[Auto Parallel] Add spmd rule No.7 for expand_as and expand_as_grad ops.

paddle/phi/api/generator/dist_api_gen.py

@@ -1817,6 +1817,10 @@ def generate_kernel_call_code(self) -> str:
                     kernel_args_type_list.append(
                         self.attrs['attr_info'][arg][0]
                     )
+                    # calculate local_shape for expand_as
+                    # TODO(ooooo): bwd reuse this function to kernel, but actually the local_shape isn't same meaning.
+                    if self.infer_meta['local_shape'] is not None:
+                        arg = "local_shape"
                 input_args.append(arg)
             elif isinstance(arg, bool):
                 input_args.append(str(arg).lower())

paddle/phi/infermeta/spmd_rules/expand_as.cc

@@ -17,6 +17,8 @@ limitations under the License. */
 
 namespace phi::distributed {
 
+using phi::distributed::auto_parallel::str_join;
+
 std::tuple<TensorDistAttr, TensorDistAttr> AlignExpandAsDistAttrs(
     const DistMetaTensor& x, const DistMetaTensor& y) {
   EXTRACT_SHAPE_AND_DIST_ATTR(x);
@@ -42,11 +44,43 @@ std::tuple<TensorDistAttr, TensorDistAttr> AlignExpandAsDistAttrs(
   return {x_dist_attr_dst, y_dist_attr_dst};
 }
 
+std::tuple<TensorDistAttr, TensorDistAttr> AlignExpandAsDistAttrsWithShape(
+    const DistMetaTensor& x, const std::vector<int64_t>& target_shape) {
+  EXTRACT_SHAPE_AND_DIST_ATTR(x);
+  std::vector<int64_t> out_dims_mapping(target_shape.size());
+  int dims_diff = target_shape.size() - x_ndim;
+  for (int i = target_shape.size() - 1; i >= dims_diff; --i) {
+    if (target_shape[i] != x_shape[i - dims_diff]) {
+      out_dims_mapping[i] = -1;
+    } else {
+      out_dims_mapping[i] = x_dims_mapping_src[i - dims_diff];
+    }
+  }
+  for (int i = 0; i < dims_diff; i++) {
+    out_dims_mapping[i] = -1;
+  }
+  TensorDistAttr out_dist_attr_dst =
+      CopyTensorDistAttrForOutput(x_dist_attr_src);
+  out_dist_attr_dst.set_dims_mapping(out_dims_mapping);
+  VLOG(4) << "x";
+  VLOG(4) << "shape: [" << str_join(x_shape) << "] "
+          << "src_dist_attr: [" << x_dist_attr_src.to_string() << "] "
+          << "dst_dist_attr: [" << x_dist_attr_src.to_string() << "]";
+  LOG_SPMD_OUTPUT(out_dist_attr_dst);
+  return {{x_dist_attr_src}, {out_dist_attr_dst}};
+}
+
 SpmdInfo ExpandAsInferSpmd(const DistMetaTensor& x,
                            const DistMetaTensor& y,
                            const std::vector<int64_t>& target_shape) {
-  auto [x_dist_attr, y_dist_attr] = AlignExpandAsDistAttrs(x, y);
-  return {{x_dist_attr, y_dist_attr}, {y_dist_attr}};
+  if (y.initialized()) {
+    auto [x_dist_attr, y_dist_attr] = AlignExpandAsDistAttrs(x, y);
+    return {{x_dist_attr, y_dist_attr}, {y_dist_attr}};
+  } else {
+    auto [x_dist_attr, out_dist_attr] =
+        AlignExpandAsDistAttrsWithShape(x, target_shape);
+    return {{x_dist_attr, TensorDistAttr()}, {out_dist_attr}};
+  }
 }
 
 SpmdInfo ExpandAsInferSpmdReverse(const DistMetaTensor& x,

paddle/phi/infermeta/spmd_rules/rules.cc

@@ -667,15 +667,13 @@ PD_REGISTER_SPMD_RULE(
     PD_INFER_SPMD(
         phi::distributed::FusedLinearParamGradAddInferSpmdFakeReverse));
 
-PD_REGISTER_SPMD_RULE(
-    expand_as,
-    PD_INFER_SPMD(phi::distributed::ExpandAsInferSpmd),
-    PD_INFER_SPMD(phi::distributed::ExpandAsInferSpmdReverse));
+PD_REGISTER_SPMD_RULE(expand_as,
+                      PD_INFER_SPMD(phi::distributed::ExpandAsInferSpmd),
+                      PD_INFER_SPMD(phi::distributed::ExpandAsGradInferSpmd));
 
