Merge pull request #5776 from typhoonzero/update_refactor_dist_train_doc

Update design of dist train refactor

doc/design/refactor/distributed_architecture.md → doc/design/dist_refactor/distributed_architecture.md

@@ -52,77 +52,121 @@ The IR for PaddlePaddle after refactoring is called a `Block`, it specifies the
 
 The user can not directly specify the parameter update rule for the parameter server in the Python module, since the parameter server does not use the same computation definition as the trainer. Instead, the update rule is baked inside the parameter server. The user can not specify the update rule explicitly.
 
-This could be fixed by making the parameter server run the same computation definition as the trainer (the user's Python module). For a detailed explanation, refer to this document -
-[Design Doc: Operation Graph Based Parameter Server](./parameter_server.md)
+This could be fixed by making the parameter server also run an IR, which can be different to the trainer side
+For a detailed explanation, refer to this document -
+[Design Doc: Parameter Server](./parameter_server.md)
 
 ## Distributed Training Architecture
 
 The revamped distributed training architecture can address the above discussed limitations. Below is the illustration of how it does so:
 
 <img src="src/distributed_architecture.png"/>
 
-The major components in the architecture are: *PaddlePaddle Python*, *PaddlePaddle converter* and *PaddlePaddle runtime*.
+The major components are: *Python API*, *Distribute Transpiler* and *Remote Executor*.
 
-### PaddlePaddle Python
+### Python API
 
-PaddlePaddle Python is the Python library that user's Python code invokes, to read the data. build the neural network topology, start training, etc.
+Python API is the Python library that user's Python code invokes, to read the data, build the neural network topology, and start training, etc.
 
 ```Python
-paddle.init()
-input = paddle.op.recordIO("/home/data/mnist.recordio") # file stored on the cluster
-img, label = input[0], input[1]
-hidden = paddle.layer.fc(input=img, size=200, act=paddle.activation.Tanh())
-prediction = paddle.layer.fc(input=img, size=10, act=paddle.activation.Softmax())
-cost = paddle.layer.classification_cost(input=prediction, label=label)
-optimizer = paddle.optimizer.SGD(cost, learning_rate=0.01)
-session = paddle.session.NewRemote(num_trainer=3, num_ps=2, GPU_per_trainer=1)
-for i in range(1000):
-	_, cost_val = session.eval(targets=[cost, optimizer])
-	print cost_val
+images = fluid.layers.data(name='pixel', shape=[1, 28, 28], dtype='float32')
+label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+...
+predict = fluid.layers.fc(input=conv_pool_2, size=10, act="softmax")
+cost = fluid.layers.cross_entropy(input=predict, label=label)
+avg_cost = fluid.layers.mean(x=cost)
+optimizer = fluid.optimizer.Adam(learning_rate=0.01)
+optimizer.minimize(avg_cost)
+
+train_reader = paddle.batch(
+    paddle.reader.shuffle(
+        paddle.dataset.mnist.train(), buf_size=500),
+    batch_size=BATCH_SIZE)
+
+place = fluid.CPUPlace()
+exe = fluid.Executor(place)
+
+for pass_id in range(10):
+    for data in train_reader():
+        loss, acc = exe.run(trainer_prog,
+                            feed=feeder.feed(data),
+                            fetch_list=[avg_cost])
 ```
 
