[Torch] emit and lowering frac, signbit, ldexp, copysign ops (#3851)

also fix `aten.exp2` with integer type

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

@@ -1483,6 +1483,51 @@ def Torch_AtenNeg_Op : Torch_Op<"aten.neg_", [
   }];
 }
 
+def Torch_AtenFracOp : Torch_Op<"aten.frac", [
+    AllowsTypeRefinement,
+    HasValueSemantics,
+    ReadOnly
+  ]> {
+  let summary = "Generated op for `aten::frac : (Tensor) -> (Tensor)`";
+  let arguments = (ins
+    AnyTorchTensorType:$self
+  );
+  let results = (outs
+    AnyTorchOptionalTensorType:$result
+  );
+  let hasCustomAssemblyFormat = 1;
+  let extraClassDefinition = [{
+    ParseResult AtenFracOp::parse(OpAsmParser &parser, OperationState &result) {
+      return parseDefaultTorchOp(parser, result, 1, 1);
+    }
+    void AtenFracOp::print(OpAsmPrinter &printer) {
+      printDefaultTorchOp(printer, *this, 1, 1);
+    }
+  }];
+}
+
+def Torch_AtenFrac_Op : Torch_Op<"aten.frac_", [
+    IsTrailingUnderscoreInplaceVariant,
+    AllowsTypeRefinement
+  ]> {
+  let summary = "Generated op for `aten::frac_ : (Tensor) -> (Tensor)`";
+  let arguments = (ins
+    Torch_NonValueTensorType:$self
+  );
+  let results = (outs
+    AnyTorchOptionalNonValueTensorType:$result
+  );
+  let hasCustomAssemblyFormat = 1;
+  let extraClassDefinition = [{
+    ParseResult AtenFrac_Op::parse(OpAsmParser &parser, OperationState &result) {
+      return parseDefaultTorchOp(parser, result, 1, 1);
+    }
+    void AtenFrac_Op::print(OpAsmPrinter &printer) {
+      printDefaultTorchOp(printer, *this, 1, 1);
+    }
+  }];
+}
+
 def Torch_AtenBitwiseNotOp : Torch_Op<"aten.bitwise_not", [
     AllowsTypeRefinement,
     HasValueSemantics,
@@ -3400,6 +3445,53 @@ def Torch_AtenFill_TensorOp : Torch_Op<"aten.fill_.Tensor", [
   }];
 }
 
+def Torch_AtenCopysignTensorOp : Torch_Op<"aten.copysign.Tensor", [
+    AllowsTypeRefinement,
+    HasValueSemantics,
+    ReadOnly
+  ]> {
+  let summary = "Generated op for `aten::copysign.Tensor : (Tensor, Tensor) -> (Tensor)`";
+  let arguments = (ins
+    AnyTorchTensorType:$self,
+    AnyTorchTensorType:$other
+  );
+  let results = (outs
+    AnyTorchOptionalTensorType:$result
+  );
+  let hasCustomAssemblyFormat = 1;
+  let extraClassDefinition = [{
+    ParseResult AtenCopysignTensorOp::parse(OpAsmParser &parser, OperationState &result) {
+      return parseDefaultTorchOp(parser, result, 2, 1);
+    }
+    void AtenCopysignTensorOp::print(OpAsmPrinter &printer) {
+      printDefaultTorchOp(printer, *this, 2, 1);
+    }
+  }];
+}
+
+def Torch_AtenCopysign_TensorOp : Torch_Op<"aten.copysign_.Tensor", [
+    IsTrailingUnderscoreInplaceVariant,
+    AllowsTypeRefinement
+  ]> {
+  let summary = "Generated op for `aten::copysign_.Tensor : (Tensor, Tensor) -> (Tensor)`";
+  let arguments = (ins
+    Torch_NonValueTensorType:$self,
+    Torch_NonValueTensorType:$other
+  );
+  let results = (outs
+    AnyTorchOptionalNonValueTensorType:$result
+  );
+  let hasCustomAssemblyFormat = 1;
+  let extraClassDefinition = [{
+    ParseResult AtenCopysign_TensorOp::parse(OpAsmParser &parser, OperationState &result) {
+      return parseDefaultTorchOp(parser, result, 2, 1);
+    }
+    void AtenCopysign_TensorOp::print(OpAsmPrinter &printer) {
+      printDefaultTorchOp(printer, *this, 2, 1);
+    }
+  }];
+}
+
 def Torch_AtenUnsqueezeOp : Torch_Op<"aten.unsqueeze", [
     AllowsTypeRefinement,
     ReadOnly
@@ -3850,6 +3942,53 @@ def Torch_AtenMul_ScalarOp : Torch_Op<"aten.mul_.Scalar", [
   }];
 }
 
