Skip to content
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

[onnx] Lower onnx.Gemm to torch #2663

Merged
merged 1 commit into from
Dec 16, 2023
Merged

[onnx] Lower onnx.Gemm to torch #2663

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
74 changes: 74 additions & 0 deletions lib/Conversion/TorchOnnxToTorch/DefaultDomainGtoP.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -71,6 +71,80 @@ void mlir::torch::onnx_c::populateDefaultDomainGtoP(
binder.op, resultType, data, constAxis, indices, sparseGrad);
return success();
});
patterns.onOp(
"Gemm", 13, [](OpBinder binder, ConversionPatternRewriter &rewriter) {
Torch::ValueTensorType resultType;
Value a, b, c;
float alpha, beta;
int64_t transA, transB;
if (binder.tensorOperandAtIndex(a, 0) ||
binder.tensorOperandAtIndex(b, 1) ||
binder.tensorOperandAtIndex(c, 2) ||
binder.s64IntegerAttr(transA, "transA", 0) ||
binder.s64IntegerAttr(transB, "transB", 0) ||
binder.f32FloatAttr(alpha, "alpha", 1.0) ||
binder.f32FloatAttr(beta, "beta", 1.0) ||
binder.tensorResultType(resultType))
return failure();

Value zero = rewriter.create<Torch::ConstantIntOp>(
binder.getLoc(), rewriter.getType<Torch::IntType>(),
rewriter.getIntegerAttr(rewriter.getIntegerType(64), 0));
Value one = rewriter.create<Torch::ConstantIntOp>(
binder.getLoc(), rewriter.getType<Torch::IntType>(),
rewriter.getIntegerAttr(rewriter.getIntegerType(64), 1));

auto transpose = [&](Value m) -> Value {
auto tty = m.getType().cast<Torch::ValueTensorType>();
auto shape = tty.getOptionalSizes();
if (shape.has_value()) {
llvm::SmallVector<int64_t> newShape(shape.value());
std::reverse(newShape.begin(), newShape.end());
shape = std::move(newShape);
}
auto oty = Torch::ValueTensorType::get(tty.getContext(), shape,
tty.getOptionalDtype());
return rewriter.create<Torch::AtenTransposeIntOp>(binder.getLoc(),
oty, m, zero, one);
};

if (transA) {
a = transpose(a);
}

if (transB) {
b = transpose(b);
}

Value mm =
rewriter.create<Torch::AtenMmOp>(binder.getLoc(), resultType, a, b);
if (alpha == 1.0 && beta == 1.0) {
rewriter.replaceOpWithNewOp<Torch::AtenAddTensorOp>(
binder.op, resultType, mm, c, one);
return success();
}

if (alpha != 1.0 && beta != 1.0) {
Value constAlpha = rewriter.create<Torch::ConstantFloatOp>(
binder.getLoc(), rewriter.getType<Torch::FloatType>(),
rewriter.getF64FloatAttr(alpha));
mm = rewriter.create<Torch::AtenMulScalarOp>(
binder.getLoc(), resultType, mm, constAlpha);
alpha = 1.0;
}

if (alpha != 1.0) {
std::swap(alpha, beta);
std::swap(mm, c);
}

Value constBeta = rewriter.create<Torch::ConstantFloatOp>(
binder.getLoc(), rewriter.getType<Torch::FloatType>(),
rewriter.getF64FloatAttr(beta));
rewriter.replaceOpWithNewOp<Torch::AtenAddTensorOp>(
binder.op, resultType, mm, c, constBeta);
return success();
});
patterns.onOp("LeakyRelu", 16,
[](OpBinder binder, ConversionPatternRewriter &rewriter) {
Torch::ValueTensorType resultType;
Expand Down
76 changes: 76 additions & 0 deletions test/Conversion/TorchOnnxToTorch/simple_ops_g_to_p.mlir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,6 +13,82 @@ func.func @test_gather_elements(%arg0: !torch.vtensor<[3,4,5],f32>, %arg1: !torc
return %0 : !torch.vtensor<[3,4,5],f32>
}

