Don't do symbolic upcasting in local_upcast_elemwise_constants

This reduces the number of rewrite passes, by avoiding constant fold of cast/expand_dims/alloc

Author: ricardoV94

pytensor/tensor/rewriting/elemwise.py

@@ -30,21 +30,16 @@
 from pytensor.scalar.math import Grad2F1Loop, _grad_2f1_loop
 from pytensor.tensor.basic import (
     MakeVector,
-    alloc,
-    cast,
     constant,
-    get_underlying_scalar_constant_value,
 )
 from pytensor.tensor.elemwise import CAReduce, DimShuffle, Elemwise
-from pytensor.tensor.exceptions import NotScalarConstantError
 from pytensor.tensor.math import add, exp, mul
 from pytensor.tensor.rewriting.basic import (
     alloc_like,
     broadcasted_by,
     register_canonicalize,
     register_specialize,
 )
-from pytensor.tensor.shape import shape_padleft
 from pytensor.tensor.variable import TensorConstant, TensorVariable
 
 
@@ -434,66 +429,40 @@ def local_upcast_elemwise_constant_inputs(fgraph, node):
 
     """
     if len(node.outputs) > 1:
-        return
-    try:
-        shape_i = fgraph.shape_feature.shape_i
-    except AttributeError:
-        shape_i = None
-    if isinstance(node.op, Elemwise):
-        scalar_op = node.op.scalar_op
-        # print "aa", scalar_op.output_types_preference
-        if getattr(scalar_op, "output_types_preference", None) in (
-            ps.upgrade_to_float,
-            ps.upcast_out,
-        ):
-            # this is the kind of op that we can screw with the input
-            # dtypes by upcasting explicitly
-            output_dtype = node.outputs[0].type.dtype
-            new_inputs = []
-            for i in node.inputs:
-                if i.type.dtype == output_dtype:
-                    new_inputs.append(i)
-                else:
-                    try:
-                        cval_i = get_underlying_scalar_constant_value(
-                            i, only_process_constants=True
-                        )
-                        if all(i.broadcastable):
-                            new_inputs.append(
-                                shape_padleft(cast(cval_i, output_dtype), i.ndim)
-                            )
-                        else:
-                            if shape_i is None:
-                                return
-                            new_inputs.append(
-                                alloc(
-                                    cast(cval_i, output_dtype),
-                                    *[shape_i(d)(i) for d in range(i.ndim)],
-                                )
-                            )
-                            # print >> sys.stderr, "AAA",
-                            # *[Shape_i(d)(i) for d in range(i.ndim)]
-                    except NotScalarConstantError:
-                        # for the case of a non-scalar
-                        if isinstance(i, TensorConstant):
-                            new_inputs.append(cast(i, output_dtype))
-                        else:
-                            new_inputs.append(i)
+        return None
+
+    if getattr(node.op.scalar_op, "output_types_preference", None) not in (
+        ps.upgrade_to_float,
+        ps.upcast_out,
+    ):
+        return None
 
-            if new_inputs != node.inputs:
-                rval = [node.op(*new_inputs)]
-                if not node.outputs[0].type.is_super(rval[0].type):
-                    # This can happen for example when floatX=float32
-                    # and we do the true division between and int64
-                    # and a constant that will get typed as int8.
+    # this is the kind of op that we can screw with the input
+    # dtypes by upcasting explicitly
+    [old_out] = node.outputs
+    output_dtype = old_out.type.dtype
+    new_inputs = list(node.inputs)
+    changed = False
+    for i, inp in enumerate(node.inputs):
+        if inp.type.dtype != output_dtype and isinstance(inp, TensorConstant):
+            new_inputs[i] = constant(inp.data.astype(output_dtype))
+            changed = True
+
+    if not changed:
+        return None
 
-                    # As this is just to allow merging more case, if
-                    # the upcast don't work, we can just skip it.
-                    return
+    rval = node.op(*new_inputs)
+    if not old_out.type.is_super(rval.type):
+        # This can happen for example when floatX=float32
+        # and we do the true division between and int64
+        # and a constant that will get typed as int8.
+        # As this is just to allow merging more case, if
+        # the upcast don't work, we can just skip it.
+        return None
 
-                # Copy over output stacktrace from before upcasting
-                copy_stack_trace(node.outputs[0], rval)
-                return rval
+    # Copy over output stacktrace from before upcasting
+    copy_stack_trace(old_out, rval)
+    return [rval]
 
 
 @node_rewriter([add, mul])