2022年5月,PyTorch官方宣布已正式支持在M1芯片版本的Mac上进行模型加速。官方对比数据显示,和CPU相比,M1上炼丹速度平均可加速7倍。
哇哦,不用单独配个GPU也能加速这么多,我迫不及待地搞到一个M1芯片的MacBook后试水了一番,并把我认为相关重要的信息梳理成了本文。
一,加速原理
-
Question1,Mac M1芯片 为什么可以用来加速 pytorch?
因为 Mac M1芯片不是一个单纯的一个CPU芯片,而是包括了CPU(中央处理器),GPU(图形处理器),NPU(神经网络引擎),以及统一内存单元等众多组件的一块集成芯片。由于Mac M1芯片集成了GPU组件,所以可以用来加速pytorch.
-
Question2,Mac M1芯片 上GPU的的显存有多大?
Mac M1芯片的CPU和GPU使用统一的内存单元。所以Mac M1芯片的能使用的显存大小就是 Mac 电脑的内存大小。
-
Question3,使用Mac M1芯片加速 pytorch 需要安装 cuda后端吗?
不需要,cuda是适配nvidia的GPU的,Mac M1芯片中的GPU适配的加速后端是mps,在Mac对应操作系统中已经具备,无需单独安装。只需要安装适配的pytorch即可。
-
Question4,为什么有些可以在Mac Intel芯片电脑安装的软件不能在Mac M1芯片电脑上安装?
Mac M1芯片为了追求高性能和节能,在底层设计上使用的是一种叫做arm架构的精简指令集,不同于Intel等常用CPU芯片采用的x86架构完整指令集。所以有些基于x86指令集开发的软件不能直接在Mac M1芯片电脑上使用。
二,环境配置
0,检查mac型号
点击桌面左上角mac图标——>关于本机——>概览,确定是m1芯片,了解内存大小(最好有16G以上,8G可能不太够用)。
1,下载 miniforge3 (miniforge3可以理解成 miniconda/annoconda 的社区版,提供了更稳定的对M1芯片的支持)
https://github.com/conda-forge/miniforge/#download
备注: annoconda 在 2022年5月开始也发布了对 mac m1芯片的官方支持,但还是推荐社区发布的miniforge3,开源且更加稳定。
2,安装 miniforge3
chmod +x ~/Downloads/Miniforge3-MacOSX-arm64.sh
sh ~/Downloads/Miniforge3-MacOSX-arm64.sh
source ~/miniforge3/bin/activate
3,安装 pytorch (v1.12版本已经正式支持了用于mac m1芯片gpu加速的mps后端。)
pip install torch>=1.12 -i https://pypi.tuna.tsinghua.edu.cn/simple
4,测试环境
import torch
print(torch.backends.mps.is_available())
print(torch.backends.mps.is_built())
如果输出都是True的话,那么恭喜你配置成功了。
三,范例代码
下面以mnist手写数字识别为例,演示使用mac M1芯片GPU的mps后端来加速pytorch的完整流程。
核心操作非常简单,和使用cuda类似,训练前把模型和数据都移动到torch.device("mps")就可以了。
import torch
from torch import nn
import torchvision
from torchvision import transforms
import torch.nn.functional as F
import os,sys,time
import numpy as np
import pandas as pd
import datetime
from tqdm import tqdm
from copy import deepcopy
from torchmetrics import Accuracy
def printlog(info):
nowtime = datetime.datetime.now().strftime('%Y-%m-%d %H:%M:%S')
print("\n"+"=========="*8 + "%s"%nowtime)
print(str(info)+"\n")
#================================================================================
# 一,准备数据
#================================================================================
transform = transforms.Compose([transforms.ToTensor()])
ds_train = torchvision.datasets.MNIST(root="mnist/",train=True,download=True,transform=transform)
ds_val = torchvision.datasets.MNIST(root="mnist/",train=False,download=True,transform=transform)
dl_train = torch.utils.data.DataLoader(ds_train, batch_size=128, shuffle=True, num_workers=2)
dl_val = torch.utils.data.DataLoader(ds_val, batch_size=128, shuffle=False, num_workers=2)
#================================================================================
# 二,定义模型
#================================================================================
def create_net():
net = nn.Sequential()
net.add_module("conv1",nn.Conv2d(in_channels=1,out_channels=64,kernel_size = 3))
net.add_module("pool1",nn.MaxPool2d(kernel_size = 2,stride = 2))
net.add_module("conv2",nn.Conv2d(in_channels=64,out_channels=512,kernel_size = 3))
net.add_module("pool2",nn.MaxPool2d(kernel_size = 2,stride = 2))
net.add_module("dropout",nn.Dropout2d(p = 0.1))
net.add_module("adaptive_pool",nn.AdaptiveMaxPool2d((1,1)))
net.add_module("flatten",nn.Flatten())
net.add_module("linear1",nn.Linear(512,1024))
net.add_module("relu",nn.ReLU())
net.add_module("linear2",nn.Linear(1024,10))
return net
net = create_net()
print(net)
# 评估指标
class Accuracy(nn.Module):
def __init__(self):
super().__init__()
self.correct = nn.Parameter(torch.tensor(0.0),requires_grad=False)
self.total = nn.Parameter(torch.tensor(0.0),requires_grad=False)
def forward(self, preds: torch.Tensor, targets: torch.Tensor):
preds = preds.argmax(dim=-1)
m = (preds == targets).sum()
