From 8ee23da846075f902d29e2c6bd10cb27bb0fd489 Mon Sep 17 00:00:00 2001
From: Helin Wang <ustc.harry@gmail.com>
Date: Thu, 3 May 2018 10:17:11 -0700
Subject: [PATCH] Fluid new API: dist train without modifying code

Works with 1 trainer 1 pserver. 2 trainer 1 pserver will stuck at the
end of first step, still investigating.

The user only need to set envrionment variables to enable distributed
training.

run pserver:

PADDLE_TRAINING_ROLE=PSERVER PADDLE_PSERVER_IPS=127.0.0.1 PADDLE_TRAINERS=2 PADDLE_CURRENT_IP=127.0.0.1 python no_test_word2vec_new_api.py

run trainer:

PADDLE_TRAINING_ROLE=TRAINER PADDLE_PSERVER_IPS=127.0.0.1 PADDLE_TRAINERS=2 PADDLE_TRAINER_ID=0 python no_test_word2vec_new_api.py
---
 python/paddle/fluid/trainer.py | 56 +++++++++++++++++++++++++++++++---
 1 file changed, 52 insertions(+), 4 deletions(-)

diff --git a/python/paddle/fluid/trainer.py b/python/paddle/fluid/trainer.py
index 0aada3deb0f..5385d798ea8 100644
--- a/python/paddle/fluid/trainer.py
+++ b/python/paddle/fluid/trainer.py
@@ -12,6 +12,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+import os
 import core
 import framework
 import executor
@@ -20,6 +21,7 @@ import contextlib
 
 # optimizer is same as the parameter of Trainer.__init__. Rename it to opt_module
 import optimizer as opt_module
+import distribute_transpiler
 
 __all__ = [
     'Trainer',
@@ -76,22 +78,61 @@ class Trainer(object):
                 raise TypeError(
                     "The optimizer should be an instance of Optimizer")
 
-            optimizer.minimize(loss)
+            optimize_ops, params_grads = optimizer.minimize(loss)
 
         self.place = Trainer._check_and_get_place(place)
 
+        self.dist_transpile_if_necessary(optimize_ops, params_grads)
+
         # 2. move the default_main_program to self.program and run the
         # default_startup program on an empty core.Scope()
         # Run startup program
-        exe = executor.Executor(place)
-        exe.run(self.startup_program, scope=self.scope)
+        with self._prog_and_scope_guard():
+            exe = executor.Executor(place)
+            exe.run(self.startup_program)
 
         if param_path:
             # load params from param_path into scope
             # TODO(yuyang): This depends on parameters implementation.
             pass
 
-        # TODO(helin): support distributed training
+    def dist_transpile_if_necessary(self, optimize_ops, params_grads):
+        if "PADDLE_TRAINING_ROLE" not in os.environ:
+            return
+
+        # the port of all pservers, needed by both trainer and pserver
+        port = os.getenv("PADDLE_PSERVER_PORT", "6174")
+        # comma separated ips of all pservers, needed by trainer and
+        # pserver
+        pserver_ips = os.getenv("PADDLE_PSERVER_IPS", "")
+        eplist = []
+        for ip in pserver_ips.split(","):
+            eplist.append(':'.join([ip, port]))
+        pserver_endpoints = ",".join(eplist)
+        # total number of workers/trainers in the job, needed by
+        # trainer and pserver
+        trainers = int(os.getenv("PADDLE_TRAINERS"))
+        # the IP of the local machine, needed by pserver only
+        current_endpoint = os.getenv("PADDLE_CURRENT_IP", "") + ":" + port
+        # the unique trainer id, starting from 0, needed by trainer
+        # only
+        trainer_id = int(os.getenv("PADDLE_TRAINER_ID", "0"))
+        # the role, should be either PSERVER or TRAINER
+        training_role = os.getenv("PADDLE_TRAINING_ROLE")
+        with self._prog_and_scope_guard():
+            t = distribute_transpiler.DistributeTranspiler()
+            t.transpile(
+                trainer_id, pservers=pserver_endpoints, trainers=trainers)
+            if training_role == "PSERVER":
+                self.train_program = t.get_pserver_program(current_endpoint)
+                self.startup_program = t.get_startup_program(current_endpoint,
+                                                             self.train_program)
+            elif training_role == "TRAINER":
+                self.train_program = t.get_trainer_program()
+            else:
+                raise ValueError(
+                    'TRAINING_ROLE environment variable must be either TRAINER or PSERVER'
+                )
 
     def train(self,
               num_epochs,
@@ -117,6 +158,13 @@ class Trainer(object):
             raise NotImplementedError(
                 "Parallel Executor version of trainer is not implemented")
 
+        training_role = os.getenv("PADDLE_TRAINING_ROLE", "")
+        if training_role == "PSERVER":
+            with self._prog_and_scope_guard():
+                exe = executor.Executor(self.place)
+                exe.run()
+                return
+
         self._train_by_executor(num_epochs, event_handler, reader, feed_order)
 
     def test(self, reader):
-- 
GitLab