From 161998f715039c442732d26e232b2a9e23ecd630 Mon Sep 17 00:00:00 2001
From: JZ-LIANG <jianzhongliang10@gmail.com>
Date: Tue, 22 Nov 2022 15:12:11 +0800
Subject: [PATCH] [Auto Parallel] Recompute Support New Graph Executor (#47846)

* add depend

* fp16 pass distinguish None & False

* engine log
---
 .../paddle/distributed/auto_parallel/utils.py | 91 +++++++++++++++++++
 .../passes/auto_parallel_recompute.py         | 12 ++-
 2 files changed, 102 insertions(+), 1 deletion(-)

diff --git a/python/paddle/distributed/auto_parallel/utils.py b/python/paddle/distributed/auto_parallel/utils.py
index be4c68d97d..c31642a9e2 100644
--- a/python/paddle/distributed/auto_parallel/utils.py
+++ b/python/paddle/distributed/auto_parallel/utils.py
@@ -1407,6 +1407,27 @@ def naive_set_dist_op_attr_for_program_by_mesh_and_mapping(
     ctx.set_op_dist_attr_for_program(new_op, new_op_dist_attr)
 
 
+def naive_set_dist_op_attr_for_program_by_mesh(
+    new_op, process_mesh, ctx, is_recompute=False
+):
+    assert process_mesh is not None
+
+    new_op_dist_attr = OperatorDistributedAttribute()
+
+    for input_varname in new_op.desc.input_arg_names():
+        var = ctx.serial_main_program.global_block().var(input_varname)
+        mapping = ctx.get_tensor_dist_attr_for_program(var).dims_mapping
+        new_op_dist_attr.set_input_dims_mapping(input_varname, mapping)
+    for output_varname in new_op.desc.output_arg_names():
+        var = ctx.serial_main_program.global_block().var(output_varname)
+        mapping = ctx.get_tensor_dist_attr_for_program(var).dims_mapping
+        new_op_dist_attr.set_output_dims_mapping(output_varname, mapping)
+
+    new_op_dist_attr.process_mesh = process_mesh
+    new_op_dist_attr.is_recompute = is_recompute
+    ctx.set_op_dist_attr_for_program(new_op, new_op_dist_attr)
+
+
 def update_op_dims_mapping_by_default_dist_impl(dist_op):
     changed = False
     op_dist_attr = dist_op.dist_attr
@@ -2102,3 +2123,73 @@ def _copy_dist_attr_from_cpp_for_graph(dist_context):
             py_dist_attr = dist_context.get_op_dist_attr_for_graph(node)
             cpp_dist_attr = node.op().dist_attr
             _copy_op_dist_attr_from_cpp(cpp_dist_attr, py_dist_attr)
+
+
+def insert_dependencies_for_two_ops(
+    block,
+    idx,
+    prior_op,
+    posterior,
+    dist_context,
+    is_recompute=False,
+    sync=False,
+):
+    """
+    dependency: prior_op should be run before posterior
+    """
+
+    assert (
+        len(prior_op.output_arg_names) >= 1
+    ), "first op of dependency should at least have one output. [{}]".format(
+        str(prior_op)
+    )
+    assert (
+        len(posterior.input_arg_names) >= 1
+    ), "second op of dependency should at least have one input. [{}]".format(
+        str(posterior)
+    )
+    prior_op_mesh = dist_context.get_op_dist_attr_for_program(
+        prior_op
+    ).process_mesh
+    posterior_mesh = dist_context.get_op_dist_attr_for_program(
+        posterior
+    ).process_mesh
+    assert (
+        prior_op_mesh == posterior_mesh
+    ), "two ops of dependency should have same mesh but got [{}] and [{}]".format(
+        str(prior_op_mesh), str(posterior_mesh)
+    )
+
+    def _select_best_depend_var(vars):
+
+        vars_with_numels = [(var, get_var_numel(var)) for var in vars]
+        vars_with_numels.sort(key=lambda x: x[1])
+
+        return vars_with_numels[-1][0]
+
+    first_var = _select_best_depend_var(
+        [block.var(name) for name in prior_op.output_arg_names]
+    )
+    second_var = _select_best_depend_var(
+        [block.var(name) for name in posterior.input_arg_names]
+    )
+
+    depend_op = block._insert_op_without_sync(
+        idx,
+        type='nop',
+        inputs={
+            "X": first_var,
+        },
+        outputs={"Out": second_var},
+    )
+    # depend_op.desc.set_type("depend")
+    depend_op._set_attr(OP_ROLE_KEY, OpRole.Backward)
+    # depend_op.desc.set_input("Dep", [first_var.name])
+    # self.desc.set_output(out_proto.name, out_arg_names)
+
+    naive_set_dist_op_attr_for_program_by_mesh(
+        depend_op, prior_op_mesh, dist_context, is_recompute
+    )
+
+    if sync:
+        block._sync_with_cpp()
diff --git a/python/paddle/distributed/passes/auto_parallel_recompute.py b/python/paddle/distributed/passes/auto_parallel_recompute.py
index 5bdbe9d2dd..44e02fb3ff 100644
--- a/python/paddle/distributed/passes/auto_parallel_recompute.py
+++ b/python/paddle/distributed/passes/auto_parallel_recompute.py
@@ -29,6 +29,7 @@ from paddle.distributed.auto_parallel.utils import (
 )
 from paddle.distributed.auto_parallel.utils import (
     naive_set_dist_op_attr_for_program_by_mesh_and_mapping,
+    insert_dependencies_for_two_ops,
 )
 
 
@@ -449,6 +450,7 @@ class RecomputePass(PassBase):
                     while idx - 1 >= 0 and ops[idx - 1].type == "sum":
                         idx -= 1
                     segment_descs = ckpt_ops_dict[fwd_op_id][1]
+                    rc_op = None
                     for _, op_desc in reversed(list(enumerate(segment_descs))):
                         rc_op = main_block._insert_op_without_sync(
                             idx, type='nop'
@@ -466,7 +468,15 @@ class RecomputePass(PassBase):
                         )
 
                     ckpt_ops_dict[fwd_op_id][0] = False
-
+                    if rc_op:
+                        insert_dependencies_for_two_ops(
+                            main_block,
+                            idx,
+                            main_block.ops[rc_op.idx - 1],
+                            rc_op,
+                            self._dist_context,
+                            sync=False,
+                        )
         main_program._sync_with_cpp()
 
     def reset_op_dist_attr(self, op, var_name_dict):
-- 
GitLab