-PD_REGISTER_SPMD_RULE(
-    expand_as_v2,
-    PD_INFER_SPMD(phi::distributed::ExpandAsInferSpmd),
-    PD_INFER_SPMD(phi::distributed::ExpandAsInferSpmdReverse));
+PD_REGISTER_SPMD_RULE(expand_as_v2,
+                      PD_INFER_SPMD(phi::distributed::ExpandAsInferSpmd),
+                      PD_INFER_SPMD(phi::distributed::ExpandAsGradInferSpmd));
 
 // scatter
 PD_REGISTER_SPMD_RULE(scatter,

paddle/phi/ops/yaml/backward.yaml

@@ -1007,6 +1007,8 @@
   infer_meta :
     func : UnchangedInferMeta
     param : [x]
+    spmd_rule : ExpandAsGradInferSpmd
+    local_shape : x_grad
   kernel :
     func : expand_as_grad
   no_need_buffer : x

paddle/phi/ops/yaml/ops.yaml

@@ -1808,6 +1808,7 @@
   infer_meta :
     func : ExpandAsInferMeta
     local_shape: out
+    spmd_rule : ExpandAsInferSpmd
   kernel :
     func : expand_as
     data_type : x

test/auto_parallel/spmd_rules/CMakeLists.txt

@@ -44,6 +44,7 @@ if(WITH_DISTRIBUTE)
   py_test_modules(test_logsumexp_rule MODULES test_logsumexp_rule)
   py_test_modules(test_nonzero_rule MODULES test_nonzero_rule)
   if(NOT WITH_ROCM)
+    py_test_modules(test_expand_as_rule MODULES test_expand_as_rule)
     py_test_modules(test_roll_rule MODULES test_roll_rule)
     py_test_modules(test_unique_rule MODULES test_unique_rule)
     py_test_modules(test_topk_rule MODULES test_topk_rule)

