您现在的位置是：首页 >技术交流 >pytorch rpc如何实现分物理机器实现model parallel网站首页技术交流

因为业务需要，最近接到一项任务，是如何利用pytorch实现model parallel以及distributed training。搜罗了网上很多资料，以及阅读了pytorch官方的教程，都没有可参考的案例。讲的比较多的是data parallel，关于model parallel的研究发现不多。
通过阅读pytorch官方主页，发现这个example是进行model parallel的，
官方博客地址：DISTRIBUTED PIPELINE PARALLELISM USING RPC
官方的example地址：Distributed Pipeline Parallel Example
通过阅读代码发现，这个代码以Resnet 50 model为例，将model直接拆分成两部分，并指定两部分在不同的worker运行，代码实现了在同一台机器上，创建多进程来拆分模型运行。关于这个代码的详细介绍可搜索关键词：pytorch RPC 的分布式管道并行，这里不多介绍。
通过在本地运行代码发现，不满足多机器运行的需求。接下来是思考的心路里程。

1. 首先通过代码发现，python main.py程序运行时，无法指定rank，那么在跨机器运行时如何知道哪台机器是worker1，worker2？这个地方，我们首先怀疑需要去修改worker，人为在代码中指定worker的IP地址，如修改main.py 代码中191行
   修改前：
   model = DistResNet50(split_size, ["worker1", "worker2"])
   修改后：
   model = DistResNet50(split_size, ["worker1@xxx.xxx.xxx.xxx", "worker2@xxx.xxx.xxx.xxx"])
   然后，很自然的就报错了，这里无法识别这样的worker名，不支持直接指定，这条路也就走不通了。
2. 接着只能重新阅读代码，到最后251行，我们发现
   mp.spawn(run_worker, args=(world_size, num_split), nprocs=world_size, join=True)
   尤其是这行代码中mp.spawn引起了我们的怀疑，这不是多进程么，这本质还是在多进程情况下来执行程序，无法跨机器啊，不符合我们的需求。
3. 最后的最后，我们重新阅读pytorch rpc机制，并通过简单测试程序，让两台机器互相通信，其中一台机器发起运算请求并传输原始数据，另外一台机器负责接收数据并进行相关运算，这个程序当时在两台物理机器上测试成功了，那说明rpc实现通信这件事并不复杂。结合前面给的代码，我们决定将worke1和worker2分开写代码，分开执行，并且在代码中需要指定这些worker所属的rank，这样理论上就能够将原始代码修改成分机器的rpc通信运行了。
   上面主要是我们的心理历程，话不多说，接下来show the code。
   实验环境，两台机器，均是cpu环境，conda安装的环境也保证了一致。
   master机器代码：

# https://github.com/pytorch/examples/blob/main/distributed/rpc/pipeline/main.py

import os
import threading
import time
import torch
import torch.nn as nn
import torch.distributed.autograd as dist_autograd
import torch.distributed.rpc as rpc
import torch.optim as optim
from torch.distributed.optim import DistributedOptimizer
from torch.distributed.rpc import RRef

from torchvision.models.resnet import Bottleneck
os.environ['MASTER_ADDR'] = 'XXX.XXX.XXX.XXX' # 指定master ip地址
os.environ['MASTER_PORT'] = '7856' # 指定master 端口号

#########################################################
#           Define Model Parallel ResNet50              #
#########################################################

# In order to split the ResNet50 and place it on two different workers, we
# implement it in two model shards. The ResNetBase class defines common
# attributes and methods shared by two shards. ResNetShard1 and ResNetShard2
# contain two partitions of the model layers respectively.


num_classes = 1000


def conv1x1(in_planes, out_planes, stride=1):
    """1x1 convolution"""
    return nn.Conv2d(in_planes, out_planes, kernel_size=1, stride=stride, bias=False)

class ResNetBase(nn.Module):
    def __init__(self, block, inplanes, num_classes=1000,
                 groups=1, width_per_group=64, norm_layer=None):
        super(ResNetBase, self).__init__()

        self._lock = threading.Lock()
        self._block = block
        self._norm_layer = nn.BatchNorm2d
        self.inplanes = inplanes
        self.dilation = 1
        self.groups = groups
        self.base_width = width_per_group

    def _make_layer(self, planes, blocks, stride=1):
        norm_layer = self._norm_layer
        downsample = None
        previous_dilation = self.dilation
        if stride != 1 or self.inplanes != planes * self._block.expansion:
            downsample = nn.Sequential(
                conv1x1(self.inplanes, planes * self._block.expansion, stride),
                norm_layer(planes * self._block.expansion),
            )

