_sgeqrt_(%0, %1, %1, %1, %arg0, %1, %arg1, %1) : (i64, i64, i64, i64, !llvm.ptr, i64, !llvm.ptr, i64) -> i64
// CPU-NEXT: llvm.store %2, %arg2 : i64, !llvm.ptr
Expand All @@ -18,7 +18,7 @@ module {
// CPU-NEXT: func.func @main(%arg0: tensor<64x64xf32>) -> (tensor<64x64xf32>, tensor<64x64xf32>, tensor<i64>) {
// CPU-NEXT: %c = stablehlo.constant dense<-1> : tensor<i64>
// CPU-NEXT: %cst = stablehlo.constant dense<0.000000e+00> : tensor<64x64xf32>
// CPU-NEXT: %0:3 = enzymexla.jit_call @enzymexla_wrapper_lapacke_sgeqrt_[[WRAPPER_ID]] (%arg0, %cst, %c) {operand_layouts = [dense<[0, 1]> : tensor<2xindex>, dense<0> : tensor<1xindex>, dense<> : tensor<0xindex>], output_operand_aliases = [#stablehlo.output_operand_alias<output_tuple_indices = [0], operand_index = 0, operand_tuple_indices = []>, #stablehlo.output_operand_alias<output_tuple_indices = [1], operand_index = 1, operand_tuple_indices = []>, #stablehlo.output_operand_alias<output_tuple_indices = [2], operand_index = 2, operand_tuple_indices = []>], result_layouts = [dense<[0, 1]> : tensor<2xindex>, dense<0> : tensor<1xindex>, dense<> : tensor<0xindex>], xla_side_effect_free} : (tensor<64x64xf32>, tensor<64x64xf32>, tensor<i64>) -> (tensor<64x64xf32>, tensor<64x64xf32>, tensor<i64>)
// CPU-NEXT: %0:3 = enzymexla.jit_call @enzymexla_wrapper_lapacke_sgeqrt_[[WRAPPER_ID]] (%arg0, %cst, %c) {operand_layouts = [dense<[0, 1]> : tensor<2xindex>, dense<[0, 1]> : tensor<2xindex>, dense<> : tensor<0xindex>], output_operand_aliases = [#stablehlo.output_operand_alias<output_tuple_indices = [0], operand_index = 0, operand_tuple_indices = []>, #stablehlo.output_operand_alias<output_tuple_indices = [1], operand_index = 1, operand_tuple_indices = []>, #stablehlo.output_operand_alias<output_tuple_indices = [2], operand_index = 2, operand_tuple_indices = []>], result_layouts = [dense<[0, 1]> : tensor<2xindex>, dense<[0, 1]> : tensor<2xindex>, dense<> : tensor<0xindex>], xla_side_effect_free} : (tensor<64x64xf32>, tensor<64x64xf32>, tensor<i64>) -> (tensor<64x64xf32>, tensor<64x64xf32>, tensor<i64>)
// CPU-NEXT: return %0#0, %0#1, %0#2 : tensor<64x64xf32>, tensor<64x64xf32>, tensor<i64>
// CPU-NEXT: }

Expand Down
4 changes: 2 additions & 2 deletions test/lit_tests/linalg/geqrt_tall_thin.mlir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,7 +8,7 @@ module {
}

// CPU: llvm.func @enzymexla_wrapper_lapacke_sgeqrt_[[WRAPPER_ID:[0-9]+]](%arg0: !llvm.ptr, %arg1: !llvm.ptr, %arg2: !llvm.ptr) {
// CPU-NEXT: %0 = llvm.mlir.constant(101 : i64) : i64
// CPU-NEXT: %0 = llvm.mlir.constant(102 : i64) : i64
// CPU-NEXT: %1 = llvm.mlir.constant(64 : i64) : i64
// CPU-NEXT: %2 = llvm.mlir.constant(32 : i64) : i64
// CPU-NEXT: %3 = llvm.call @enzymexla_lapacke_sgeqrt_(%0, %1, %2, %2, %arg0, %1, %arg1, %2) : (i64, i64, i64, i64, !llvm.ptr, i64, !llvm.ptr, i64) -> i64
Expand All @@ -19,7 +19,7 @@ module {
// CPU-NEXT: func.func @main(%arg0: tensor<64x32xf32>) -> (tensor<64x32xf32>, tensor<32x32xf32>, tensor<i64>) {
// CPU-NEXT: %c = stablehlo.constant dense<-1> : tensor<i64>
// CPU-NEXT: %cst = stablehlo.constant dense<0.000000e+00> : tensor<32x32xf32>
// CPU-NEXT: %0:3 = enzymexla.jit_call @enzymexla_wrapper_lapacke_sgeqrt_[[WRAPPER_ID]] (%arg0, %cst, %c) {operand_layouts = [dense<[0, 1]> : tensor<2xindex>, dense<0> : tensor<1xindex>, dense<> : tensor<0xindex>], output_operand_aliases = [#stablehlo.output_operand_alias<output_tuple_indices = [0], operand_index = 0, operand_tuple_indices = []>, #stablehlo.output_operand_alias<output_tuple_indices = [1], operand_index = 1, operand_tuple_indices = []>, #stablehlo.output_operand_alias<output_tuple_indices = [2], operand_index = 2, operand_tuple_indices = []>], result_layouts = [dense<[0, 1]> : tensor<2xindex>, dense<0> : tensor<1xindex>, dense<> : tensor<0xindex>], xla_side_effect_free} : (tensor<64x32xf32>, tensor<32x32xf32>, tensor<i64>) -> (tensor<64x32xf32>, tensor<32x32xf32>, tensor<i64>)
// CPU-NEXT: %0:3 = enzymexla.jit_call @enzymexla_wrapper_lapacke_sgeqrt_[[WRAPPER_ID]] (%arg0, %cst, %c) {operand_layouts = [dense<[0, 1]> : tensor<2xindex>, dense<[0, 1]> : tensor<2xindex>, dense<> : tensor<0xindex>], output_operand_aliases = [#stablehlo.output_operand_alias<output_tuple_indices = [0], operand_index = 0, operand_tuple_indices = []>, #stablehlo.output_operand_alias<output_tuple_indices = [1], operand_index = 1, operand_tuple_indices = []>, #stablehlo.output_operand_alias<output_tuple_indices = [2], operand_index = 2, operand_tuple_indices = []>], result_layouts = [dense<[0, 1]> : tensor<2xindex>, dense<[0, 1]> : tensor<2xindex>, dense<> : tensor<0xindex>], xla_side_effect_free} : (tensor<64x32xf32>, tensor<32x32xf32>, tensor<i64>) -> (tensor<64x32xf32>, tensor<32x32xf32>, tensor<i64>)
// CPU-NEXT: return %0#0, %0#1, %0#2 : tensor<64x32xf32>, tensor<32x32xf32>, tensor<i64>
// CPU-NEXT: }

Expand Down
Loading
Loading