Revert to old behaviour of not padding strides if shape or stride is dynamic (pytorch#163639)

Differential Revision: D83053287

Pull Request resolved: pytorch#163639
Approved by: https://github.com/blaine-rister

Author: nandesuka

torch/_inductor/ir.py

@@ -3741,6 +3741,15 @@ def _pad_strides(
         ):
             return in_strides
 
+        # Skip padding the strides for dynamic shapes based on config.pad_dynamic_shape
+        # Checking both shape and strides, as there are cases where only one is dynamic
+        is_dynamic = not all(
+            isinstance(s, (int, sympy.Integer))
+            for s in itertools.chain(in_strides, size)
+        )
+        if not config.pad_dynamic_shapes and is_dynamic:
+            return in_strides
+
         shape_env = V.graph._shape_env if hasattr(V.graph, "_shape_env") else None
 
         def contains_unbacked_symints(expr: sympy.Expr | int) -> bool: