[Torch] support aten.trunc (#3219)

decompose `trunc(x)` to `sign(x) * floor(abs(x))`

include/torch-mlir/Dialect/Torch/IR/GeneratedTorchOps.td

@@ -4223,6 +4223,52 @@ def Torch_AtenRound_Op : Torch_Op<"aten.round_", [
   }];
 }
 
+def Torch_AtenTruncOp : Torch_Op<"aten.trunc", [
+    AllowsTypeRefinement,
+    HasValueSemantics,
+    ReadOnly
+  ]> {
+  let summary = "Generated op for `aten::trunc : (Tensor) -> (Tensor)`";
+  let arguments = (ins
+    AnyTorchTensorType:$self
+  );
+  let results = (outs
+    AnyTorchOptionalTensorType:$result
+  );
+  let hasCustomAssemblyFormat = 1;
+  let extraClassDefinition = [{
+    ParseResult AtenTruncOp::parse(OpAsmParser &parser, OperationState &result) {
+      return parseDefaultTorchOp(parser, result, 1, 1);
+    }
+    void AtenTruncOp::print(OpAsmPrinter &printer) {
+      printDefaultTorchOp(printer, *this, 1, 1);
+    }
+  }];
+  let hasFolder = 1;
+}
+
+def Torch_AtenTrunc_Op : Torch_Op<"aten.trunc_", [
+    IsTrailingUnderscoreInplaceVariant,
+    AllowsTypeRefinement
+  ]> {
+  let summary = "Generated op for `aten::trunc_ : (Tensor) -> (Tensor)`";
+  let arguments = (ins
+    Torch_NonValueTensorType:$self
+  );
+  let results = (outs
+    AnyTorchOptionalNonValueTensorType:$result
+  );
+  let hasCustomAssemblyFormat = 1;
+  let extraClassDefinition = [{
+    ParseResult AtenTrunc_Op::parse(OpAsmParser &parser, OperationState &result) {
+      return parseDefaultTorchOp(parser, result, 1, 1);
+    }
+    void AtenTrunc_Op::print(OpAsmPrinter &printer) {
+      printDefaultTorchOp(printer, *this, 1, 1);
+    }
+  }];
+}
+
 def Torch_AtenSignOp : Torch_Op<"aten.sign", [
     AllowsTypeRefinement,
     HasValueSemantics,

lib/Dialect/Torch/IR/TorchOps.cpp

@@ -1834,6 +1834,19 @@ OpFoldResult AtenRoundOp::fold(FoldAdaptor adaptor) {
   return {};
 }
 
+//===----------------------------------------------------------------------===//
+// AtenTruncOp
+//===----------------------------------------------------------------------===//
+
+OpFoldResult AtenTruncOp::fold(FoldAdaptor adaptor) {
+  auto resultType = getType().dyn_cast<ValueTensorType>();
+  if (resultType && resultType.hasDtype() &&
+      resultType.getDtype().isa<mlir::IntegerType>()) {
+    return getSelf();
+  }
+  return {};
+}
+
 //===----------------------------------------------------------------------===//
 // AtenSignOp
 //===----------------------------------------------------------------------===//

lib/Dialect/Torch/Transforms/AbstractInterpLibrary.cpp

@@ -6502,6 +6502,10 @@ StringRef mlir::torch::Torch::getAbstractInterpLibrary() {
 "    %0 = call @__torch__.torch.jit._shape_functions.unary(%arg0) : (!torch.list<int>) -> !torch.list<int>\n"
 "    return %0 : !torch.list<int>\n"
 "  }\n"
+"  func.func @\"__torch_mlir_shape_fn.aten.trunc\"(%arg0: !torch.list<int>) -> !torch.list<int> {\n"
+"    %0 = call @__torch__.torch.jit._shape_functions.unary(%arg0) : (!torch.list<int>) -> !torch.list<int>\n"
+"    return %0 : !torch.list<int>\n"
+"  }\n"
 "  func.func @\"__torch_mlir_shape_fn.aten.log\"(%arg0: !torch.list<int>) -> !torch.list<int> {\n"
 "    %0 = call @__torch__.torch.jit._shape_functions.unary(%arg0) : (!torch.list<int>) -> !torch.list<int>\n"
 "    return %0 : !torch.list<int>\n"
@@ -10003,6 +10007,10 @@ StringRef mlir::torch::Torch::getAbstractInterpLibrary() {
 "    %0:2 = torch.prim.TupleUnpack %arg0 : !torch.tuple<int, int> -> !torch.int, !torch.int\n"
 "    return %0#1 : !torch.int\n"
 "  }\n"
+"  func.func @\"__torch_mlir_dtype_fn.aten.trunc\"(%arg0: !torch.tuple<int, int>) -> !torch.int {\n"
+"    %0:2 = torch.prim.TupleUnpack %arg0 : !torch.tuple<int, int> -> !torch.int, !torch.int\n"
+"    return %0#1 : !torch.int\n"
+"  }\n"
 "  func.func @\"__torch_mlir_dtype_fn.aten.clamp_max\"(%arg0: !torch.tuple<int, int>, %arg1: !torch.number) -> !torch.int {\n"
 "    %int4 = torch.constant.int 4\n"
 "    %int11 = torch.constant.int 11\n"