        layers = []
        layers.append(self._block(self.inplanes, planes, stride, downsample, self.groups,
                                  self.base_width, previous_dilation, norm_layer))
        self.inplanes = planes * self._block.expansion
        for _ in range(1, blocks):
            layers.append(self._block(self.inplanes, planes, groups=self.groups,
                                      base_width=self.base_width, dilation=self.dilation,
                                      norm_layer=norm_layer))

        return nn.Sequential(*layers)

    def parameter_rrefs(self):
        r"""
        Create one RRef for each parameter in the given local module, and return a
        list of RRefs.
        """
        return [RRef(p) for p in self.parameters()]


class ResNetShard1(ResNetBase):
    """
    The first part of ResNet.
    """
    def __init__(self, device, *args, **kwargs):
        super(ResNetShard1, self).__init__(
            Bottleneck, 64, num_classes=num_classes, *args, **kwargs)

        self.device = device
        self.seq = nn.Sequential(
            nn.Conv2d(3, self.inplanes, kernel_size=7, stride=2, padding=3, bias=False),
            self._norm_layer(self.inplanes),
            nn.ReLU(inplace=True),
            nn.MaxPool2d(kernel_size=3, stride=2, padding=1),
            self._make_layer(64, 3),
            self._make_layer(128, 4, stride=2)
        ).to(self.device)

        for m in self.modules():
            if isinstance(m, nn.Conv2d):
                nn.init.kaiming_normal_(m.weight, mode='fan_out', nonlinearity='relu')
            elif isinstance(m, nn.BatchNorm2d):
                nn.init.ones_(m.weight)
                nn.init.zeros_(m.bias)

    def forward(self, x_rref):
        x = x_rref.to_here().to(self.device)
        with self._lock:
            out =  self.seq(x)
        return out.cpu()


class ResNetShard2(ResNetBase):
    """
    The second part of ResNet.
    """
    def __init__(self, device, *args, **kwargs):
        super(ResNetShard2, self).__init__(
            Bottleneck, 512, num_classes=num_classes, *args, **kwargs)

        self.device = device
        self.seq = nn.Sequential(
            self._make_layer(256, 6, stride=2),
            self._make_layer(512, 3, stride=2),
            nn.AdaptiveAvgPool2d((1, 1)),
        ).to(self.device)

        self.fc =  nn.Linear(512 * self._block.expansion, num_classes).to(self.device)

    def forward(self, x_rref):
        x = x_rref.to_here().to(self.device)
        with self._lock:
            out = self.fc(torch.flatten(self.seq(x), 1))
        return out.cpu()


class DistResNet50(nn.Module):
    """
    Assemble two parts as an nn.Module and define pipelining logic
    """
    def __init__(self, split_size, workers, *args, **kwargs):
        super(DistResNet50, self).__init__()

        self.split_size = split_size

        # Put the first part of the ResNet50 on workers[0]
        self.p1_rref = rpc.remote(
            workers[0],
            ResNetShard1,
            args = ("cuda:0",) + args,
            kwargs = kwargs
        )

        # Put the second part of the ResNet50 on workers[1]
        self.p2_rref = rpc.remote(
            workers[1],
            ResNetShard2,
            args = ("cpu",) + args,
            kwargs = kwargs
        )

    def forward(self, xs):
        # Split the input batch xs into micro-batches, and collect async RPC
        # futures into a list
        out_futures = []
        for x in iter(xs.split(self.split_size, dim=0)):
            x_rref = RRef(x)
            y_rref = self.p1_rref.remote().forward(x_rref)
            print(y_rref)
            z_fut = self.p2_rref.rpc_async().forward(y_rref)
            print(z_fut)
            out_futures.append(z_fut)

        # collect and cat all output tensors into one tensor.
        return torch.cat(torch.futures.wait_all(out_futures))

    def parameter_rrefs(self):
        remote_params = []
        remote_params.extend(self.p1_rref.remote().parameter_rrefs().to_here())
        remote_params.extend(self.p2_rref.remote().parameter_rrefs().to_here())
        return remote_params


#########################################################
#                   Run RPC Processes                   #
#########################################################

num_batches = 3
batch_size = 8
image_w = 128
image_h = 128

if __name__=="__main__":
    options = rpc.TensorPipeRpcBackendOptions(num_worker_threads=256, rpc_timeout=300)
    # 初始化主节点的RPC连接
    rpc.init_rpc("master", rank=0, world_size=2, rpc_backend_options=options)

    for num_split in [1,2]:
        tik = time.time()

        model = DistResNet50(num_split, ["master", "worker"])
        loss_fn = nn.MSELoss()
        opt = DistributedOptimizer(
            optim.SGD,
            model.parameter_rrefs(),
            lr=0.05,
        )
        one_hot_indices = torch.LongTensor(batch_size) 
            .random_(0, num_classes) 
            .view(batch_size, 1)

        for i in range(num_batches):
            print(f"Processing batch {i}")
            # generate random inputs and labels
            inputs = torch.randn(batch_size, 3, image_w, image_h)
            labels = torch.zeros(batch_size, num_classes) 
                .scatter_(1, one_hot_indices, 1)

            with dist_autograd.context() as context_id:
                outputs = model(inputs)
                dist_autograd.backward(context_id, [loss_fn(outputs, labels)])
                opt.step(context_id)

        tok = time.time()
        print(f"number of splits = {num_split}, execution time = {tok - tik}")

