NNCFOVQuantizer and NNCFFXQuantizer

nncf/experimental/torch/fx/constant_folding.py

@@ -0,0 +1,268 @@
+# Copyright (c) 2024 Intel Corporation
+# 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 collections
+from typing import Any, Callable, Dict, List, Optional
+
+import torch
+import torch.utils._pytree as pytree
+
+aten = torch.ops.aten
+
+
+def _replace_node_with_constant(
+    gm: torch.fx.GraphModule,
+    node: torch.fx.Node,
+    constant: torch.Tensor,
+    name: Optional[str] = None,
+) -> None:
+    g = gm.graph
+
+    if name:
+        qualname = name
+    else:
+        if not hasattr(gm, "_frozen_param_count"):
+            gm._frozen_param_count = 0  # type: ignore[assignment]
+        i = gm._frozen_param_count
+
+        while True:
+            qualname = f"_frozen_param{i}"
+            if not hasattr(gm, qualname):
+                break
+            i += 1
+
+        gm._frozen_param_count = i + 1
+
+    with g.inserting_before(node):
+        new_input_node = g.create_node("get_attr", qualname, (), {})
+        node.replace_all_uses_with(new_input_node)
+        new_input_node.meta.update(node.meta)
+        g.erase_node(node)
+
+    # needed to suppress `does not reference an nn.Module, nn.Parameter, or buffer` warning
+    gm.register_buffer(qualname, constant)
+    setattr(gm, qualname, constant)
+
+
+def _is_const_source(node: torch.fx.Node, lifted_constants: Optional[Dict[str, Any]]) -> bool:
+    return node.op == "get_attr" or (
+        node.op == "placeholder" and lifted_constants is not None and node.name in lifted_constants
+    )
+
+
+class _ConstantFolder(torch.fx.Interpreter):
+    def __init__(
+        self,
+        gm: torch.fx.GraphModule,
+        skip_constructors: bool = False,
+        lifted_constants: Optional[Dict[str, torch.Tensor]] = None,
+        skip_folding_node_fn: Optional[Callable[[torch.fx.Node], bool]] = None,
+    ) -> None:
+        super().__init__(gm)
+        self.node_replacements: Dict[torch.fx.Node, Any] = {}
+        self.replaced_uses: Dict[torch.fx.Node, int] = collections.Counter()
+        self.unknown_value = object()
+        self.skip_constructors: bool = skip_constructors
+
+        # overwrite this to deallocate env values if their only remaining use
+        # is the output
+        self.user_to_last_uses = self.node_to_last_non_output_use()
+        self.lifted_constants = lifted_constants
+
+    def _support_dynamic_shape(self) -> bool:
+        # ConstantFolder not support dynamic shape now
+        return False
+
+    def _deduce_value(self, node: torch.fx.Node) -> Any:
+        return super().run_node(node)
+
+    def is_impure(self, node: torch.fx.node.Node) -> bool:
+        def is_woq_int8_pattern(node: torch.fx.node.Node) -> bool:
+            return (
+                node.target == torch.ops.prims.convert_element_type.default  # type: ignore[return-value]
+                and isinstance(node.args[0], torch.fx.Node)
+                and "val" in node.args[0].meta
+                and node.args[0].meta["val"].dtype == torch.int8  # type: ignore[union-attr]
+                and node.args[1] == torch.bfloat16
+            )
+
+        if (
+            is_woq_int8_pattern(node)
+            or (
+                node.target == torch.ops.aten.permute.default
+                and len(node.users) == 1
+                and is_woq_int8_pattern(next(iter(node.users)))
+            )
+        ) and _is_const_source(
+            node.args[0], self.lifted_constants  # type: ignore[arg-type]
+        ):
+            # Case 1: int8_weight -> dq -> bf16_weight
+            # Case 2: int8_weight -> permute -> dq -> bf16_weight
+            return True
+
+        quant_registered = getattr(torch.ops.quantized_decomposed, "dequantize_per_channel", None) is not None
+        if quant_registered and node.target in [
+            torch.ops.quantized_decomposed.dequantize_per_channel.default,
+            torch.ops.quantized_decomposed.dequantize_per_tensor.default,
+            torch.ops.quantized_decomposed.dequantize_per_tensor.tensor,
+        ]:
+            # For the pattern fp32_weight -> q -> dq
+            # We only folding fp32_weight -> q
+            # int8_weight and leave dq in graph to be fused
+            return True
+        return False
+
+    def node_to_last_non_output_use(self) -> Dict[torch.fx.Node, List[torch.fx.Node]]:
+        last_non_output_use = collections.defaultdict(list)
+        seen_uses = set()
+        output_node = next(iter(reversed(self.module.graph.nodes)))
+
+        for node in reversed(self.module.graph.nodes):
+            if node.target == "output":
+                continue
+
+            def add_use(inp: torch.fx.Node) -> None:
+                if inp in seen_uses:
+                    return
+
+                seen_uses.add(inp)
+                last_non_output_use[node].append(inp)
+
+            # In-place is fine since we don't mutate
+            pytree.tree_map_only_(torch.fx.Node, add_use, (node.args, node.kwargs))
+
