Multi-dimensional Hybrid Parallel Case Based on Double Recursive Search

Overview

Multi-dimensional hybrid parallel based on double recursive search means that the user can configure optimization methods such as recomputation, optimizer parallel, pipeline parallel. Based on the user configurations, the operator-level strategy is automatically searched by the double recursive strategy search algorithm, which generates the optimal parallel strategy.

Operation Practice

The following is a multi-dimensional hybrid parallel case based on double recursive search using Ascend or GPU single-machine 8-card as an example:

Example Code Description

Download the complete example code: multiple_mix.

The directory structure is as follows:

└─ sample_code
    ├─ multiple_mix
       ├── sapp_mix_train.py
       └── run_sapp_mix_train.sh
    ...

sapp_mix_train.py is the script that defines the network structure and the training process. run_sapp_mix_train.sh is the execution script.

Configuring Distributed Environment

Specify the run mode, run device, run card number, etc. through the context interface. Unlike single-card scripts, parallel scripts also need to specify the parallel mode parallel_mode as auto-parallel and the search mode search_mode as double recursive strategy search mode recursive_programming for auto-slicing of the data parallel and model parallel, and initialize HCCL or NCCL communication with init. pipeline_stages is the number of stages in pipeline parallel, and optimizer parallel is enabled by enabling enable_parallel_optimizer. device_target is automatically specified as the backend hardware device corresponding to the MindSpore package.

import mindspore as ms
from mindspore.communication import init

ms.set_context(mode=ms.GRAPH_MODE, save_graphs=2)
ms.set_context(max_device_memory="25GB")
ms.set_auto_parallel_context(parallel_mode=ms.ParallelMode.AUTO_PARALLEL, search_mode="recursive_programming")
ms.set_auto_parallel_context(pipeline_stages=2, enable_parallel_optimizer=True)
init()
ms.set_seed(1)

Network Definition

The network definition adds recomputation, pipeline parallel to the data parallel and model parallel provided by the double recursive strategy search algorithm:

from mindspore import nn

class Network(nn.Cell):
    def __init__(self):
        super().__init__()
        self.flatten = nn.Flatten()
        self.layer1 = nn.Dense(28*28, 512)
        self.relu1 = nn.ReLU()
        self.layer2 = nn.Dense(512, 512)
        self.relu2 = nn.ReLU()
        self.layer3 = nn.Dense(512, 1)

    def construct(self, x):
        x = self.flatten(x)
        x = self.layer1(x)
        x = self.relu1(x)
        x = self.layer2(x)
        x = self.relu2(x)
        logits = self.layer3(x)
        return logits

net = Network()
# Configure the pipeline_stage number for each layer in pipeline parallel
net.layer1.pipeline_stage = 0
net.relu1.pipeline_stage = 0
net.layer2.pipeline_stage = 1
net.relu2.pipeline_stage = 1
net.layer3.pipeline_stage = 1
# Configure recomputation of relu operators
net.relu1.recompute()
net.relu2.recompute()

Loading the Datasets

The dataset is loaded in the same way as the single-card model, with the following code:

import os
import mindspore.dataset as ds

def create_dataset(batch_size):
    dataset_path = os.getenv("DATA_PATH")
    dataset = ds.MnistDataset(dataset_path)
    image_transforms = [
        ds.vision.Rescale(1.0 / 255.0, 0),
        ds.vision.Normalize(mean=(0.1307,), std=(0.3081,)),
        ds.vision.HWC2CHW()
    ]
    label_transform = ds.transforms.TypeCast(ms.int32)
    dataset = dataset.map(image_transforms, 'image')
    dataset = dataset.map(label_transform, 'label')
    dataset = dataset.batch(batch_size)
    return dataset

data_set = create_dataset(32)

Training the Network

This part is consistent with the pipeline parallel training code. Two additional interfaces need to be called based on the stand-alone training code: nn.WithLossCell for wrapping the network and loss function, and nn.PipelineCell for wrapping the LossCell and configuring the MicroBatch size. The code is as follows:

import mindspore as ms
from mindspore import nn, train

optimizer = nn.SGD(net.trainable_params(), 1e-2)
loss_fn = nn.MAELoss()
loss_cb = train.LossMonitor()
net_with_grads = nn.PipelineCell(nn.WithLossCell(net, loss_fn), 4)
model = ms.Model(net_with_grads, optimizer=optimizer)
model.train(10, data_set, callbacks=[loss_cb], dataset_sink_mode=True)

Running a Stand-alone Eight-Card Script

Next, the corresponding scripts are invoked by commands, using the mpirun startup method and the 8-card distributed training script as an example of distributed training:

bash run_sapp_mix_train.sh

The results are saved in log_output/1/rank.*/stdout, and the example is as follows:

epoch: 1 step: 1875, loss is 11.6961808800697327
epoch: 2 step: 1875, loss is 10.2737872302532196
epoch: 3 step: 1875, loss is 8.87508840560913086
epoch: 4 step: 1875, loss is 8.1057268142700195