refactor tile op and in expander open on gpu

mindspore/_extends/graph_kernel/expanders/tile.py

@@ -13,7 +13,7 @@
 # limitations under the License.
 # ===========================================================================
 """generate json desc for Tile"""
-from mindspore._extends.graph_kernel.model.op_infer import Tile as TileInfer
+from mindspore._extends.graph_kernel.model.model import GraphKernelUnsupportedException as GKException
 from mindspore._extends.graph_kernel.model.model import DataFormat as DF
 from ._utils import Expander, ExpanderInfoValidator as VLD
 
@@ -23,14 +23,32 @@ from ._utils import Expander, ExpanderInfoValidator as VLD
 class Tile(Expander):
     """Tile expander"""
 
+    def _get_output_shape(self):
+        """Get output shape"""
+        shape = self.inputs[0].shape
+        multiples = self.attrs["multiples"]
+
+        shape = list(shape)
+        multiples = list(multiples)
+        diff_len = len(multiples) - len(shape)
+        if diff_len < 0:
+            raise GKException("Dimensions of multiples{} < dimensions of input{} in Tile".format(multiples, shape))
+        if diff_len > 0:
+            for _ in range(diff_len):
+                shape.insert(0, 1)
+
+        output_shape = []
+
+        for sh, mul in list(zip(shape, multiples)):
+            if sh != 1 and mul != 1:
+                raise GKException("Tile op in expander only Support Automatic Broadcast!")
+            dim = sh * mul
+            output_shape.append(dim)
+        return output_shape
+
     def _expand(self, graph_builder):
         input_x = self.inputs[0]
-        multiples = self.attrs['multiples']
-
-        tile_infer = TileInfer(self.name, self.inputs, self.attrs)
-        output_shape, _, _ = tile_infer.infer()
-        if tile_infer.broadcast_compatible:
-            result = graph_builder.emit('BroadcastTo', [input_x], attrs={'shape': output_shape})
-        else:
-            result = graph_builder.emit('Tile', [input_x], attrs={'multiples': multiples})
+        output_shape = self._get_output_shape()
+
+        result = graph_builder.emit('BroadcastTo', [input_x], attrs={'shape': output_shape})
         return result

mindspore/_extends/graph_kernel/model/op_infer.py

@@ -198,51 +198,6 @@ class BroadcastTo(OpInfer):
         return self.inputs[0].data_format
 
 
-class Tile(OpInfer):
-    """Op Tile"""
-
-    def __init__(self, op_name, inputs, attrs):
-        super().__init__(op_name, inputs, attrs)
-        self.input_reshape = None
-        self.output_reshape = None
-        self.broadcast_compatible = True
-
-    def _infer_shape(self):
-        shape = self.inputs[0].shape
-        multiples = self.attrs["multiples"]
-
-        shape = list(shape)
-        multiples = list(multiples)
-        diff_len = len(multiples) - len(shape)
-        if diff_len < 0:
-            raise ValueError("Dimensions of multiples{} < dimensions of input{} in Tile".format(multiples, shape))
-        if diff_len > 0:
-            for _ in range(diff_len):
-                shape.insert(0, 1)
-
-        self.broadcast_compatible = True
-        output_shape = []
-        self.input_reshape = []
-        self.output_reshape = []
-        for sh, mul in list(zip(shape, multiples)):
-            dim = sh * mul
-            output_shape.append(dim)
-            if sh == 1 or mul == 1:
-                self.input_reshape.append(sh)
-                self.output_reshape.append(dim)
-            else:
-                self.broadcast_compatible = False
-                self.input_reshape.append(1)
-                self.input_reshape.append(sh)
-                self.output_reshape.append(mul)
-                self.output_reshape.append(sh)
-
-        return output_shape
-
-    def _infer_format(self):
-        return DF.DEFAULT
-
-
 class _CompareOp(_Elemwise):
     """Compare operators"""
 

mindspore/_extends/graph_kernel/splitter.py

@@ -61,7 +61,7 @@ def _dump_split_info(flags, graph_json, graph_desc, subgraphs, graph_mode):
         f.write("********** main graph: {} **********\n".format(graph_desc.name))
         f.write("input json:\n{}\n".format(graph_json))
         f.write("graph desc:\n{}\n".format(str(graph_desc)))
-        if len(subgraphs) > 1 or subgraphs[0].stitch_info is not None:
+        if len(subgraphs) > 1:
             for i, g in enumerate(subgraphs):
                 f.write("-------- subgraph {}, mode: {} --------\n".format(i, graph_mode[i]))
                 f.write("{}\n".format(str(g)))

mindspore/ccsrc/backend/optimizer/graph_kernel/graph_kernel_expander.cc

@@ -50,8 +50,8 @@ std::vector<PrimitivePtr> GetExpandOps() {
     prim::kPrimLayerNorm,
     prim::kPrimLayerNormGrad,
     prim::kPrimExpandDims,
-#if ENABLE_D
     prim::kPrimTile,
+#if ENABLE_D
     prim::kPrimSqrtGrad,
     prim::kPrimClipByNormNoDivSum,
     prim::kLambApplyOptimizerAssign,