+            # if this node is only used in output, we want to gc it right away
+            if len(node.users) == 1 and output_node in node.users:
+                last_non_output_use[node].append(node)
+
+        return last_non_output_use
+
+    def run_node(self, node: torch.fx.Node) -> Any:
+        if node.target == "output":
+            # because we remove nodes from env on last non output use,
+            # re-define them now or we'll get error in interpreter
+            def set_env(arg: torch.fx.Node) -> None:
+                self.env[arg] = self.unknown_value
+
+            # In-place is fine since we don't mutate
+            pytree.tree_map_only_(torch.fx.Node, set_env, node.args)
+            return super().run_node(node)
+
+        args, kwargs = self.fetch_args_kwargs_from_env(node)
+        flattened_inputs = pytree.arg_tree_leaves(*args, **kwargs)
+
+        # We need to do this weird thing because in cases where flattened_inputs
+        # contains a ScriptObject, equality checking results in a type error if
+        # the types are different.
+        if any(
+            type(self.unknown_value) is type(input_) and self.unknown_value == input_ for input_ in flattened_inputs
+        ):
+            return self.unknown_value
+
+        # TODO - fix errors with this
+        if node.op == "call_function" and node.target == aten._efficientzerotensor.default:
+            return self.unknown_value
+
+        # TODO - constant folding triton kernel returns the inputs -- fix this
+        if node.op == "call_function" and node.name == "triton_kernel_wrapper_functional_proxy":
+            return self.unknown_value
+
+        # skip constructors, since inductor generates optimal code for them already
+        # and turning into tensor would result in an additional global memory read
+        # TODO - more complicated strategy
+        if (
+            self.skip_constructors
+            and not _is_const_source(node, self.lifted_constants)
+            and not any(isinstance(e, torch.Tensor) for e in flattened_inputs)
+        ):
+            return self.unknown_value
+
+        # All mutations should either be removed or on inputs which we did not make constant
+        if isinstance(node.target, torch._ops.OpOverload) and torch.Tag.nondeterministic_seeded in node.target.tags:
+            return self.unknown_value
+
+        out = self._deduce_value(node)
+        if out == self.unknown_value:
+            return self.unknown_value
+
+        if not _is_const_source(node, self.lifted_constants) and isinstance(out, torch.Tensor):
+            if out.device.type == "meta":
+                return out
+
+            if not self.insertable_tensor_check(out):
+                return out
+
+            if self.is_impure(node):
+                return self.unknown_value
+
+            self.add_node_replacement(node, out)
+
+            flattened_node_inps = pytree.arg_tree_leaves(*node.args, **node.kwargs)
+
+            for n in flattened_node_inps:
+                if not isinstance(n, torch.fx.Node):
+                    continue
+
+                self.replaced_uses[n] += 1
+
+            for to_delete in self.user_to_last_uses.get(node, []):
+                if self.replaced_uses[to_delete] == len(to_delete.users):
+                    self.node_replacements.pop(to_delete, None)
+
+        return out
+
+    def insertable_tensor_check(self, tensor: torch.Tensor) -> bool:
+        return True
+
+    def add_node_replacement(self, node: torch.fx.Node, tensor: torch.Tensor) -> None:
+        self.node_replacements[node] = tensor
+
+    def run(self) -> Any:  # type: ignore[override]
+        env: Dict[torch.fx.Node, Any] = {}
+        self.insert_placerholder_values(env)
+        return super().run(initial_env=env)
+
+    def insert_placerholder_values(self, env: Dict[torch.fx.Node, Any]) -> None:
+        for n in self.module.graph.find_nodes(op="placeholder"):
+            if self.lifted_constants is not None and n.name in self.lifted_constants:
+                env[n] = self.lifted_constants[n.name]
+            else:
+                env[n] = self.unknown_value  # type: ignore[assignment]
+
+
+def constant_fold(
+    gm: torch.fx.GraphModule,
+) -> None:
+    """
+    Calcualtes constant subgraphs values and replaces them with a constan node inplace.
+
+    :param gm: Given graph model.
+    """
+    with torch.utils._python_dispatch._disable_current_modes():
+        cf = _ConstantFolder(gm, skip_constructors=True)
+        cf.run()
+
+        for node, constant in cf.node_replacements.items():
+            _replace_node_with_constant(gm, node, constant)
+
+        erased_params = []
+        for node in gm.graph.find_nodes(op="get_attr"):
+            if len(node.users) == 0:
+                if hasattr(gm, node.target):
+                    delattr(gm, node.target)
+                erased_params.append(node)
+
+        for node in erased_params:
+            gm.graph.erase_node(node)
+
+        gm.graph.eliminate_dead_code()
+        gm.graph.lint()
+        gm.recompile()