test/auto_parallel/spmd_rules/test_expand_as_rule.py

@@ -0,0 +1,270 @@
+# Copyright (c) 2025 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import unittest
+from collections import OrderedDict
+
+from paddle.distributed.auto_parallel.static.dist_attribute import (
+    DistTensorSpec,
+    TensorDistAttr,
+)
+from paddle.distributed.fleet import auto
+from paddle.framework import core
+
+
+class TestExpandAsSPMDRule(unittest.TestCase):
+    def setUp(self):
+        x_shape = [1, 48]
+        y_shape = [2, 32, 48]
+        out_shape = [2, 32, 48]
+        process_mesh = auto.ProcessMesh(mesh=[[0, 1], [2, 3]])
+
+        x_tensor_dist_attr = TensorDistAttr()
+        x_tensor_dist_attr.dims_mapping = [-1, -1]
+        x_tensor_dist_attr.process_mesh = process_mesh
+        self.x_dist_tensor_spec = DistTensorSpec(x_shape, x_tensor_dist_attr)
+        y_tensor_dist_attr = TensorDistAttr()
+        y_tensor_dist_attr.dims_mapping = [0, 1, -1]
+        y_tensor_dist_attr.process_mesh = process_mesh
+        self.y_dist_tensor_spec = DistTensorSpec(y_shape, y_tensor_dist_attr)
+        out_tensor_dist_attr = TensorDistAttr()
+        out_tensor_dist_attr.dims_mapping = [0, 1, -1]
+        out_tensor_dist_attr.process_mesh = process_mesh
+        self.out_dist_tensor_spec = DistTensorSpec(
+            out_shape, out_tensor_dist_attr
+        )
+
+        self.rule = core.get_phi_spmd_rule("expand_as")
+        self.attrs = OrderedDict()
+        self.attrs['target_shape'] = [2, 32, 48]
+
+    def test_expand_as_forward_with_y(self):
+        # [1, 48], [2, 32, 48]
+        # [-1, -1], [0, 1, -1] --> [-1, -1], [0, 1, -1],[0, 1, -1]
+        result_dist_attrs = self.rule.infer_forward(
+            self.x_dist_tensor_spec,
+            self.y_dist_tensor_spec,
+            self.attrs['target_shape'],
+        )
+
+        self.assertEqual(len(result_dist_attrs), 2)
+        inferred_input_dist_attrs = result_dist_attrs[0]
+        inferred_output_dist_attrs = result_dist_attrs[1]
+        self.assertEqual(len(inferred_input_dist_attrs), 2)
+        self.assertEqual(len(inferred_output_dist_attrs), 1)
+        self.assertEqual(inferred_input_dist_attrs[0].dims_mapping, [-1, -1])
+        self.assertEqual(inferred_input_dist_attrs[1].dims_mapping, [0, 1, -1])
+        self.assertEqual(inferred_output_dist_attrs[0].dims_mapping, [0, 1, -1])
+
+        # [1, 48], [2, 32, 48]
+        # [-1, -1], [0, -1, 1] --> [-1, 1], [0, -1, 1], [0, -1, 1]
+        self.y_dist_tensor_spec.set_dims_mapping([0, -1, 1])
+        result_dist_attrs = self.rule.infer_forward(
+            self.x_dist_tensor_spec,
+            self.y_dist_tensor_spec,
+            self.attrs['target_shape'],
+        )
+
+        self.assertEqual(len(result_dist_attrs), 2)
+        inferred_input_dist_attrs = result_dist_attrs[0]
+        inferred_output_dist_attrs = result_dist_attrs[1]
+        self.assertEqual(len(inferred_input_dist_attrs), 2)
+        self.assertEqual(len(inferred_output_dist_attrs), 1)
+        self.assertEqual(inferred_input_dist_attrs[0].dims_mapping, [-1, 1])
+        self.assertEqual(inferred_input_dist_attrs[1].dims_mapping, [0, -1, 1])
+        self.assertEqual(inferred_output_dist_attrs[0].dims_mapping, [0, -1, 1])
+
+        # [2, 1, 48], [2, 32, 48]
+        # [0, -1, -1], [0, 1, -1] --> [0, -1, -1], [0, 1, -1], [0, 1, -1]
+        self.x_dist_tensor_spec.shape = [2, 1, 48]
+        self.x_dist_tensor_spec.set_dims_mapping([0, -1, -1])
+        self.y_dist_tensor_spec.set_dims_mapping([0, 1, -1])
+        result_dist_attrs = self.rule.infer_forward(
+            self.x_dist_tensor_spec,
+            self.y_dist_tensor_spec,
+            self.attrs['target_shape'],
+        )
+        self.assertEqual(len(result_dist_attrs), 2)
+        inferred_input_dist_attrs = result_dist_attrs[0]
+        inferred_output_dist_attrs = result_dist_attrs[1]
+        self.assertEqual(len(inferred_input_dist_attrs), 2)
+        self.assertEqual(len(inferred_output_dist_attrs), 1)
+        self.assertEqual(inferred_input_dist_attrs[0].dims_mapping, [0, -1, -1])
+        self.assertEqual(inferred_input_dist_attrs[1].dims_mapping, [0, 1, -1])