// -----

// CHECK-LABEL: func.func @test_gemm_default
func.func @test_gemm_default(%arg0: !torch.vtensor<[3,5],f32>, %arg1: !torch.vtensor<[5,4],f32>, %arg2: !torch.vtensor<[1,4],f32>) -> !torch.vtensor<[3,4],f32> attributes {torch.onnx_meta.ir_version = 7 : si64, torch.onnx_meta.opset_version = 13 : si64} {
// CHECK: %[[I1:.+]] = torch.constant.int 1
// CHECK: %[[MM:.+]] = torch.aten.mm %arg0, %arg1 : !torch.vtensor<[3,5],f32>, !torch.vtensor<[5,4],f32> -> !torch.vtensor<[3,4],f32>
// CHECK: torch.aten.add.Tensor %[[MM]], %arg2, %[[I1]] : !torch.vtensor<[3,4],f32>, !torch.vtensor<[1,4],f32>, !torch.int -> !torch.vtensor<[3,4],f32>
%0 = torch.operator "onnx.Gemm"(%arg0, %arg1, %arg2) : (!torch.vtensor<[3,5],f32>, !torch.vtensor<[5,4],f32>, !torch.vtensor<[1,4],f32>) -> !torch.vtensor<[3,4],f32>
return %0 : !torch.vtensor<[3,4],f32>
}

// -----

// CHECK-LABEL: func.func @test_gemm_transposeA
func.func @test_gemm_transposeA(%arg0: !torch.vtensor<[5,3],f32>, %arg1: !torch.vtensor<[5,4],f32>, %arg2: !torch.vtensor<[1,4],f32>) -> !torch.vtensor<[3,4],f32> attributes {torch.onnx_meta.ir_version = 7 : si64, torch.onnx_meta.opset_version = 13 : si64} {
// CHECK: %[[I0:.+]] = torch.constant.int 0
// CHECK: %[[I1:.+]] = torch.constant.int 1
// CHECK: %[[TRANS:.+]] = torch.aten.transpose.int %arg0, %[[I0]], %[[I1]] : !torch.vtensor<[5,3],f32>, !torch.int, !torch.int -> !torch.vtensor<[3,5],f32>
// CHECK: %[[MM:.+]] = torch.aten.mm %[[TRANS]], %arg1 : !torch.vtensor<[3,5],f32>, !torch.vtensor<[5,4],f32> -> !torch.vtensor<[3,4],f32>
// CHECK: torch.aten.add.Tensor %[[MM]], %arg2, %[[I1]] : !torch.vtensor<[3,4],f32>, !torch.vtensor<[1,4],f32>, !torch.int -> !torch.vtensor<[3,4],f32>
%0 = torch.operator "onnx.Gemm"(%arg0, %arg1, %arg2) {torch.onnx.transA = 1 : si64} : (!torch.vtensor<[5,3],f32>, !torch.vtensor<[5,4],f32>, !torch.vtensor<[1,4],f32>) -> !torch.vtensor<[3,4],f32>
return %0 : !torch.vtensor<[3,4],f32>
}

// -----

// CHECK-LABEL: func.func @test_gemm_transposeB
func.func @test_gemm_transposeB(%arg0: !torch.vtensor<[3,5],f32>, %arg1: !torch.vtensor<[4,5],f32>, %arg2: !torch.vtensor<[1,4],f32>) -> !torch.vtensor<[3,4],f32> attributes {torch.onnx_meta.ir_version = 7 : si64, torch.onnx_meta.opset_version = 13 : si64} {
// CHECK: %[[I0:.+]] = torch.constant.int 0
// CHECK: %[[I1:.+]] = torch.constant.int 1
// CHECK: %[[TRANS:.+]] = torch.aten.transpose.int %arg1, %[[I0]], %[[I1]] : !torch.vtensor<[4,5],f32>, !torch.int, !torch.int -> !torch.vtensor<[5,4],f32>
// CHECK: %[[MM:.+]] = torch.aten.mm %arg0, %[[TRANS]] : !torch.vtensor<[3,5],f32>, !torch.vtensor<[5,4],f32> -> !torch.vtensor<[3,4],f32>
// CHECK: torch.aten.add.Tensor %[[MM]], %arg2, %[[I1]] : !torch.vtensor<[3,4],f32>, !torch.vtensor<[1,4],f32>, !torch.int -> !torch.vtensor<[3,4],f32>
%0 = torch.operator "onnx.Gemm"(%arg0, %arg1, %arg2) {torch.onnx.transB = 1 : si64} : (!torch.vtensor<[3,5],f32>, !torch.vtensor<[4,5],f32>, !torch.vtensor<[1,4],f32>) -> !torch.vtensor<[3,4],f32>
return %0 : !torch.vtensor<[3,4],f32>
}