n = targets.shape[0]
self.correct += m
self.total += n
return m/n
def compute(self):
return self.correct.float() / self.total
def reset(self):
self.correct -= self.correct
self.total -= self.total
#================================================================================
# 三,训练模型
#================================================================================
loss_fn = nn.CrossEntropyLoss()
optimizer= torch.optim.Adam(net.parameters(),lr = 0.01)
metrics_dict = nn.ModuleDict({"acc":Accuracy()})
# =========================移动模型到mps上==============================
device = torch.device("mps" if torch.backends.mps.is_available() else "cpu")
net.to(device)
loss_fn.to(device)
metrics_dict.to(device)
# ====================================================================
epochs = 20
ckpt_path='checkpoint.pt'
#early_stopping相关设置
monitor="val_acc"
patience=5
mode="max"
history = {}
for epoch in range(1, epochs+1):
printlog("Epoch {0} / {1}".format(epoch, epochs))
# 1,train -------------------------------------------------
net.train()
total_loss,step = 0,0
loop = tqdm(enumerate(dl_train), total =len(dl_train),ncols=100)
train_metrics_dict = deepcopy(metrics_dict)
for i, batch in loop:
features,labels = batch
# =========================移动数据到mps上==============================
features = features.to(device)
labels = labels.to(device)
# ====================================================================
#forward
preds = net(features)
loss = loss_fn(preds,labels)
#backward
loss.backward()
optimizer.step()
optimizer.zero_grad()
#metrics
step_metrics = {"train_"+name:metric_fn(preds, labels).item()
for name,metric_fn in train_metrics_dict.items()}
step_log = dict({"train_loss":loss.item()},**step_metrics)
total_loss += loss.item()
step+=1
if i!=len(dl_train)-1:
loop.set_postfix(**step_log)
else:
epoch_loss = total_loss/step
epoch_metrics = {"train_"+name:metric_fn.compute().item()
for name,metric_fn in train_metrics_dict.items()}
epoch_log = dict({"train_loss":epoch_loss},**epoch_metrics)
loop.set_postfix(**epoch_log)
for name,metric_fn in train_metrics_dict.items():
metric_fn.reset()
for name, metric in epoch_log.items():
history[name] = history.get(name, []) + [metric]
# 2,validate -------------------------------------------------
net.eval()
total_loss,step = 0,0
loop = tqdm(enumerate(dl_val), total =len(dl_val),ncols=100)
val_metrics_dict = deepcopy(metrics_dict)
with torch.no_grad():
for i, batch in loop:
features,labels = batch
# =========================移动数据到mps上==============================
features = features.to(device)
labels = labels.to(device)
# ====================================================================
#forward
preds = net(features)
loss = loss_fn(preds,labels)
#metrics
step_metrics = {"val_"+name:metric_fn(preds, labels).item()
for name,metric_fn in val_metrics_dict.items()}
step_log = dict({"val_loss":loss.item()},**step_metrics)
total_loss += loss.item()
step+=1
if i!=len(dl_val)-1:
loop.set_postfix(**step_log)
else:
epoch_loss = (total_loss/step)
epoch_metrics = {"val_"+name:metric_fn.compute().item()
for name,metric_fn in val_metrics_dict.items()}
epoch_log = dict({"val_loss":epoch_loss},**epoch_metrics)
loop.set_postfix(**epoch_log)
for name,metric_fn in val_metrics_dict.items():
metric_fn.reset()
epoch_log["epoch"] = epoch
for name, metric in epoch_log.items():
history[name] = history.get(name, []) + [metric]
# 3,early-stopping -------------------------------------------------
arr_scores = history[monitor]
best_score_idx = np.argmax(arr_scores) if mode=="max" else np.argmin(arr_scores)
if best_score_idx==len(arr_scores)-1:
torch.save(net.state_dict(),ckpt_path)