-The above code is what a typical Python trainer code is, the neural network topology is built using the helper functions such as `paddle.layer.fc`. Training is done by calling `session.eval` iteratively.
-
-#### session.eval
-
-As shown in the graph, `session.eval` sends the IR and the evaluation inputs or targets to the PaddlePaddle cluster for evaluation.
-The targets can be any variable in the computation graph. When the target is say, the `optimizer` variable, the neural network will be optimized once. When the target is the `cost` variable, `session.eval` returns the cost value. Based on what the target is, an appropriate action is taken.
-
-The Python `session` is a wrapper of the C++ `Session` class. For more information about `Session`, refer to this document - [Design Doc: Session](./session.md).
-
-### PaddlePaddle Converter
-
-The PaddlePaddle converter automatically converts the IR in the request (IR and evaluation inputs/targets) from PaddlePaddle Python to partitioned IRs and dispatches the new IRs and evaluation inputs/targets to different PaddlePaddle runtimes. Below are the steps that are followed :
-
-1. Add a `feed` OP that feeds the eval inputs, and a `fetch` OP that fetches the eval targets to the IR.
-
-2. Extract a new computation (sub)graph with the `feed` and `fetch` OPs as the boundary. The runtime does not need to run the OP that is not dependent on the `fetch` OP.
-
-3. Optimize the computation graph.
-
-4. Place the OPs in the graph onto different devices on different PaddlePaddle runtime according to a placement algorithm and the device constraints specified by the user.
-
-5. Partition the graph according to runtime boundaries and add `send` / `recv` OP pair on the runtime boundaries.
+The code above is a typical local training program, the "Training Program" is built using helper functions such as
+`fluid.layer.fc`. The training is done by calling `Executor.run`
+iteratively.
+
+For more details, the implementation of IR is [Program](../program.md), and `ProgramDesc` is the protobuf type.
+
+[Executor](../executor.md) simply runs the `ProgramDesc`. For local training you generally use
+`Executor` to run the program locally. For any kind of distributed training, you can use
+`RemoteExecutor` to specify desired distributed training method with some optional arguments.
+
+### Distributed Transpiler
+
+The Distributed Transpiler automatically converts the IR (in protobuf format) to partitioned IRs. Then
+the Remote Executor dispatches the new IRs to Remote Executors across the cluster.
+Below are the steps that are followed :
+
+1. User only need to change `Executor` to `RemoteExecutor` to change local program to distributed program.
+1. `RemoteExecutor` calls `Distributed Transpiler` to "transpile" user's program to several IRs representing a
+   distributed training program:
+   1. Parse configurations from `RemoteExecutor`.
+   1. Determine the type of distributed program, can be DataParallelism, ModelParallelism or Streaming.
+   1. Partition the `ProgramDesc` according to type and add `send` / `recv` OP pair on the boundaries. Take
+      DataParallelism type for example, it removes the optimization operators and add a `send` OP to the
+      "trainer" role, then add the optimization operators to the parameter server role within the `recv` OP.
+1. Dispatch the partitioned graph to different `RemoteExecutor` in the cluster.
+1. `RemoteExecutor` on each node run the received `ProgramDesc` utill the end.
+
+
+### RemoteExecutor
+
+As shown in the graph, `RemoteExecutor.run` sends the IR to the cluster for Execution.
+You can also use parameter `fetch_list` to interactively fetch variable back to local for
+log printing.
+
+The Python `RemoteExecutor` is derived from `Executor` class.
+
+```python
+exe = RemoteExecutor(
+    feed=feeder.feed(data),
+    fetch_list=[avg_cost],
+    job_desc=JobDesc(
+      jobname,
+      num_trainer,
+      num_pserver,
+      cpu_per_trainer,
+      gpu_per_trainer,
+      mem_per_trainer,
+      cpu_per_pserver,
+      mem_per_pserver
+    ))
+for data in train_reader():
+    loss, acc = exe.run(trainer_prog,
+                        feed=feeder.feed(data),
+                        fetch_list=[avg_cost])
+```
 
-6. Dispatch the partitioned graph to different PaddlePaddle runtimes.
+`JobDesc` object describe the distributed job resource specification to run on
+Cluster environment.
 
-7. PaddlePaddle runtimes with the `fetch` OP reports evaluation results back to the converter, the converter reports the evaluation results back to the PaddlePaddle Python.
+<img src="src/remote_executor.png"/>
 
-The output IRs will be cached to optimize the conversion latency.
+`RemoteExecutor.run` sends the `ProgramDesc` and
+[TrainingJob](https://github.com/PaddlePaddle/cloud/blob/develop/doc/autoscale/README.md#training-job-resource)
+to a server in the cluster which executes `RemoteExecutor.listen`. This server is responsible
+to start the final Kubernetes Jobs to run the different role of `ProgramDesc`.
 
 
-#### Placement Algorithm
+### Placement Algorithm
 
 Our first implementation will only support "trainer-parameter server" placement: the parameters, initializers, and optimizers are all placed on the PaddlePaddle runtimes with the parameter server role. Everything else will be placed on the PaddlePaddle runtimes with the trainer role. This has the same functionality as the "trainer-parameter server" architecture of PaddlePaddle v0.10.0, but is more generic and flexible.
 
 In the future, a more general placement algorithm should be implemented, which makes placements according to the input IR, and a model of device computation time and device communication time. Model parallelism requires the generic placement algorithm.
 