lib/Dialect/Torch/Transforms/DecomposeComplexOps.cpp

@@ -5886,6 +5886,32 @@ class DecomposeAtenCosineSimilarityOp
 };
 } // namespace
 
+namespace {
+// decompose `trunc(x)` to `sign(x) * floor(abs(x))`
+class DecomposeAtenTruncOp : public OpRewritePattern<AtenTruncOp> {
+  using OpRewritePattern::OpRewritePattern;
+  LogicalResult matchAndRewrite(AtenTruncOp op,
+                                PatternRewriter &rewriter) const override {
+    Location loc = op.getLoc();
+    Value self = op.getSelf();
+
+    auto resultTy = dyn_cast<ValueTensorType>(op.getType());
+    if (!resultTy || !resultTy.hasDtype()) {
+      return rewriter.notifyMatchFailure(op, "result must have dtype");
+    }
+
+    if (isa<mlir::FloatType>(resultTy.getDtype())) {
+      Value sign = rewriter.create<AtenSgnOp>(loc, resultTy, self);
+      Value abs = rewriter.create<AtenAbsOp>(loc, resultTy, self);
+      Value floor = rewriter.create<AtenFloorOp>(loc, resultTy, abs);
+      rewriter.replaceOpWithNewOp<AtenMulTensorOp>(op, resultTy, sign, floor);
+      return success();
+    }
+    return failure();
+  }
+};
+} // namespace
+
 namespace {
 // Decompose `aten.baddbmm` op into `aten.bmm`, `aten.mul.Scalar`, and
 // `aten.add.Tensor` op.
@@ -7700,6 +7726,7 @@ public:
     addPatternIfTargetOpIsIllegal<DecomposeAtenClampMinTensorOp>(patterns);
     addPatternIfTargetOpIsIllegal<DecomposeAtenClampMaxOp>(patterns);
     addPatternIfTargetOpIsIllegal<DecomposeAtenCosineSimilarityOp>(patterns);
+    addPatternIfTargetOpIsIllegal<DecomposeAtenTruncOp>(patterns);
     addPatternIfTargetOpIsIllegal<DecomposeAtenBaddbmmOp>(patterns);
     addPatternIfTargetOpIsIllegal<DecomposeAtenFloorDivideOp>(patterns);
     addPatternIfTargetOpIsIllegal<DecomposeAtenFloorDivideScalarOp>(patterns);

lib/Dialect/Torch/Transforms/LowerToBackendContract.cpp

@@ -512,6 +512,7 @@ static void markDecomposedOpsAsIllegal(MLIRContext *context,
   target.addIllegalOp<AtenNormalFunctionalOp>();
   target.addIllegalOp<AtenVarMeanOp>();
   target.addIllegalOp<AtenCosineSimilarityOp>();
+  target.addIllegalOp<AtenTruncOp>();
   target.addIllegalOp<AtenNewEmptyStridedOp>();
   target.addIllegalOp<AtenEmptyStridedOp>();
   target.addIllegalOp<AtenBucketizeTensorOp>();

projects/pt1/e2e_testing/xfail_sets.py

@@ -1479,6 +1479,8 @@ STABLEHLO_PASS_SET = {
     "ElementwiseCoshModule_basic",
     "ElementwiseSinhIntModule_basic",
     "ElementwiseSinhModule_basic",
+    "ElementwiseTruncIntModule_basic",
+    "ElementwiseTruncModule_basic",
 }
 
 STABLEHLO_CRASHING_SET =  {
@@ -1488,6 +1490,8 @@ STABLEHLO_CRASHING_SET =  {
 # Write the TOSA set as a "passing" set as it is very early in development
 # and very few tests work yet.
 TOSA_PASS_SET = {
+    "ElementwiseTruncModule_basic",
+    "ElementwiseTruncIntModule_basic",
     "ElementwiseSgnModule_basic",
     "ElementwiseSignIntModule_basic",
     "AdaptiveAvgPool2dNonUnitOutputSizeStaticModule_basic",
@@ -2344,6 +2348,8 @@ ONNX_XFAIL_SET = {
     "ElementwiseSinhModule_basic",
     "ElementwiseCoshIntModule_basic",
     "ElementwiseCoshModule_basic",
+    "ElementwiseTruncIntModule_basic",
+    "ElementwiseTruncModule_basic",
     "ElementwiseDequantizePerChannelModule_basic",
     "ElementwiseDequantizePerTensorModule_basic",
     "ElementwiseDivTensorRoundingModeFloorIntStaticModule_basic",

projects/pt1/python/torch_mlir/jit_ir_importer/build_tools/abstract_interp_lib_gen.py

