#9681: set __name__ attribute for ttnn operations when fast runtime m…

…ode is disabled

ttnn/ttnn/decorators.py

@@ -380,6 +380,10 @@ class Operation:
     is_cpp_function: bool
     is_experimental: bool
 
+    @property
+    def __name__(self):
+        return self.python_fully_qualified_name
+
     def __gt__(self, other):
         return self.python_fully_qualified_name < other.python_fully_qualified_name
 
@@ -748,6 +752,18 @@ def operation_decorator(function: callable):
         global OPERATION_TO_GOLDEN_FUNCTION
         global OPERATION_TO_FALLBACK_FUNCTION
 
+        is_cpp_function = hasattr(function, "__ttnn__")
+
+        python_fully_qualified_name = name
+        if is_cpp_function:
+            if doc is not None:
+                raise RuntimeError(f"Registering {name}: documentation for C++ function has to be set from C++")
+            if python_fully_qualified_name is not None:
+                raise RuntimeError(f"Registering {name}: name is not allowed for ttnn functions")
+            python_fully_qualified_name = function.python_fully_qualified_name  # Replace C++ name with python
+        elif not is_experimental:
+            logger.warning(f"{name} should be migrated to C++!")
+
         def fallback_function(*function_args, **function_kwargs):
             preprocess_inputs = preprocess_golden_function_inputs or default_preprocess_golden_function_inputs
             postprocess_outputs = postprocess_golden_function_outputs or default_postprocess_golden_function_outputs
@@ -759,19 +775,21 @@ def fallback_function(*function_args, **function_kwargs):
             return output
 
         if ttnn.CONFIG.enable_fast_runtime_mode:
+
+            def name_decorator(function):
+                @wraps(function)
+                def wrapper(*args, **kwargs):
+                    return function(*args, **kwargs)
+
+                return wrapper
+
+            function = name_decorator(function)
+            function.__name__ = python_fully_qualified_name
+
             OPERATION_TO_GOLDEN_FUNCTION[function] = golden_function
             OPERATION_TO_FALLBACK_FUNCTION[function] = fallback_function
-            return function
-
-        is_cpp_function = hasattr(function, "__ttnn__")
 
-        python_fully_qualified_name = name
-        if is_cpp_function:
-            if doc is not None:
-                raise RuntimeError(f"Registering {name}: documentation for C++ functiomn has to be set from C++")
-            if python_fully_qualified_name is not None:
-                raise RuntimeError(f"Registering {name}: name is not allowed for ttnn functions")
-            python_fully_qualified_name = function.python_fully_qualified_name  # Replace C++ name with python
+            return function
 
         # Wrap functions before attaching documentation to avoid errors
         if doc is not None:

ttnn/ttnn/operations/binary.py

@@ -49,7 +49,6 @@ def binary_function(
         return output_tensor
 
     if isinstance(binary_function, ttnn.decorators.Operation):
-        binary_function.__name__ = f"ttnn.{name}"
         binary_function.decorated_function.__doc__ = doc + (
             binary_function.__doc__ if binary_function.__doc__ is not None else ""
         )
@@ -348,7 +347,6 @@ def elt_binary_function(
         return output_tensor
 
     if isinstance(elt_binary_function, ttnn.decorators.Operation):
-        elt_binary_function.__name__ = f"ttnn.{name}"
         elt_binary_function.decorated_function.__doc__ = f"""{name}(input_tensor_a: ttnn.Tensor, input_tensor_b: ttnn.Tensor, *, memory_config: ttnn.MemoryConfig = ttnn.DRAM_MEMORY_CONFIG) -> ttnn.Tensor
 
             Performs eltwise-binary {op_name} operation on two tensors :attr:`input_a` and :attr:`input_b`.

ttnn/ttnn/operations/losses.py

@@ -79,7 +79,6 @@ def loss_function(
         return output_tensor
 
     if isinstance(loss_function, ttnn.decorators.Operation):
-        loss_function.__name__ = f"ttnn.{name}"
         loss_function.decorated_function.__doc__ = f"""{name}(input_tensor_a: ttnn.Tensor, input_tensor_b: ttnn.Tensor, loss_mode: str, *, memory_config: ttnn.MemoryConfig = ttnn.DRAM_MEMORY_CONFIG) -> ttnn.Tensor
 