+def Torch_AtenLdexpTensorOp : Torch_Op<"aten.ldexp.Tensor", [
+    AllowsTypeRefinement,
+    HasValueSemantics,
+    ReadOnly
+  ]> {
+  let summary = "Generated op for `aten::ldexp.Tensor : (Tensor, Tensor) -> (Tensor)`";
+  let arguments = (ins
+    AnyTorchTensorType:$self,
+    AnyTorchTensorType:$other
+  );
+  let results = (outs
+    AnyTorchOptionalTensorType:$result
+  );
+  let hasCustomAssemblyFormat = 1;
+  let extraClassDefinition = [{
+    ParseResult AtenLdexpTensorOp::parse(OpAsmParser &parser, OperationState &result) {
+      return parseDefaultTorchOp(parser, result, 2, 1);
+    }
+    void AtenLdexpTensorOp::print(OpAsmPrinter &printer) {
+      printDefaultTorchOp(printer, *this, 2, 1);
+    }
+  }];
+}
+
+def Torch_AtenSignbitOp : Torch_Op<"aten.signbit", [
+    AllowsTypeRefinement,
+    HasValueSemantics,
+    ReadOnly
+  ]> {
+  let summary = "Generated op for `aten::signbit : (Tensor) -> (Tensor)`";
+  let arguments = (ins
+    AnyTorchTensorType:$self
+  );
+  let results = (outs
+    AnyTorchOptionalTensorType:$result
+  );
+  let hasCustomAssemblyFormat = 1;
+  let extraClassDefinition = [{
+    ParseResult AtenSignbitOp::parse(OpAsmParser &parser, OperationState &result) {
+      return parseDefaultTorchOp(parser, result, 1, 1);
+    }
+    void AtenSignbitOp::print(OpAsmPrinter &printer) {
+      printDefaultTorchOp(printer, *this, 1, 1);
+    }
+  }];
+}
+
 def Torch_AtenEqTensorOp : Torch_Op<"aten.eq.Tensor", [
     AllowsTypeRefinement,
     HasValueSemantics,