@@ -245,6 +245,9 @@ def aten〇hardtanh_backward〡shape(grad_output: List[int], self: List[int], mi
 def aten〇ceil〡shape(self: List[int]) -> List[int]:
     return upstream_shape_functions.unary(self)
 
+def aten〇trunc〡shape(self: List[int]) -> List[int]:
+    return upstream_shape_functions.unary(self)
+
 def aten〇log〡shape(self: List[int]) -> List[int]:
     return upstream_shape_functions.unary(self)
 
@@ -2227,6 +2230,11 @@ def aten〇ceil〡dtype(self_rank_dtype: Tuple[int, int]) -> int:
     self_rank, self_dtype = self_rank_dtype
     return self_dtype
 
+@check_dtype_function(_check_tensors_with_the_same_dtype(num_of_tensors=1))
+def aten〇trunc〡dtype(self_rank_dtype: Tuple[int, int]) -> int:
+    self_rank, self_dtype = self_rank_dtype
+    return self_dtype
+
 @check_dtype_function(_check_tensors_with_the_same_dtype(num_of_tensors=1, max=0))
 def aten〇clamp_max〡dtype(self_rank_dtype: Tuple[int, int], max: Union[int, float, complex]) -> int:
     self_rank, self_dtype = self_rank_dtype

projects/pt1/python/torch_mlir/jit_ir_importer/build_tools/torch_ods_gen.py

@@ -359,6 +359,7 @@ def emit_ops(emitter_td: TextEmitter, registry: Registry):
     emit_with_mutating_variants("aten::floor : (Tensor) -> (Tensor)", has_folder=True)
     emit_with_mutating_variants("aten::ceil : (Tensor) -> (Tensor)", has_folder=True)
     emit_with_mutating_variants("aten::round : (Tensor) -> (Tensor)", has_folder=True)
+    emit_with_mutating_variants("aten::trunc : (Tensor) -> (Tensor)", has_folder=True)
     emit_with_mutating_variants("aten::sign : (Tensor) -> (Tensor)", has_canonicalizer=True)
     emit_with_mutating_variants("aten::masked_fill.Tensor : (Tensor, Tensor, Tensor) -> (Tensor)", has_canonicalizer=True)
 

projects/pt1/python/torch_mlir_e2e_test/test_suite/elementwise.py

@@ -2077,6 +2077,50 @@ def ElementwiseCeilModule_basic(module, tu: TestUtils):
 # ==============================================================================
 
 
+class ElementwiseTruncModule(torch.nn.Module):
+
+    def __init__(self):
+        super().__init__()
+
+    @export
+    @annotate_args([
+        None,
+        ([1, 6], torch.float32, True),
+    ])
+    def forward(self, a):
+        return torch.trunc(a)
+
+
+@register_test_case(module_factory=lambda: ElementwiseTruncModule())
+def ElementwiseTruncModule_basic(module, tu: TestUtils):
+    module.forward(torch.tensor([[-torch.inf, torch.inf, torch.nan, -2.3, 0.0, 1.5]]))
+
+
+# ==============================================================================
+
+
+class ElementwiseTruncIntModule(torch.nn.Module):
+
+    def __init__(self):
+        super().__init__()
+
+    @export
+    @annotate_args([
+        None,
+        ([-1, -1], torch.int32, True),
+    ])
+    def forward(self, a):
+        return torch.trunc(a)
+
+
+@register_test_case(module_factory=lambda: ElementwiseTruncIntModule())
+def ElementwiseTruncIntModule_basic(module, tu: TestUtils):
+    module.forward(tu.randint(3, 4, low=-100, high=100).to(torch.int32))
+
+
+# ==============================================================================
+
+
 class ElementwiseSignModule(torch.nn.Module):
 
     def __init__(self):

test/Dialect/Torch/canonicalize.mlir

@@ -2308,6 +2308,14 @@ func.func @torch.aten.floor$canonicalize(%arg0: !torch.vtensor<[?,?],si64>) -> !
   return %0 : !torch.vtensor<[?,?],si64>
 }
 
+// CHECK-LABEL:   func.func @torch.aten.trunc$canonicalize
+// CHECK-SAME:            %[[ARG:.*]]: !torch.vtensor<[?,?],si64>
+// CHECK-NEXT:      return %[[ARG]] : !torch.vtensor<[?,?],si64>
+func.func @torch.aten.trunc$canonicalize(%arg0: !torch.vtensor<[?,?],si64>) -> !torch.vtensor<[?,?],si64> {
+  %0 = torch.aten.trunc %arg0 : !torch.vtensor<[?,?],si64> -> !torch.vtensor<[?,?],si64>
+  return %0 : !torch.vtensor<[?,?],si64>
+}
+
 // CHECK-LABEL:   func.func @torch.aten.numel$canonicalize
 // CHECK-SAME:            %[[ARG:.*]]: !torch.vtensor<[3,4],f32>
 // CHECK-NEXT:      %int12 = torch.constant.int 12