+        self.assertEqual(inferred_output_dist_attrs[0].dims_mapping, [0, 1, -1])
+
+        # [2, 1, 48], [2, 32, 48]
+        # [-1, -1, 0], [0, 1, -1] --> [0, -1, -1], [0, 1, -1], [0, 1, -1]
+        self.x_dist_tensor_spec.set_dims_mapping([-1, -1, 0])
+        self.y_dist_tensor_spec.set_dims_mapping([0, 1, -1])
+        result_dist_attrs = self.rule.infer_forward(
+            self.x_dist_tensor_spec,
+            self.y_dist_tensor_spec,
+            self.attrs['target_shape'],
+        )
+        self.assertEqual(len(result_dist_attrs), 2)
+        inferred_input_dist_attrs = result_dist_attrs[0]
+        inferred_output_dist_attrs = result_dist_attrs[1]
+        self.assertEqual(len(inferred_input_dist_attrs), 2)
+        self.assertEqual(len(inferred_output_dist_attrs), 1)
+        self.assertEqual(inferred_input_dist_attrs[0].dims_mapping, [0, -1, -1])
+        self.assertEqual(inferred_input_dist_attrs[1].dims_mapping, [0, 1, -1])
+        self.assertEqual(inferred_output_dist_attrs[0].dims_mapping, [0, 1, -1])
+
+    def test_expand_as_forward_without_y(self):
+        # [1, 48], target =[2, 32, 48]
+        # [-1, -1] --> [-1, -1], [-1, -1, -1]
+        self.attrs['target_shape'] = [2, 32, 48]
+        self.x_dist_tensor_spec.shape = [1, 48]
+        self.x_dist_tensor_spec.set_dims_mapping([-1, -1])
+        result_dist_attrs = self.rule.infer_forward(
+            self.x_dist_tensor_spec,
+            DistTensorSpec(),
+            self.attrs['target_shape'],
+        )
+        self.assertEqual(len(result_dist_attrs), 2)
+        inferred_input_dist_attrs = result_dist_attrs[0]
+        inferred_output_dist_attrs = result_dist_attrs[1]
+        self.assertEqual(len(inferred_input_dist_attrs), 2)
+        self.assertEqual(len(inferred_output_dist_attrs), 1)
+        self.assertEqual(inferred_input_dist_attrs[0].dims_mapping, [-1, -1])
+        self.assertEqual(inferred_input_dist_attrs[1], TensorDistAttr())
+        self.assertEqual(
+            inferred_output_dist_attrs[0].dims_mapping, [-1, -1, -1]
+        )
+
+        # [1, 48], target =[2, 32, 48]
+        # [-1, 0] --> [-1, 0], [-1, -1, 0]
+        self.x_dist_tensor_spec.set_dims_mapping([-1, 0])
+        result_dist_attrs = self.rule.infer_forward(
+            self.x_dist_tensor_spec,
+            DistTensorSpec(),
+            self.attrs['target_shape'],
+        )
+        self.assertEqual(len(result_dist_attrs), 2)
+        inferred_input_dist_attrs = result_dist_attrs[0]
+        inferred_output_dist_attrs = result_dist_attrs[1]
+        self.assertEqual(len(inferred_input_dist_attrs), 2)
+        self.assertEqual(len(inferred_output_dist_attrs), 1)
+        self.assertEqual(inferred_input_dist_attrs[0].dims_mapping, [-1, 0])
+        self.assertEqual(inferred_input_dist_attrs[1], TensorDistAttr())
+        self.assertEqual(
+            inferred_output_dist_attrs[0].dims_mapping, [-1, -1, 0]
+        )
+
+        # [2, 1, 48], target =[2, 32, 48]
+        # [0, -1, -1] --> [0, -1, -1], [0, -1, -1]
+        self.x_dist_tensor_spec.shape = [2, 1, 48]
+        self.x_dist_tensor_spec.set_dims_mapping([0, -1, -1])
+        result_dist_attrs = self.rule.infer_forward(
+            self.x_dist_tensor_spec,
+            DistTensorSpec(),
+            self.attrs['target_shape'],
+        )
+        self.assertEqual(len(result_dist_attrs), 2)
+        inferred_input_dist_attrs = result_dist_attrs[0]
+        inferred_output_dist_attrs = result_dist_attrs[1]
+        self.assertEqual(len(inferred_input_dist_attrs), 2)
+        self.assertEqual(len(inferred_output_dist_attrs), 1)
+        self.assertEqual(inferred_input_dist_attrs[0].dims_mapping, [0, -1, -1])
+        self.assertEqual(inferred_input_dist_attrs[1], TensorDistAttr())
+        self.assertEqual(
+            inferred_output_dist_attrs[0].dims_mapping, [0, -1, -1]
+        )
+
+        # [2, 32, 48], target =[2, 32, 48]
+        # [0, 1, -1] --> [0, 1, -1], [0, 1, -1]
+        self.x_dist_tensor_spec.shape = [2, 32, 48]
+        self.x_dist_tensor_spec.set_dims_mapping([0, 1, -1])
+        result_dist_attrs = self.rule.infer_forward(
+            self.x_dist_tensor_spec,