lib/Dialect/Torch/Transforms/AbstractInterpLibrary.cpp

@@ -9154,6 +9154,20 @@ StringRef mlir::torch::Torch::getAbstractInterpLibrary() {
 "    %0 = call @__torch__.torch.jit._shape_functions.broadcast(%arg0, %arg1) : (!torch.list<int>, !torch.list<int>) -> !torch.list<int>\n"
 "    return %0 : !torch.list<int>\n"
 "  }\n"
+"  func.func @\"__torch_mlir_shape_fn.aten.frac\"(%arg0: !torch.list<int>) -> !torch.list<int> {\n"
+"    return %arg0 : !torch.list<int>\n"
+"  }\n"
+"  func.func @\"__torch_mlir_shape_fn.aten.signbit\"(%arg0: !torch.list<int>) -> !torch.list<int> {\n"
+"    return %arg0 : !torch.list<int>\n"
+"  }\n"
+"  func.func @\"__torch_mlir_shape_fn.aten.ldexp.Tensor\"(%arg0: !torch.list<int>, %arg1: !torch.list<int>) -> !torch.list<int> {\n"
+"    %0 = call @__torch__.torch.jit._shape_functions.broadcast(%arg0, %arg1) : (!torch.list<int>, !torch.list<int>) -> !torch.list<int>\n"
+"    return %0 : !torch.list<int>\n"
+"  }\n"
+"  func.func @\"__torch_mlir_shape_fn.aten.copysign.Tensor\"(%arg0: !torch.list<int>, %arg1: !torch.list<int>) -> !torch.list<int> {\n"
+"    %0 = call @__torch__.torch.jit._shape_functions.broadcast(%arg0, %arg1) : (!torch.list<int>, !torch.list<int>) -> !torch.list<int>\n"
+"    return %0 : !torch.list<int>\n"
+"  }\n"
 "  func.func @\"__torch_mlir_shape_fn.aten.__and__.Tensor\"(%arg0: !torch.list<int>, %arg1: !torch.list<int>) -> !torch.list<int> {\n"
 "    %0 = call @__torch__.torch.jit._shape_functions.broadcast(%arg0, %arg1) : (!torch.list<int>, !torch.list<int>) -> !torch.list<int>\n"
 "    return %0 : !torch.list<int>\n"
@@ -12825,6 +12839,93 @@ StringRef mlir::torch::Torch::getAbstractInterpLibrary() {
 "    %4 = call @__torch__.torch_mlir.jit_ir_importer.build_tools.library_generator.promote_dtypes(%1, %3) : (!torch.list<optional<int>>, !torch.list<int>) -> !torch.int\n"
 "    return %4 : !torch.int\n"
 "  }\n"
+"  func.func @\"__torch_mlir_dtype_fn.aten.frac\"(%arg0: !torch.tuple<int, int>) -> !torch.int {\n"
+"    %none = torch.constant.none\n"
+"    %str = torch.constant.str \"AssertionError: \"\n"
+"    %int11 = torch.constant.int 11\n"
+"    %0:2 = torch.prim.TupleUnpack %arg0 : !torch.tuple<int, int> -> !torch.int, !torch.int\n"
+"    %1 = torch.aten.ne.int %0#1, %int11 : !torch.int, !torch.int -> !torch.bool\n"
+"    torch.prim.If %1 -> () {\n"
+"      torch.prim.If.yield\n"
+"    } else {\n"
+"      torch.prim.RaiseException %str, %none : !torch.str, !torch.none\n"
+"      torch.prim.If.yield\n"
+"    }\n"
+"    return %0#1 : !torch.int\n"
+"  }\n"
+"  func.func @\"__torch_mlir_dtype_fn.aten.signbit\"(%arg0: !torch.tuple<int, int>) -> !torch.int {\n"
+"    %int11 = torch.constant.int 11\n"
+"    return %int11 : !torch.int\n"
+"  }\n"
+"  func.func @\"__torch_mlir_dtype_fn.aten.ldexp.Tensor\"(%arg0: !torch.tuple<int, int>, %arg1: !torch.tuple<int, int>) -> !torch.int {\n"
+"    %int6 = torch.constant.int 6\n"
+"    %false = torch.constant.bool false\n"
+"    %int7 = torch.constant.int 7\n"
+"    %0:2 = torch.prim.TupleUnpack %arg0 : !torch.tuple<int, int> -> !torch.int, !torch.int\n"
+"    %1:2 = torch.prim.TupleUnpack %arg1 : !torch.tuple<int, int> -> !torch.int, !torch.int\n"
+"    %2 = torch.prim.ListConstruct %0#0, %1#0 : (!torch.int, !torch.int) -> !torch.list<optional<int>>\n"
+"    %3 = torch.prim.ListConstruct %0#1, %1#1 : (!torch.int, !torch.int) -> !torch.list<int>\n"
+"    %4 = torch.aten.eq.int %0#1, %int7 : !torch.int, !torch.int -> !torch.bool\n"
+"    %5 = torch.prim.If %4 -> (!torch.bool) {\n"
+"      %7 = func.call @__torch__.torch_mlir.jit_ir_importer.build_tools.library_generator.is_complex_dtype(%1#1) : (!torch.int) -> !torch.bool\n"
