create input dist module for sharded ec (#1923)

Summary:
Pull Request resolved: #1923

create separate input dist module for sharded qec, to enable input dist split

Reviewed By: gnahzg

Differential Revision: D56455884

fbshipit-source-id: 59751bc734575325cee269146388a16f968f2bcd

Author: jiayisuse

torchrec/distributed/quant_embedding.py

@@ -574,34 +574,6 @@ def _generate_permute_coordinates_per_feature_per_sharding(
                     torch.tensor(permuted_coordinates)
                 )
 
-    def _create_input_dist(
-        self,
-        input_feature_names: List[str],
-        device: torch.device,
-        input_dist_device: Optional[torch.device] = None,
-    ) -> None:
-        feature_names: List[str] = []
-        self._feature_splits: List[int] = []
-        for sharding in self._sharding_type_to_sharding.values():
-            self._input_dists.append(
-                sharding.create_input_dist(device=input_dist_device)
-            )
-            feature_names.extend(sharding.feature_names())
-            self._feature_splits.append(len(sharding.feature_names()))
-        self._features_order: List[int] = []
-        for f in feature_names:
-            self._features_order.append(input_feature_names.index(f))
-        self._features_order = (
-            []
-            if self._features_order == list(range(len(self._features_order)))
-            else self._features_order
-        )
-        self.register_buffer(
-            "_features_order_tensor",
-            torch.tensor(self._features_order, device=device, dtype=torch.int32),
-            persistent=False,
-        )
-
     def _create_lookups(
         self,
         fused_params: Optional[Dict[str, Any]],
@@ -627,46 +599,34 @@ def input_dist(
         features: KeyedJaggedTensor,
     ) -> ListOfKJTList:
         if self._has_uninitialized_input_dist:
-            self._create_input_dist(
+            self._intput_dist = ShardedQuantEcInputDist(
                 input_feature_names=features.keys() if features is not None else [],
-                device=features.device(),
-                input_dist_device=self._device,
+                sharding_type_to_sharding=self._sharding_type_to_sharding,
+                device=self._device,
+                feature_device=features.device(),
             )
             self._has_uninitialized_input_dist = False
         if self._has_uninitialized_output_dist:
             self._create_output_dist(features.device())
             self._has_uninitialized_output_dist = False
-        ret: List[KJTList] = []
+
+        (
+            input_dist_result_list,
+            features_by_sharding,
+            unbucketize_permute_tensor_list,
+        ) = self._intput_dist(features)
+
         with torch.no_grad():
-            features_by_sharding = []
-            if self._features_order:
-                features = features.permute(
-                    self._features_order,
-                    self._features_order_tensor,
-                )
-            features_by_sharding = (
-                [features]
-                if len(self._feature_splits) == 1
-                else features.split(self._feature_splits)
-            )
+            for i in range(len(self._sharding_type_to_sharding)):
 
-            for i in range(len(self._input_dists)):
-                input_dist = self._input_dists[i]
-                input_dist_result = input_dist.forward(features_by_sharding[i])
-                ret.append(input_dist_result)
                 ctx.sharding_contexts.append(
                     InferSequenceShardingContext(
-                        features=input_dist_result,
+                        features=input_dist_result_list[i],
                         features_before_input_dist=features_by_sharding[i],
-                        unbucketize_permute_tensor=(
-                            input_dist.unbucketize_permute_tensor
-                            if isinstance(input_dist, InferRwSparseFeaturesDist)
-                            or isinstance(input_dist, InferCPURwSparseFeaturesDist)
-                            else None
-                        ),
+                        unbucketize_permute_tensor=unbucketize_permute_tensor_list[i],
                     )
                 )
-        return ListOfKJTList(ret)
+        return input_dist_result_list
 
     def _embedding_dim_for_sharding_type(self, sharding_type: str) -> int:
         return (
@@ -680,7 +640,10 @@ def compute(
     ) -> List[List[torch.Tensor]]:
         ret: List[List[torch.Tensor]] = []
 
-        for lookup, features in zip(self._lookups, dist_input):
+        # for lookup, features in zip(self._lookups, dist_input):
+        for i in range(len(self._lookups)):
+            lookup = self._lookups[i]
+            features = dist_input[i]
             ret.append(lookup.forward(features))
         return ret
 