             Applies {name} to :attr:`input_tensor_a` and :attr:`input_tensor_b` with loss_mode :attr:`loss_mode`.

ttnn/ttnn/operations/ternary.py

@@ -99,7 +99,6 @@ def ternary_function(
         return output_tensor
 
     if isinstance(ternary_function, ttnn.decorators.Operation):
-        ternary_function.__name__ = f"ttnn.{name}"
         ternary_function.decorated_function.__doc__ = f"""{name}(input_tensor: ttnn.Tensor, input_tensor1: ttnn.Tensor, input_tensor2: ttnn.Tensor, *, memory_config: ttnn.MemoryConfig = ttnn.DRAM_MEMORY_CONFIG) -> ttnn.Tensor
 
             Returns tensor with the {name} of all of elements of the input tensors input, tensor1, tensor2.
@@ -221,7 +220,6 @@ def ternary_function(
         return output_tensor
 
     if isinstance(ternary_function, ttnn.decorators.Operation):
-        ternary_function.__name__ = f"ttnn.{(name)}"
         ternary_function.decorated_function.__doc__ = f"""{(name)}(input_tensor: ttnn.Tensor, input_tensor1: ttnn.Tensor, input_tensor2: ttnn.Tensor, parameter, *, memory_config: ttnn.MemoryConfig = ttnn.DRAM_MEMORY_CONFIG) -> ttnn.Tensor
 
             Performs the element-wise {op_name} of tensor1 by tensor2, multiplies the result by the scalar value and adds it to input input.

ttnn/ttnn/operations/unary.py

@@ -260,7 +260,6 @@ def unary_function(
         return output_tensor
 
     if isinstance(unary_function, ttnn.decorators.Operation):
-        unary_function.__name__ = f"ttnn.{(name)}"
         unary_function.decorated_function.__doc__ = f"""{(name)}(input_tensor: ttnn.Tensor, parameter, *, memory_config: ttnn.MemoryConfig = ttnn.DRAM_MEMORY_CONFIG) -> ttnn.Tensor
 
             Applies the {name} function to the elements of the input tensor :attr:`input_tensor` with :attr:`{param}` parameter.
@@ -346,7 +345,6 @@ def activation_function(
         return output_tensor
 
     if isinstance(activation_function, ttnn.decorators.Operation):
-        activation_function.__name__ = f"ttnn.{(name)}"
         activation_function.decorated_function.__doc__ = f"""{(name)}(input_tensor: ttnn.Tensor, parameter, *, memory_config: ttnn.MemoryConfig = ttnn.DRAM_MEMORY_CONFIG) -> ttnn.Tensor
 
             Applies the {name} function to the elements of the input tensor :attr:`input_tensor` with :attr:`{param}` parameter.
@@ -430,7 +428,6 @@ def activation_function(
         return output_tensor
 
     if isinstance(activation_function, ttnn.decorators.Operation):
-        activation_function.__name__ = f"ttnn.{(name)}"
         activation_function.decorated_function.__doc__ = f"""{(name)}(input_tensor: ttnn.Tensor, parameter, *, memory_config: ttnn.MemoryConfig = ttnn.DRAM_MEMORY_CONFIG) -> ttnn.Tensor
 
             Applies the {name} function to the elements of the input tensor :attr:`input_tensor` with :attr:`{param1_name}` and :attr:`{param2_name}`  parameters.
@@ -550,7 +547,6 @@ def activation_function(
         return output_tensor
 
     if isinstance(activation_function, ttnn.decorators.Operation):
-        activation_function.__name__ = f"ttnn.{(name)}"
         activation_function.decorated_function.__doc__ = f"""{(name)}(input_tensor: ttnn.Tensor, dim: int = -1, *, memory_config: ttnn.MemoryConfig = ttnn.DRAM_MEMORY_CONFIG) -> ttnn.Tensor
 
             Applies the {name} function to the elements of the input tensor :attr:`input_tensor` split along :attr:`{param}`.