[frontend] Add verbose mode for torch dynamo compiler.

frontend/Python/frontend.py

@@ -62,6 +62,7 @@ def __init__(
         func_name: str = "forward",
         primary_registry: Optional[dict] = None,
         aot_autograd_decomposition: Optional[dict] = None,
+        verbose=False,
     ) -> None:
         """
         Initializes the Dynamo Compiler.
@@ -71,21 +72,29 @@ def __init__(
             primary_registry (dict, optional): The primary operations registry.
             aot_autograd_decomposition (Optional[dict], optional):
             The ahead-of-time autograd decomposition dictionary.
+            verbose (bool): Controls whether to print additional information for
+                debugging purposes. The default value is False, indicating that
+                no extra debug information will be printed.
         Attributes:
             _func_name: The function name to be used.
             _aot_autograd_decomposition (Optional[dict], optional):
             The ahead-of-time autograd decomposition dictionary.
+            _verbose: The option for the verbosity option of output.
             _imported_graphs: The buddy graphs from dynamo importer.
             _ops_registry (dict, optional): The buddy operations' lower func
             registry.
             _imported_params: The model params extract from torch.
             _ops_map: The torch aten ops map with buddy ops.
 
         """
+        # Make custom dynamo compiler take effect.
+        dynamo.reset()
+        # Initialize the attributes.
         if primary_registry is None:
             primary_registry = {}
         self._func_name = func_name
         self._aot_autograd_decomposition = aot_autograd_decomposition
+        self._verbose = verbose
         self._imported_graphs = []
         self._ops_registry = {}
         self._imported_params = {}
@@ -243,6 +252,10 @@ def _compile_fx(
         }
         params_flat, _ = pytree.tree_flatten(params)
 
+        if self._verbose:
+            print("Graph in tabular form:")
+            gm.graph.print_tabular()
+
         def _compiler(_gm: torch.fx.GraphModule, _inputs: List[torch.Tensor]):
             """Compile a FX graph in Aten/Prims IR to MLIR."""
             nonlocal params_flat

requirements.txt

@@ -7,3 +7,4 @@ accelerate
 protobuf
 pybind11 == 2.11.1
 torchvision
+tabulate

tests/Python/test_verbose_mode.py

@@ -0,0 +1,39 @@
+# RUN: %PYTHON %s 2>&1 | FileCheck %s
+import torch
+
+from buddy.compiler.frontend import DynamoCompiler
+
+
+# Define the target function or model.
+def foo(x, y):
+    return x * y + x
+
+
+# Define the input data.
+float32_in1 = torch.randn(10).to(torch.float32)
+float32_in2 = torch.randn(10).to(torch.float32)
+
+# Test the default dynamo compiler importer mode.
+dynamo_compiler_default = DynamoCompiler()
+graphs = dynamo_compiler_default.importer(foo, *(float32_in1, float32_in2))
+
+# Ensure no output is printed in the default mode.
+# CHECK-NOT: .
+
+# Test the dynamo compiler verbose mode.
+dynamo_compiler_verbose_on = DynamoCompiler(verbose=True)
+graphs = dynamo_compiler_verbose_on.importer(foo, *(float32_in1, float32_in2))
+
+# Test output in the verbose mode.
+# CHECK: placeholder
+# CHECK: placeholder
+# CHECK: call_function
+# CHECK: call_function
+# CHECK: output
+
+# Test the dynamo compiler verbose mode off.
+dynamo_compiler_verbose_off = DynamoCompiler(verbose=False)
+graphs = dynamo_compiler_verbose_off.importer(foo, *(float32_in1, float32_in2))
+
+# Ensure no output is printed when the verbose mode is off.
+# CHECK-NOT: .