+"      torch.prim.If.yield %7 : !torch.bool\n"
+"    } else {\n"
+"      torch.prim.If.yield %false : !torch.bool\n"
+"    }\n"
+"    %6 = torch.prim.If %5 -> (!torch.int) {\n"
+"      torch.prim.If.yield %1#1 : !torch.int\n"
+"    } else {\n"
+"      %7 = func.call @__torch__.torch_mlir.jit_ir_importer.build_tools.library_generator.is_complex_dtype(%0#1) : (!torch.int) -> !torch.bool\n"
+"      %8 = torch.prim.If %7 -> (!torch.bool) {\n"
+"        %10 = torch.aten.eq.int %1#1, %int7 : !torch.int, !torch.int -> !torch.bool\n"
+"        torch.prim.If.yield %10 : !torch.bool\n"
+"      } else {\n"
+"        torch.prim.If.yield %false : !torch.bool\n"
+"      }\n"
+"      %9 = torch.prim.If %8 -> (!torch.int) {\n"
+"        torch.prim.If.yield %0#1 : !torch.int\n"
+"      } else {\n"
+"        %10 = func.call @__torch__.torch_mlir.jit_ir_importer.build_tools.library_generator.is_integer_dtype(%0#1) : (!torch.int) -> !torch.bool\n"
+"        %11 = torch.prim.If %10 -> (!torch.bool) {\n"
+"          %13 = func.call @__torch__.torch_mlir.jit_ir_importer.build_tools.library_generator.is_integer_dtype(%1#1) : (!torch.int) -> !torch.bool\n"
+"          torch.prim.If.yield %13 : !torch.bool\n"
+"        } else {\n"
+"          torch.prim.If.yield %false : !torch.bool\n"
+"        }\n"
+"        %12 = torch.prim.If %11 -> (!torch.int) {\n"
+"          torch.prim.If.yield %int6 : !torch.int\n"
+"        } else {\n"
+"          %13 = func.call @__torch__.torch_mlir.jit_ir_importer.build_tools.library_generator.promote_dtypes(%2, %3) : (!torch.list<optional<int>>, !torch.list<int>) -> !torch.int\n"
+"          torch.prim.If.yield %13 : !torch.int\n"
+"        }\n"
+"        torch.prim.If.yield %12 : !torch.int\n"
+"      }\n"
+"      torch.prim.If.yield %9 : !torch.int\n"
+"    }\n"
+"    return %6 : !torch.int\n"
+"  }\n"
+"  func.func @\"__torch_mlir_dtype_fn.aten.copysign.Tensor\"(%arg0: !torch.tuple<int, int>, %arg1: !torch.tuple<int, int>) -> !torch.int {\n"
+"    %int6 = torch.constant.int 6\n"
+"    %false = torch.constant.bool false\n"
+"    %0:2 = torch.prim.TupleUnpack %arg0 : !torch.tuple<int, int> -> !torch.int, !torch.int\n"
+"    %1:2 = torch.prim.TupleUnpack %arg1 : !torch.tuple<int, int> -> !torch.int, !torch.int\n"
+"    %2 = torch.prim.ListConstruct %0#0, %1#0 : (!torch.int, !torch.int) -> !torch.list<optional<int>>\n"
+"    %3 = torch.prim.ListConstruct %0#1, %1#1 : (!torch.int, !torch.int) -> !torch.list<int>\n"
+"    %4 = call @__torch__.torch_mlir.jit_ir_importer.build_tools.library_generator.is_integer_dtype(%0#1) : (!torch.int) -> !torch.bool\n"
+"    %5 = torch.prim.If %4 -> (!torch.bool) {\n"
+"      %7 = func.call @__torch__.torch_mlir.jit_ir_importer.build_tools.library_generator.is_integer_dtype(%1#1) : (!torch.int) -> !torch.bool\n"
+"      torch.prim.If.yield %7 : !torch.bool\n"
+"    } else {\n"
+"      torch.prim.If.yield %false : !torch.bool\n"
+"    }\n"
+"    %6 = torch.prim.If %5 -> (!torch.int) {\n"
+"      torch.prim.If.yield %int6 : !torch.int\n"
+"    } else {\n"
+"      %7 = func.call @__torch__.torch_mlir.jit_ir_importer.build_tools.library_generator.promote_dtypes(%2, %3) : (!torch.list<optional<int>>, !torch.list<int>) -> !torch.int\n"
+"      torch.prim.If.yield %7 : !torch.int\n"
+"    }\n"
+"    return %6 : !torch.int\n"
+"  }\n"
 "  func.func @\"__torch_mlir_dtype_fn.aten.__and__.Scalar\"(%arg0: !torch.tuple<int, int>, %arg1: !torch.number) -> !torch.int {\n"
 "    %none = torch.constant.none\n"
 "    %0:2 = torch.prim.TupleUnpack %arg0 : !torch.tuple<int, int> -> !torch.int, !torch.int\n"