+            DistTensorSpec(),
+            self.attrs['target_shape'],
+        )
+        self.assertEqual(len(result_dist_attrs), 2)
+        inferred_input_dist_attrs = result_dist_attrs[0]
+        inferred_output_dist_attrs = result_dist_attrs[1]
+        self.assertEqual(len(inferred_input_dist_attrs), 2)
+        self.assertEqual(len(inferred_output_dist_attrs), 1)
+        self.assertEqual(inferred_input_dist_attrs[0].dims_mapping, [0, 1, -1])
+        self.assertEqual(inferred_input_dist_attrs[1], TensorDistAttr())
+        self.assertEqual(inferred_output_dist_attrs[0].dims_mapping, [0, 1, -1])
+
+    def test_expand_as_backward(self):
+        # [1, 48], [2, 32, 48]
+        # [-1, -1], [0, 1, -1] --> [-1, -1], [0, 1, -1], [-1, -1](partial {0, 1})
+        self.attrs['target_shape'] = [2, 32, 48]
+        self.x_dist_tensor_spec.shape = [1, 48]
+        self.x_dist_tensor_spec.set_dims_mapping([-1, -1])
+        self.out_dist_tensor_spec.shape = [2, 32, 48]
+        self.out_dist_tensor_spec.set_dims_mapping([0, 1, -1])
+        result_dist_attrs = self.rule.infer_backward(
+            self.x_dist_tensor_spec,
+            self.out_dist_tensor_spec,
+            self.attrs['target_shape'],
+        )
+        self.assertEqual(len(result_dist_attrs), 2)
+        inferred_input_dist_attrs = result_dist_attrs[0]
+        inferred_output_dist_attrs = result_dist_attrs[1]
+        self.assertEqual(len(inferred_input_dist_attrs), 2)
+        self.assertEqual(len(inferred_output_dist_attrs), 1)
+        self.assertEqual(inferred_input_dist_attrs[0].dims_mapping, [-1, -1])
+        self.assertEqual(inferred_input_dist_attrs[1].dims_mapping, [0, 1, -1])
+        self.assertEqual(inferred_output_dist_attrs[0].dims_mapping, [-1, -1])
+        self.assertEqual(inferred_output_dist_attrs[0]._is_partial(), True)
+        self.assertEqual(inferred_output_dist_attrs[0]._partial_dims(), {0, 1})
+
+        # [1, 48], [2, 32, 48]
+        # [-1, 1], [0, -1, 1] --> [-1, 1], [0, -1, 1], [-1, 1](partial {0})
+        self.x_dist_tensor_spec.set_dims_mapping([-1, 1])
+        self.out_dist_tensor_spec.set_dims_mapping([0, -1, 1])
+        result_dist_attrs = self.rule.infer_backward(
+            self.x_dist_tensor_spec,
+            self.out_dist_tensor_spec,
+            self.attrs['target_shape'],
+        )
+        self.assertEqual(len(result_dist_attrs), 2)
+        inferred_input_dist_attrs = result_dist_attrs[0]
+        inferred_output_dist_attrs = result_dist_attrs[1]
+        self.assertEqual(len(inferred_input_dist_attrs), 2)
+        self.assertEqual(len(inferred_output_dist_attrs), 1)
+        self.assertEqual(inferred_input_dist_attrs[0].dims_mapping, [-1, 1])
+        self.assertEqual(inferred_input_dist_attrs[1].dims_mapping, [0, -1, 1])
+        self.assertEqual(inferred_output_dist_attrs[0].dims_mapping, [-1, 1])
+        self.assertEqual(inferred_output_dist_attrs[0]._is_partial(), True)
+        self.assertEqual(inferred_output_dist_attrs[0]._partial_dims(), {0})
+
+        # [1, 48], [2, 32, 48]
+        # [-1, -1], [0, -1, 1] --> [-1, 1], [0, -1, 1], [-1, 1](partial {0})
+        self.x_dist_tensor_spec.set_dims_mapping([-1, -1])
+        self.out_dist_tensor_spec.set_dims_mapping([0, -1, 1])
+        result_dist_attrs = self.rule.infer_backward(
+            self.x_dist_tensor_spec,
+            self.out_dist_tensor_spec,
+            self.attrs['target_shape'],
+        )
+        self.assertEqual(len(result_dist_attrs), 2)
+        inferred_input_dist_attrs = result_dist_attrs[0]
+        inferred_output_dist_attrs = result_dist_attrs[1]
+        self.assertEqual(len(inferred_input_dist_attrs), 2)
+        self.assertEqual(len(inferred_output_dist_attrs), 1)
+        self.assertEqual(inferred_input_dist_attrs[0].dims_mapping, [-1, 1])
+        self.assertEqual(inferred_input_dist_attrs[1].dims_mapping, [0, -1, 1])
+        self.assertEqual(inferred_output_dist_attrs[0].dims_mapping, [-1, 1])
+        self.assertEqual(inferred_output_dist_attrs[0]._is_partial(), True)
+        self.assertEqual(inferred_output_dist_attrs[0]._partial_dims(), {0})
+
+
+if __name__ == "__main__":
+    unittest.main()