    # 关闭RPC连接
    rpc.shutdown()

worker端的代码

# https://github.com/pytorch/examples/blob/main/distributed/rpc/pipeline/main.py

import os
import threading
import time
from functools import wraps

import torch
import torch.nn as nn
import torch.distributed.rpc as rpc
from torch.distributed.rpc import RRef

from torchvision.models.resnet import Bottleneck
os.environ['MASTER_ADDR'] = 'XXX.XXX.XXX.XXX' # 指定master 端口号
os.environ['MASTER_PORT'] = '7856' # 指定master 端口号

#########################################################
#           Define Model Parallel ResNet50              #
#########################################################

# In order to split the ResNet50 and place it on two different workers, we
# implement it in two model shards. The ResNetBase class defines common
# attributes and methods shared by two shards. ResNetShard1 and ResNetShard2
# contain two partitions of the model layers respectively.


num_classes = 1000


def conv1x1(in_planes, out_planes, stride=1):
    """1x1 convolution"""
    return nn.Conv2d(in_planes, out_planes, kernel_size=1, stride=stride, bias=False)

class ResNetBase(nn.Module):
    def __init__(self, block, inplanes, num_classes=1000,
                 groups=1, width_per_group=64, norm_layer=None):
        super(ResNetBase, self).__init__()

        self._lock = threading.Lock()
        self._block = block
        self._norm_layer = nn.BatchNorm2d
        self.inplanes = inplanes
        self.dilation = 1
        self.groups = groups
        self.base_width = width_per_group

    def _make_layer(self, planes, blocks, stride=1):
        norm_layer = self._norm_layer
        downsample = None
        previous_dilation = self.dilation
        if stride != 1 or self.inplanes != planes * self._block.expansion:
            downsample = nn.Sequential(
                conv1x1(self.inplanes, planes * self._block.expansion, stride),
                norm_layer(planes * self._block.expansion),
            )

        layers = []
        layers.append(self._block(self.inplanes, planes, stride, downsample, self.groups,
                                  self.base_width, previous_dilation, norm_layer))
        self.inplanes = planes * self._block.expansion
        for _ in range(1, blocks):
            layers.append(self._block(self.inplanes, planes, groups=self.groups,
                                      base_width=self.base_width, dilation=self.dilation,
                                      norm_layer=norm_layer))

        return nn.Sequential(*layers)

    def parameter_rrefs(self):
        r"""
        Create one RRef for each parameter in the given local module, and return a
        list of RRefs.
        """
        return [RRef(p) for p in self.parameters()]

class ResNetShard2(ResNetBase):
    """
    The second part of ResNet.
    """
    def __init__(self, device, *args, **kwargs):
        super(ResNetShard2, self).__init__(
            Bottleneck, 512, num_classes=num_classes, *args, **kwargs)

        self.device = device
        self.seq = nn.Sequential(
            self._make_layer(256, 6, stride=2),
            self._make_layer(512, 3, stride=2),
            nn.AdaptiveAvgPool2d((1, 1)),
        ).to(self.device)

        self.fc =  nn.Linear(512 * self._block.expansion, num_classes).to(self.device)

    def forward(self, x_rref):
        x = x_rref.to_here().to(self.device)
        print(x)
        with self._lock:
            out = self.fc(torch.flatten(self.seq(x), 1))
        return out.cpu()


#########################################################
#                   Run RPC Processes                   #
#########################################################



if __name__=="__main__":
    options = rpc.TensorPipeRpcBackendOptions(num_worker_threads=256, rpc_timeout=300)
    # 初始化工作节点的RPC连接
    rpc.init_rpc("worker", rank=1, world_size=2, rpc_backend_options=options)

    # 等待主节点的调用
    rpc.shutdown()

代码中的MASTER_ADDR和port需要指定为一致，分别在master机器上运行master.py，worker机器上运行worker.py，这样就可以实现Resnet 50 model在两台物理机器上model parallel。
注意事项

• 确保物理机器能够互相ping通，同时关闭防火墙
• 两个物理机器最好都是linux环境，我们的实验发现pytorch的分布式不支持在Windows环境运行
• 两个物理机器的python运行环境要求保持一致