// -----

// CHECK-LABEL: func.func @test_gemm_alpha
func.func @test_gemm_alpha(%arg0: !torch.vtensor<[3,5],f32>, %arg1: !torch.vtensor<[5,4],f32>, %arg2: !torch.vtensor<[1,4],f32>) -> !torch.vtensor<[3,4],f32> attributes {torch.onnx_meta.ir_version = 7 : si64, torch.onnx_meta.opset_version = 13 : si64} {
// CHECK-DAG: %[[MM:.+]] = torch.aten.mm %arg0, %arg1 : !torch.vtensor<[3,5],f32>, !torch.vtensor<[5,4],f32> -> !torch.vtensor<[3,4],f32>
// CHECK-DAG: %[[ALPHA:.+]] = torch.constant.float 5.000000e-01
// CHECK: torch.aten.add.Tensor %arg2, %[[MM]], %[[ALPHA]] : !torch.vtensor<[1,4],f32>, !torch.vtensor<[3,4],f32>, !torch.float -> !torch.vtensor<[3,4],f32>
%0 = torch.operator "onnx.Gemm"(%arg0, %arg1, %arg2) {torch.onnx.alpha = 5.000000e-01 : f32} : (!torch.vtensor<[3,5],f32>, !torch.vtensor<[5,4],f32>, !torch.vtensor<[1,4],f32>) -> !torch.vtensor<[3,4],f32>
return %0 : !torch.vtensor<[3,4],f32>
}

// -----

// CHECK-LABEL: func.func @test_gemm_beta
func.func @test_gemm_beta(%arg0: !torch.vtensor<[3,5],f32>, %arg1: !torch.vtensor<[5,4],f32>, %arg2: !torch.vtensor<[1,4],f32>) -> !torch.vtensor<[3,4],f32> attributes {torch.onnx_meta.ir_version = 7 : si64, torch.onnx_meta.opset_version = 13 : si64} {
// CHECK-DAG: %[[MM:.+]] = torch.aten.mm %arg0, %arg1 : !torch.vtensor<[3,5],f32>, !torch.vtensor<[5,4],f32> -> !torch.vtensor<[3,4],f32>
// CHECK-DAG: %[[BETA:.+]] = torch.constant.float 5.000000e-01
// CHECK: torch.aten.add.Tensor %[[MM]], %arg2, %[[BETA]] : !torch.vtensor<[3,4],f32>, !torch.vtensor<[1,4],f32>, !torch.float -> !torch.vtensor<[3,4],f32>
%0 = torch.operator "onnx.Gemm"(%arg0, %arg1, %arg2) {torch.onnx.beta = 5.000000e-01 : f32} : (!torch.vtensor<[3,5],f32>, !torch.vtensor<[5,4],f32>, !torch.vtensor<[1,4],f32>) -> !torch.vtensor<[3,4],f32>
return %0 : !torch.vtensor<[3,4],f32>
}

// -----

// CHECK-LABEL: func.func @test_gemm_alpha_beta
func.func @test_gemm_alpha_beta(%arg0: !torch.vtensor<[3,5],f32>, %arg1: !torch.vtensor<[5,4],f32>, %arg2: !torch.vtensor<[1,4],f32>) -> !torch.vtensor<[3,4],f32> attributes {torch.onnx_meta.ir_version = 7 : si64, torch.onnx_meta.opset_version = 13 : si64} {
// CHECK-DAG: %[[I0:.+]] = torch.constant.int 0
// CHECK-DAG: %[[I1:.+]] = torch.constant.int 1
// CHECK-DAG: %[[MM:.+]] = torch.aten.mm %arg0, %arg1 : !torch.vtensor<[3,5],f32>, !torch.vtensor<[5,4],f32> -> !torch.vtensor<[3,4],f32>
// CHECK-DAG: %[[ALPHA:.+]] = torch.constant.float 5.000000e-01
// CHECK-DAG: %[[BETA:.+]] = torch.constant.float 2.500000e-01
// CHECK-DAG: %[[MUL:.+]] = torch.aten.mul.Scalar %[[MM]], %[[ALPHA]] : !torch.vtensor<[3,4],f32>, !torch.float -> !torch.vtensor<[3,4],f32>
// CHECK: torch.aten.add.Tensor %[[MUL]], %arg2, %[[BETA]] : !torch.vtensor<[3,4],f32>, !torch.vtensor<[1,4],f32>, !torch.float -> !torch.vtensor<[3,4],f32>
%0 = torch.operator "onnx.Gemm"(%arg0, %arg1, %arg2) {torch.onnx.alpha = 5.000000e-01 : f32, torch.onnx.beta = 2.500000e-01 : f32} : (!torch.vtensor<[3,5],f32>, !torch.vtensor<[5,4],f32>, !torch.vtensor<[1,4],f32>) -> !torch.vtensor<[3,4],f32>
return %0 : !torch.vtensor<[3,4],f32>
}

// -----

// CHECK-LABEL: func.func @test_leaky_relu
func.func @test_leaky_relu(%arg0: !torch.vtensor<[3,4,5],f32>) -> !torch.vtensor<[3,4,5],f32> attributes {torch.onnx_meta.opset_version = 16 : si64} {
// CHECK-DAG: %[[F2:.+]] = torch.constant.float 2
Expand Down