lib/Dialect/Torch/Transforms/DecomposeComplexOps.cpp

@@ -7490,6 +7490,160 @@ class DecomposeAtenTruncOp : public OpRewritePattern<AtenTruncOp> {
 };
 } // namespace
 
+namespace {
+// decompose `signbit(x)` to `view.dtype(x, si32/si64) < 0 `
+class DecomposeAtenSignbitOp : public OpRewritePattern<AtenSignbitOp> {
+  using OpRewritePattern::OpRewritePattern;
+  LogicalResult matchAndRewrite(AtenSignbitOp op,
+                                PatternRewriter &rewriter) const override {
+    Location loc = op.getLoc();
+    Value self = op.getSelf();
+
+    auto operandTy = dyn_cast<ValueTensorType>(self.getType());
+    auto resultTy = dyn_cast<ValueTensorType>(op.getType());
+    if (!operandTy || !operandTy.hasDtype() || !resultTy ||
+        !resultTy.hasDtype()) {
+      return rewriter.notifyMatchFailure(op,
+                                         "operand and result must have dtype");
+    }
+
+    if (isa<mlir::FloatType>(operandTy.getDtype())) {
+      mlir::IntegerType intType = rewriter.getIntegerType(
+          operandTy.getDtype().getIntOrFloatBitWidth(), /*isSigned*/ true);
+      Value dtype = getDtypeIntValueForType(rewriter, loc, intType);
+      Value view = rewriter.create<AtenViewDtypeOp>(
+          loc,
+          operandTy.getWithSizesAndDtype(operandTy.getOptionalSizes(), intType),
+          self, dtype);
+      Value zero =
+          rewriter.create<ConstantIntOp>(loc, rewriter.getI64IntegerAttr(0));
+      Value shift = rewriter.create<AtenLtScalarOp>(loc, resultTy, view, zero);
+      rewriter.replaceOp(op, shift);
+      return success();
+    } else if (isa<mlir::IntegerType>(operandTy.getDtype())) {
+      Value zero =
+          rewriter.create<ConstantIntOp>(loc, rewriter.getI64IntegerAttr(0));
+      Value shift = rewriter.create<AtenLtScalarOp>(loc, resultTy, self, zero);
+      rewriter.replaceOp(op, shift);
+    }
+    return failure();
+  }
+};
+} // namespace
+
+namespace {
+// decompose `frac(x)` to `x - trunc(x)`
+class DecomposeAtenFracOp : public OpRewritePattern<AtenFracOp> {
+  using OpRewritePattern::OpRewritePattern;
+  LogicalResult matchAndRewrite(AtenFracOp op,
+                                PatternRewriter &rewriter) const override {
+    Location loc = op.getLoc();
+    Value self = op.getSelf();
+    auto resultTy = op.getType();
+
+    Value one =
+        rewriter.create<ConstantIntOp>(loc, rewriter.getI64IntegerAttr(1));
+    Value trunc = rewriter.create<AtenTruncOp>(loc, resultTy, self);
+    rewriter.replaceOpWithNewOp<AtenSubTensorOp>(op, resultTy, self, trunc,
+                                                 /*alpha=*/one);
+    return success();
+  }
+};
+} // namespace
+
+namespace {
+// decompose `copysign(x, y)` to `signbit(y) ? -abs(x) : abs(x)`
+class DecomposeAtenCopysignTensorOp
+    : public OpRewritePattern<AtenCopysignTensorOp> {
+  using OpRewritePattern::OpRewritePattern;
+  LogicalResult matchAndRewrite(AtenCopysignTensorOp op,
+                                PatternRewriter &rewriter) const override {
+    Location loc = op.getLoc();
+    Value self = op.getSelf();
+    Value other = op.getOther();
+    auto selfTy = self.getType();
+    auto otherTy = cast<BaseTensorType>(other.getType());
+    auto resultTy = op.getType();
+
+    Value signbit = rewriter.create<AtenSignbitOp>(
+        loc,
+        otherTy.getWithSizesAndDtype(otherTy.getOptionalSizes(),
+                                     rewriter.getI1Type()),
+        other);
+    Value abs = rewriter.create<AtenAbsOp>(loc, selfTy, self);
+    Value neg = rewriter.create<AtenNegOp>(loc, selfTy, abs);
+    rewriter.replaceOpWithNewOp<AtenWhereSelfOp>(op, resultTy, signbit, neg,
+                                                 abs);
+    return success();
+  }
+};
+} // namespace
+
+namespace {
+// decompose `ldexp(x, y)` to `x * 2^y`
+class DecomposeAtenLdexpTensorOp : public OpRewritePattern<AtenLdexpTensorOp> {
+  using OpRewritePattern::OpRewritePattern;
+  LogicalResult matchAndRewrite(AtenLdexpTensorOp op,
+                                PatternRewriter &rewriter) const override {
+    Location loc = op.getLoc();
+    Value self = op.getSelf();
+    Value other = op.getOther();
+
+    auto otherTy = dyn_cast<BaseTensorType>(other.getType());
+    auto resultTy = dyn_cast<ValueTensorType>(op.getType());
+    if (!resultTy || !resultTy.hasDtype()) {
+      return rewriter.notifyMatchFailure(op, "result must have dtype");
+    }
+
+    Value exp2 = rewriter.create<AtenExp2Op>(
+        loc,
+        resultTy.getWithSizesAndDtype(otherTy.getOptionalSizes(),
+                                      resultTy.getDtype()),
+        other);
+    rewriter.replaceOpWithNewOp<AtenMulTensorOp>(op, resultTy, self, exp2);
+    return success();
+  }
+};
+} // namespace
+
+namespace {
+// decompose `fmod(x, y)` to `x - trunc(x/y) * y`
+class DecomposeAtenFmodTensorOp : public OpRewritePattern<AtenFmodTensorOp> {
+  using OpRewritePattern::OpRewritePattern;
+  LogicalResult matchAndRewrite(AtenFmodTensorOp op,
+                                PatternRewriter &rewriter) const override {
+    Location loc = op.getLoc();
+    Value self = op.getSelf();
+    Value other = op.getOther();
+
+    auto resultTy = dyn_cast<ValueTensorType>(op.getType());
+    if (!resultTy || !resultTy.hasDtype()) {
+      return rewriter.notifyMatchFailure(op, "result must have dtype");
+    }
+
+    if (isa<mlir::IntegerType>(resultTy.getDtype())) {
+      Value div = rewriter.create<AtenDivTensorOp>(loc, resultTy, self, other);
+      Value mul = rewriter.create<AtenMulTensorOp>(loc, resultTy, div, other);
+      Value alpha =
+          rewriter.create<ConstantFloatOp>(loc, rewriter.getF64FloatAttr(1));
+      rewriter.replaceOpWithNewOp<AtenSubTensorOp>(op, resultTy, self, mul,
+                                                   alpha);
+      return success();
+    } else if (isa<mlir::FloatType>(resultTy.getDtype())) {
+      Value div = rewriter.create<AtenDivTensorOp>(loc, resultTy, self, other);
+      Value trunc = rewriter.create<AtenTruncOp>(loc, resultTy, div);
+      Value mul = rewriter.create<AtenMulTensorOp>(loc, resultTy, trunc, other);
+      Value alpha =
+          rewriter.create<ConstantFloatOp>(loc, rewriter.getF64FloatAttr(1));
+      rewriter.replaceOpWithNewOp<AtenSubTensorOp>(op, resultTy, self, mul,
+                                                   alpha);
+      return success();
+    }
+    return failure();
+  }
+};
+} // namespace
+
 namespace {
 // Decompose `aten.baddbmm` op into `aten.bmm`, `aten.mul.Scalar`, and
 // `aten.add.Tensor` op.
@@ -8598,10 +8752,16 @@ class DecomposeAtenExp2Op : public OpRewritePattern<AtenExp2Op> {
     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");
+    }
+
     auto two =
         rewriter.create<ConstantIntOp>(loc, rewriter.getI64IntegerAttr(2));
-    rewriter.replaceOpWithNewOp<AtenPowScalarOp>(op, op.getType(), two, self);
+    Value to = convertTensorToDtype(rewriter, loc, self, resultTy.getDtype());
-
+    Value pow = rewriter.create<AtenPowScalarOp>(loc, resultTy, two, to);
+    rewriter.replaceOp(op, pow);
     return success();
   }
 };
@@ -9696,6 +9856,11 @@ public:
     addPatternIfTargetOpIsIllegal<DecomposeAtenRad2degOp>(patterns);
     addPatternIfTargetOpIsIllegal<DecomposeAtenCosineSimilarityOp>(patterns);
     addPatternIfTargetOpIsIllegal<DecomposeAtenTruncOp>(patterns);
+    addPatternIfTargetOpIsIllegal<DecomposeAtenSignbitOp>(patterns);
+    addPatternIfTargetOpIsIllegal<DecomposeAtenFracOp>(patterns);
+    addPatternIfTargetOpIsIllegal<DecomposeAtenCopysignTensorOp>(patterns);
+    addPatternIfTargetOpIsIllegal<DecomposeAtenLdexpTensorOp>(patterns);
+    addPatternIfTargetOpIsIllegal<DecomposeAtenFmodTensorOp>(patterns);
     addPatternIfTargetOpIsIllegal<DecomposeAtenBaddbmmOp>(patterns);
     addPatternIfTargetOpIsIllegal<DecomposeAtenFloorDivideOp>(patterns);
     addPatternIfTargetOpIsIllegal<DecomposeAtenFloorDivideScalarOp>(patterns);