nncf/experimental/torch/fx/model_transformer.py

@@ -126,11 +126,14 @@ def _apply_model_extraction(
         def remap_fn(node: torch.fx.Node):
             return value_remap.get(node)  # noqa F821
 
+        visited_outputs_names = []
         for node in model.graph.nodes:
-            if node.name not in visited or node.op == "output":
+            if node.name not in visited:
+                continue
+            if node.op == "output":
+                visited_outputs_names.append(node.name)
                 continue
             value_remap[node] = extracted_graph.node_copy(node, remap_fn)
-        del value_remap
 
         for input_name in transformation.input_node_names:
             node_with_input = get_graph_node_by_name(extracted_graph, input_name)
@@ -146,7 +149,29 @@ def remap_fn(node: torch.fx.Node):
             args[0] = graph_input
             node_with_input.args = tuple(args)
 
-        nodes_with_output = [get_graph_node_by_name(extracted_graph, name) for name in transformation.output_node_names]
+        # Merge new output with the original output in case
+        # the original output is requested in the extracted graph.
+        nodes_with_output = []
+        for name in transformation.output_node_names:
+            nodes_with_output.append(
+                name if name in visited_outputs_names else get_graph_node_by_name(extracted_graph, name)
+            )
+
+        for idx, node in enumerate(nodes_with_output):
+            if isinstance(node, torch.fx.Node):
+                continue
+            output_node = get_graph_node_by_name(model.graph, node)
+            args = output_node.args[0]
+            if isinstance(args, torch.fx.Node):
+                args = value_remap[args]
+            else:
+                args = [value_remap[n] for n in args]
+                # Unpack target output args in case
+                # only one arg is presented.
+                if len(args) == 1:
+                    args = args[0]
+            nodes_with_output[idx] = args
+
         last_node = list(extracted_graph.nodes)[-1]
         with extracted_graph.inserting_after(last_node):
             graph_output_name = "output"

nncf/experimental/torch/fx/node_utils.py

@@ -66,11 +66,10 @@ def is_node_with_bias(node: NNCFNode, nncf_graph: NNCFGraph) -> bool:
     :return: True if the node has a bias, False otherwise.
     """
     # Assumes that all biases were unfused
-    if node.metatype in FX_OPERATORS_WITH_BIAS_METATYPES:
-        next_nodes = nncf_graph.get_next_nodes(node)
-        if len(next_nodes) != 1:
-            return False
-        return next_nodes[0].metatype in (om.PTAddMetatype,)
+    if node.metatype not in FX_OPERATORS_WITH_BIAS_METATYPES or len(nncf_graph.get_input_edges(node)) != 3:
+        return False
+    const_node = nncf_graph.get_input_edge_by_port_id(node, 2).from_node
+    return const_node.metatype is om.PTConstNoopMetatype
 
 
 def get_bias_value(node: NNCFNode, nncf_graph: NNCFGraph, model: torch.fx.GraphModule) -> Tensor:
@@ -82,7 +81,7 @@ def get_bias_value(node: NNCFNode, nncf_graph: NNCFGraph, model: torch.fx.GraphM
     :param model: Target GraphModule.
     :return: Bias value of the given node.
     """
-    bias_node = nncf_graph.get_next_nodes(node)[0]
+    bias_node = nncf_graph.get_input_edge_by_port_id(node, 2).from_node
     # TODO(dlyakhov): make a node_name_vs_node map to speed up the process
-    graph_bias_node = get_graph_node_by_name(model.graph, bias_node.node_name)
-    return Tensor(get_tensor_constant_from_node(graph_bias_node.all_input_nodes[1], model))
+    graph_bias_const = get_graph_node_by_name(model.graph, bias_node.node_name)
+    return Tensor(get_tensor_constant_from_node(graph_bias_const, model))