@@ -848,3 +811,126 @@ def shard(
     @property
     def module_type(self) -> Type[QuantEmbeddingCollection]:
         return QuantEmbeddingCollection
+
+
+class ShardedQuantEcInputDist(torch.nn.Module):
+    """
+    This module implements distributed inputs of a ShardedQuantEmbeddingCollection.
+
+    Args:
+        input_feature_names (List[str]): EmbeddingCollection feature names.
+        sharding_type_to_sharding (Dict[
+            str,
+            EmbeddingSharding[
+                InferSequenceShardingContext,
+                KJTList,
+                List[torch.Tensor],
+                List[torch.Tensor],
+            ],
+        ]): map from sharding type to EmbeddingSharding.
+        device (Optional[torch.device]): default compute device.
+        feature_device (Optional[torch.device]): runtime feature device.
+
+    Example::
+
+        sqec_input_dist = ShardedQuantEcInputDist(
+            sharding_type_to_sharding={
+                ShardingType.TABLE_WISE: InferTwSequenceEmbeddingSharding(
+                    [],
+                    ShardingEnv(
+                        world_size=2,
+                        rank=0,
+                        pg=0,
+                    ),
+                    torch.device("cpu")
+                )
+            },
+            device=torch.device("cpu"),
+        )
+
+        features = KeyedJaggedTensor(
+            keys=["f1", "f2"],
+            values=torch.tensor([0, 1, 2, 3, 4, 5, 6, 7]),
+            offsets=torch.tensor([0, 2, 2, 3, 4, 5, 8]),
+        )
+
+        sqec_input_dist(features)
+    """
+
+    def __init__(
+        self,
+        input_feature_names: List[str],
+        sharding_type_to_sharding: Dict[
+            str,
+            EmbeddingSharding[
+                InferSequenceShardingContext,
+                KJTList,
+                List[torch.Tensor],
+                List[torch.Tensor],
+            ],
+        ],
+        device: Optional[torch.device] = None,
+        feature_device: Optional[torch.device] = None,
+    ) -> None:
+        super().__init__()
+        self._sharding_type_to_sharding = sharding_type_to_sharding
+        self._input_dists = torch.nn.ModuleList([])
+        self._feature_splits: List[int] = []
+        self._features_order: List[int] = []
+
+        self._has_features_permute: bool = True
+
+        feature_names: List[str] = []
+        for sharding in sharding_type_to_sharding.values():
+            self._input_dists.append(sharding.create_input_dist(device=device))
+            feature_names.extend(sharding.feature_names())
+            self._feature_splits.append(len(sharding.feature_names()))
+
+        for f in feature_names:
+            self._features_order.append(input_feature_names.index(f))
+        self._features_order = (
+            []
+            if self._features_order == list(range(len(self._features_order)))
+            else self._features_order
+        )
+        self.register_buffer(
+            "_features_order_tensor",
+            torch.tensor(
+                self._features_order, device=feature_device, dtype=torch.int32
+            ),
+            persistent=False,
+        )
+
+    def forward(
+        self, features: KeyedJaggedTensor
+    ) -> Tuple[List[KJTList], List[KeyedJaggedTensor], List[Optional[torch.Tensor]]]:
+        with torch.no_grad():
+            ret: List[KJTList] = []
+            unbucketize_permute_tensor = []
+            if self._features_order:
+                features = features.permute(
+                    self._features_order,
+                    self._features_order_tensor,
+                )
+            features_by_sharding = (
+                [features]
+                if len(self._feature_splits) == 1
+                else features.split(self._feature_splits)
+            )
+
+            for i in range(len(self._input_dists)):
+                input_dist = self._input_dists[i]
+                input_dist_result = input_dist(features_by_sharding[i])
+                ret.append(input_dist_result)
+                unbucketize_permute_tensor.append(
+                    input_dist.unbucketize_permute_tensor
+                    if isinstance(input_dist, InferRwSparseFeaturesDist)
+                    or isinstance(input_dist, InferCPURwSparseFeaturesDist)
+                    else None
+                )
+
+            return (
+                ret,
+                features_by_sharding,
+                unbucketize_permute_tensor,
+            )