lib/Dialect/Torch/Transforms/LowerToBackendContract.cpp

@@ -535,6 +535,10 @@ static void markDecomposedOpsAsIllegal(MLIRContext *context,
   target.addIllegalOp<AtenRad2degOp>();
   target.addIllegalOp<AtenCosineSimilarityOp>();
   target.addIllegalOp<AtenTruncOp>();
+  target.addIllegalOp<AtenSignbitOp>();
+  target.addIllegalOp<AtenFracOp>();
+  target.addIllegalOp<AtenCopysignTensorOp>();
+  target.addIllegalOp<AtenLdexpTensorOp>();
   target.addIllegalOp<AtenNewEmptyStridedOp>();
   target.addIllegalOp<AtenEmptyStridedOp>();
   target.addIllegalOp<AtenBucketizeTensorOp>();

projects/pt1/e2e_testing/xfail_sets.py

@@ -500,6 +500,52 @@ FX_IMPORTER_XFAIL_SET = {
     "ViewSizeFromOtherTensor_basic",
     "ViewDtypeStaticModule_basic",
     "WeightNormInterfaceModule_basic",
+    # Error: `aten.as_strided` op is not supported
+    "ChunkListUnpackDynamic_Module_basic",
+    "ChunkListUnpackUnevenDynamic_Module_basic",
+    "ChunkListUnpackUneven_Module_basic",
+    "ChunkListUnpack_Module_basic",
+    "SplitTensorGetItem_Module_basic",
+    "SplitTensorLastSmallerModule_basic",
+    "SplitTensorListUnpackModule_basic",
+    "SplitTensorNegativeDimModule_basic",
+    "SplitWithSizesListUnpackModule_basic",
+    "SplitWithSizes_Module_basic",
+    "AdaptiveAvgPool1dGeneralDynamic_basic",
+    "AdaptiveAvgPool1dStaticEvenMultiple_basic",
+    "AdaptiveAvgPool1dStaticLargerOutput_basic",
+    "AdaptiveAvgPool2dDynamicNoBatch_basic",
+    "AdaptiveAvgPool2dDynamic_basic",
+    "AdaptiveMaxPool1dDynamicNoBatch_basic",
+    "AdaptiveMaxPool1dDynamic_basic",
+    "AdaptiveMaxPool1dStatic_basic",
+    "CrossEntropyLossModule_basic",
+    "CrossEntropyLossNoReductionModule_basic",
+    "ElementwiseExpm1IntModule_basic",
+    "ElementwiseExpm1Module_basic",
+    "IndexPutImpl1DFloatAccumulateModule_basic",
+    "IndexPutImpl1DFloatNonAccumulateModule_basic",
+    "IndexPutImpl1DIntAccumulateModule_basic",
+    "IndexPutImpl1DIntNonAccumulateModule_basic",
+    "IndexPutImpl2DFloatNonAccumulateModule_basic",
+    "IndexPutImpl2DImplicitModule_basic",
+    "IndexPutImpl2DIndexModule_basic",
+    "IndexPutImpl2DNoneIndexStaticModule_basic",
+    "IndexPutImpl3DFloatNonAccumulateModule_basic",
+    "IndexPutImplIndexWithNoneModule_basic",
+    "InterpolateDynamicModule_sizes_nearest",
+    "IouOfModule_basic",
+    "MeshgridIndexingIJ_basic",
+    "MeshgridIndexingXY_basic",
+    "Meshgrid_basic",
+    "OneHotModule_basic",
+    # RuntimeError: cannot mutate tensors with frozen storage
+    "ElementwiseRreluTrainModule_basic",
+    "ElementwiseRreluTrainStaticModule_basic",
+    "ElementwiseRreluWithNoiseTrainModule_basic",
+    "ElementwiseRreluWithNoiseTrainStaticModule_basic",
+    "ElementwiseSignbitModule_basic",
+    "ElementwiseCopysignModule_basic",
 }
 