 
-### PaddlePaddle Runtime
-
-The PaddlePaddle runtime owns multiple devices (e.g., CPUs, GPUs) and runs the IR. The runtime does not need to do OP placement since it is already done by the converter.
-
-
 ### Local Training Architecture
 
 The local training architecture will be the same as the distributed training architecture, the difference is that everything runs locally, and there is just one PaddlePaddle runtime:
@@ -132,9 +176,18 @@ The local training architecture will be the same as the distributed training arc
 
 ### Training Data
 
-In PaddlePaddle v0.10.0, training data is typically read with a [data reader](../reader/README.md) from Python. This approach is no longer efficient when training in a distributed fashion since the Python process no longer runs on the same node with the trainer processes. The Python reader will need to read from the distributed filesystem (assuming it has the required access) and send to the trainers, doubling the network traffic.
-
-When doing distributed training, the user can still use Python data reader: the training data are sent with `session.eval`. However this should be used for debugging purpose only. The users are encouraged to use the read data OPs.
+In PaddlePaddle v0.10.0, training data is typically read
+with [data reader](../reader/README.md) from Python. This approach is
+no longer efficient when training distributedly since the Python
+process no longer runs on the same node with the trainer processes,
+the Python reader will need to read from the distributed filesystem
+(assuming it has the access) and send to the trainers, doubling the
+network traffic.
+
+When doing distributed training, the user can still use Python data
+reader: the training data are sent with `Executor.run`. However, should
+be used for debugging purpose only. The users are encouraged to use
+the read data OPs.
 
 
 ## References:

doc/design/refactor/parameter_server.md → doc/design/dist_refactor/parameter_server.md

@@ -1,4 +1,4 @@
-# Design Doc: Operation Graph Based Parameter Server
+# Design Doc: Parameter Server
 
 ## Abstract
 
@@ -10,7 +10,7 @@ different purposes.
 ## Background
 
 The previous implementations of the parameter server does not run a
-subgraph. parameter initialization, optimizer computation, network
+fluid sub-program. Parameter initialization, optimizer computation, network
 communication and checkpointing are implemented twice on both the
 trainer and the parameter server.
 
@@ -23,18 +23,17 @@ server becomes a natural extension.
 
 ## Design
 
-### Graph Converter
+### Distributed Transpiler
 
-The *graph converter* converts the user-defined operation (OP) graph
-into subgraphs to be scheduled on different nodes with the following
+The *Distributed Transpiler* converts the user-defined fluid program
+into sub-programs to be scheduled on different nodes with the following
 steps:
 
 1. OP placement: the OPs will be placed on different nodes according
    to heuristic that minimizes estimated total computation
    time. Currently we will use a simple heuristic that puts parameter
    varable on parameter server workers and everything else on trainer
    workers.
-
 1. Add communication OPs to enable the communication between nodes.
 
 We will need these OPs: *Send*, *Recv*, *Enqueue*, *Dequeue*.
@@ -48,8 +47,8 @@ After converting:
 
 <img src="src/dist-graph.png" width="700"/>
 
-1. The parameter variable W and it's optimizer subgraph are placed on the parameter server.
-1. Operators are added to the subgraphs.
+1. The parameter variable W and it's optimizer program are placed on the parameter server.
+1. Operators are added to the program.
    - *Send* sends data to the connected *Recv* operator.  The
 	 scheduler on the receive node will only schedule *Recv* operator
 	 to run when the *Send* operator has ran (the *Send* OP will mark
@@ -64,39 +63,30 @@ After converting:
 ### Benefits
 
 - Model parallelism become easier to implement: it's an extension to
-  the trainer - parameter server approach. we already have the
-  communication OPs, but need to extend the graph converter's
-  placement functionality.
-
+  the trainer - parameter server approach. We can have several "Transpilers"
+  to achieve different goals.
 - User-defined optimizer is easier to add - user can now express it as
-  a subgraph.
-
+  a sub-program.
 - No more duplication logic inside the trainer and the parameter
   server mentioned in the background section.
 
 ### Challenges
 
-- It might be hard for the graph converter to cut a general graph
-  (without any hint for which subgraph is the optimizer). We may need
-  to label which subgraph inside the OP graph is the optimizer.
-
 - It's important to balance the parameter shards of on multiple
   parameter server. If a single parameter is very big (some
   word-embedding, fully connected, softmax layer), we need to
   automatically partition the single parameter onto different
   parameter servers when possible (only element-wise optimizer depends
   on the parameter variable).
+- In the "Aync SGD" figure, the "W" variable on the parameter server
+  could be read and wrote concurrently. See
+  [here](https://github.com/PaddlePaddle/Paddle/pull/6394) for more
+  details about concurrent program in fluid.
 
 ### Discussion
 
-- In the "Aync SGD" figure, the "W" variable on the parameter server
-  could be read and wrote concurrently, what is our locking strategy?
-  E.g., each variable have a lock cpp method to be invoked by every
-  OP, or, have a lock OP.
-
 - Can the Enqueue OP be implemented under our current tensor design
   (puts the input tensor into the queue tensor)?
-
 - *Dequeue* OP will have variable numbers of output (depends on the
   `min_count` attribute), does our current design support it? (similar
   question for the *Add* OP)