 FX_IMPORTER_CRASHING_SET = LINALG_CRASHING_SET | {
@@ -2845,6 +2891,12 @@ ONNX_XFAIL_SET = {
     "ElementwiseToDtypeI64ToUI8Module_basic",
     "ElementwiseUnaryIntModule_basic",
     "ElementwiseFloatTensorGtIntTensorModule_basic",
+    "ElementwiseSignbitModule_basic",
+    "ElementwiseSignbitIntModule_basic",
+    "ElementwiseFracModule_basic",
+    "ElementwiseCopysignModule_basic",
+    "ElementwiseLdexpModule_basic",
+    "Exp2StaticIntModule_basic",
     "MaskedFillTensorFloatValueModule_basic",
     "MultinomialModule_basic",
     "MultinomialModule2D_basic",

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

@@ -1471,6 +1471,18 @@ def aten〇floor_divide〡shape(self: List[int], other: List[int]) -> List[int]:
 def aten〇atan2〡shape(self: List[int], other: List[int]) -> List[int]:
     return upstream_shape_functions.broadcast(self, other)
 
+def aten〇frac〡shape(self: List[int]) -> List[int]:
+    return self
+
+def aten〇signbit〡shape(self: List[int]) -> List[int]:
+    return self
+
+def aten〇ldexp〇Tensor〡shape(self: List[int], other: List[int]) -> List[int]:
+    return upstream_shape_functions.broadcast(self, other)
+
+def aten〇copysign〇Tensor〡shape(self: List[int], other: List[int]) -> List[int]:
+    return upstream_shape_functions.broadcast(self, other)
+
 def aten〇__and__〇Tensor〡shape(self: List[int], other: List[int]) -> List[int]:
     return upstream_shape_functions.broadcast(self, other)
 
@@ -3746,6 +3758,42 @@ def aten〇rsub〇Scalar〡dtype(self_rank_dtype: Tuple[int, int], other: Union[
     self_rank, self_dtype = self_rank_dtype
     return promote_dtypes([self_rank, None], [self_dtype, get_dtype_of_scalar(other)])
 
+@check_dtype_function(_check_tensors_with_the_same_dtype(num_of_tensors=1, error_types={torch.bool}))
+def aten〇frac〡dtype(self_rank_dtype: Tuple[int, int]) -> int:
+    self_rank, self_dtype = self_rank_dtype
+    assert self_dtype != torch.bool
+    return self_dtype
+
+@check_dtype_function(_check_tensors_with_the_same_dtype(num_of_tensors=1))
+def aten〇signbit〡dtype(self_rank_dtype: Tuple[int, int]) -> int:
+    return torch.bool
+
+@check_dtype_function(_check_two_tensor_op())
+def aten〇ldexp〇Tensor〡dtype(self_rank_dtype: Tuple[int, int], other_rank_dtype: Tuple[int, int]) -> int:
+    self_rank, self_dtype = self_rank_dtype
+    other_rank, other_dtype = other_rank_dtype
+    ranks: List[Optional[int]] = [self_rank, other_rank]
+    dtypes = [self_dtype, other_dtype]
+    if self_dtype == torch.double and is_complex_dtype(other_dtype):
+        return other_dtype
+    elif is_complex_dtype(self_dtype) and other_dtype == torch.double:
+        return self_dtype
+    elif is_integer_dtype(self_dtype) and is_integer_dtype(other_dtype):
+        return torch.float
+    else:
+        return promote_dtypes(ranks, dtypes)
+
+@check_dtype_function(_check_two_tensor_op())
+def aten〇copysign〇Tensor〡dtype(self_rank_dtype: Tuple[int, int], other_rank_dtype: Tuple[int, int]) -> int:
+    self_rank, self_dtype = self_rank_dtype
+    other_rank, other_dtype = other_rank_dtype
+    ranks: List[Optional[int]] = [self_rank, other_rank]
+    dtypes = [self_dtype, other_dtype]
+    if is_integer_dtype(self_dtype) and is_integer_dtype(other_dtype):
+        return torch.float
+    else:
+        return promote_dtypes(ranks, dtypes)
+
 @check_dtype_function(
     _check_tensors_with_the_same_dtype(num_of_tensors=1, other=0.0) +
     _check_tensors_with_the_same_dtype(num_of_tensors=1, other=0))

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

@@ -328,6 +328,7 @@ def emit_ops(emitter_td: TextEmitter, registry: Registry):
         "aten::atanh : (Tensor) -> (Tensor)",
         "aten::atan2 : (Tensor, Tensor) -> (Tensor)",
         "aten::neg : (Tensor) -> (Tensor)",
+        "aten::frac : (Tensor) -> (Tensor)",
         "aten::bitwise_not : (Tensor) -> (Tensor)",
         "aten::div.Tensor : (Tensor, Tensor) -> (Tensor)",
         "aten::logical_or : (Tensor, Tensor) -> (Tensor)",
@@ -369,6 +370,7 @@ def emit_ops(emitter_td: TextEmitter, registry: Registry):
         "aten::zero : (Tensor) -> (Tensor)",
         "aten::fill.Scalar : (Tensor, Scalar) -> (Tensor)",
         "aten::fill.Tensor : (Tensor, Tensor) -> (Tensor)",
+        "aten::copysign.Tensor : (Tensor, Tensor) -> (Tensor)",
     ]:
         emit_with_mutating_variants(key)
     # Shape manipulations:
@@ -417,6 +419,12 @@ def emit_ops(emitter_td: TextEmitter, registry: Registry):
         has_canonicalizer=True,
         has_folder=True,
     )
+    emit(
+        "aten::ldexp.Tensor : (Tensor, Tensor) -> (Tensor)",
+    )
+    emit(
+        "aten::signbit : (Tensor) -> (Tensor)",
+    )
     emit_with_mutating_variants(
         "aten::eq.Tensor : (Tensor, Tensor) -> (Tensor)", has_folder=True
     )

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

@@ -2701,6 +2701,130 @@ def ElementwiseTruncIntModule_basic(module, tu: TestUtils):
 # ==============================================================================
 
 
+class ElementwiseSignbitModule(torch.nn.Module):
+    def __init__(self):
+        super().__init__()
+
+    @export
+    @annotate_args(
+        [
+            None,
+            ([1, 8], torch.float32, True),
+        ]
+    )
+    def forward(self, a):
+        return torch.signbit(a)
+
+
+@register_test_case(module_factory=lambda: ElementwiseSignbitModule())
+def ElementwiseSignbitModule_basic(module, tu: TestUtils):
+    module.forward(
+        torch.tensor(
+            [[-torch.inf, torch.inf, torch.nan, -torch.nan, 2.3, -2.3, 0.0, -0.0]]
+        )
+    )
+
+
+class ElementwiseSignbitIntModule(torch.nn.Module):
+    def __init__(self):
+        super().__init__()
+
+    @export
+    @annotate_args(
+        [
+            None,
+            ([3, 4], torch.int32, True),
+        ]
+    )
+    def forward(self, a):
+        return torch.signbit(a)
+
+
+@register_test_case(module_factory=lambda: ElementwiseSignbitIntModule())
+def ElementwiseSignbitIntModule_basic(module, tu: TestUtils):
+    module.forward(tu.randint(3, 4, low=-100, high=100).to(torch.int32))
+
+
+# ==============================================================================
+
+
+class ElementwiseFracModule(torch.nn.Module):
+    def __init__(self):
+        super().__init__()
+
+    @export
+    @annotate_args(
+        [
+            None,
+            ([1, 6], torch.float32, True),
+        ]
+    )
+    def forward(self, a):
+        return torch.frac(a)
+
+
+@register_test_case(module_factory=lambda: ElementwiseFracModule())
+def ElementwiseFracModule_basic(module, tu: TestUtils):
+    module.forward(torch.tensor([[2.3, -2.3, 0.0, -0.0, 2.0, -2.0]]))
+
+
+# ==============================================================================
+
+
+class ElementwiseCopysignModule(torch.nn.Module):
+    def __init__(self):
+        super().__init__()
+
+    @export
+    @annotate_args(
+        [
+            None,
+            ([1, 1], torch.float32, True),
+            ([1, 6], torch.float32, True),
+        ]
+    )
+    def forward(self, a, b):
+        return torch.copysign(a, b)
+
+
+@register_test_case(module_factory=lambda: ElementwiseCopysignModule())
+def ElementwiseCopysignModule_basic(module, tu: TestUtils):
+    module.forward(
+        torch.tensor([[1.0]]),
+        torch.tensor([[2.3, -2.3, 0.0, -0.0, torch.inf, -torch.inf]]),
+    )
+
+
+# ==============================================================================
+
+
+class ElementwiseLdexpModule(torch.nn.Module):
+    def __init__(self):
+        super().__init__()
+
+    @export
+    @annotate_args(
+        [
+            None,
+            ([1, 6], torch.float32, True),
+            ([1, 1], torch.int64, True),
+        ]
+    )
+    def forward(self, a, b):
+        return torch.ldexp(a, b)
+
+
+@register_test_case(module_factory=lambda: ElementwiseLdexpModule())
+def ElementwiseLdexpModule_basic(module, tu: TestUtils):
+    module.forward(
+        torch.tensor([[2.3, -2.3, 0.0, -0.0, 4.5, -4.5]]),
+        torch.tensor([[2]]),
+    )
+
+
+# ==============================================================================
+
+
 class ElementwiseSignModule(torch.nn.Module):
     def __init__(self):
         super().__init__()
@@ -2792,6 +2916,26 @@ def Exp2StaticModule_basic(module, tu: TestUtils):
     module.forward(tu.rand(3, 2))
 
 
+class Exp2StaticIntModule(torch.nn.Module):
+    def __init__(self):
+        super().__init__()
+
+    @export
+    @annotate_args(
+        [
+            None,
+            ([3, 4], torch.int64, True),
+        ]
+    )
+    def forward(self, x):
+        return torch.ops.aten.exp2(x)
+
+
+@register_test_case(module_factory=lambda: Exp2StaticIntModule())
+def Exp2StaticIntModule_basic(module, tu: TestUtils):
+    module.forward(tu.randint(3, 4, low=-20, high=20))
+
+
 # ==============================================================================