print("<<<<<< reach best {0} : {1} >>>>>>".format(monitor,
arr_scores[best_score_idx]),file=sys.stderr)
if len(arr_scores)-best_score_idx>patience:
print("<<<<<< {} without improvement in {} epoch, early stopping >>>>>>".format(
monitor,patience),file=sys.stderr)
break
net.load_state_dict(torch.load(ckpt_path))
dfhistory = pd.DataFrame(history)
四,使用torchkeras支持Mac M1芯片加速
我在最新的3.3.0的torchkeras版本中引入了对 mac m1芯片的支持,当存在可用的 mac m1芯片/ GPU 时,会默认使用它们进行加速,无需做任何配置。
使用范例如下。😋😋😋
!pip install torchkeras>=3.3.0
import numpy as np
import pandas as pd
from matplotlib import pyplot as plt
import torch
from torch import nn
import torch.nn.functional as F
from torch.utils.data import Dataset,DataLoader
import torchkeras #Attention this line
#================================================================================
# 一,准备数据
#================================================================================
import torchvision
from torchvision import transforms
transform = transforms.Compose([transforms.ToTensor()])
ds_train = torchvision.datasets.MNIST(root="mnist/",train=True,download=True,transform=transform)
ds_val = torchvision.datasets.MNIST(root="mnist/",train=False,download=True,transform=transform)
dl_train = torch.utils.data.DataLoader(ds_train, batch_size=128, shuffle=True, num_workers=2)
dl_val = torch.utils.data.DataLoader(ds_val, batch_size=128, shuffle=False, num_workers=2)
for features,labels in dl_train:
break
#================================================================================
# 二,定义模型
#================================================================================
def create_net():
net = nn.Sequential()
net.add_module("conv1",nn.Conv2d(in_channels=1,out_channels=64,kernel_size = 3))
net.add_module("pool1",nn.MaxPool2d(kernel_size = 2,stride = 2))
net.add_module("conv2",nn.Conv2d(in_channels=64,out_channels=512,kernel_size = 3))
net.add_module("pool2",nn.MaxPool2d(kernel_size = 2,stride = 2))
net.add_module("dropout",nn.Dropout2d(p = 0.1))
net.add_module("adaptive_pool",nn.AdaptiveMaxPool2d((1,1)))
net.add_module("flatten",nn.Flatten())
net.add_module("linear1",nn.Linear(512,1024))
net.add_module("relu",nn.ReLU())
net.add_module("linear2",nn.Linear(1024,10))
return net
net = create_net()
print(net)
# 评估指标
class Accuracy(nn.Module):
def __init__(self):
super().__init__()
self.correct = nn.Parameter(torch.tensor(0.0),requires_grad=False)
self.total = nn.Parameter(torch.tensor(0.0),requires_grad=False)
def forward(self, preds: torch.Tensor, targets: torch.Tensor):
preds = preds.argmax(dim=-1)
m = (preds == targets).sum()
n = targets.shape[0]
self.correct += m
self.total += n
return m/n
def compute(self):
return self.correct.float() / self.total
def reset(self):
self.correct -= self.correct
self.total -= self.total
#================================================================================
# 三,训练模型
#================================================================================
model = torchkeras.KerasModel(net,
loss_fn = nn.CrossEntropyLoss(),
optimizer= torch.optim.Adam(net.parameters(),lr=0.001),
metrics_dict = {"acc":Accuracy()}
)
from torchkeras import summary
summary(model,input_data=features);
# if gpu/mps is available, will auto use it, otherwise cpu will be used.
dfhistory=model.fit(train_data=dl_train,
val_data=dl_val,
epochs=15,
patience=5,
monitor="val_acc",mode="max",
ckpt_path='checkpoint.pt')
#================================================================================
# 四,评估模型
#================================================================================
model.evaluate(dl_val)
#================================================================================
# 五,使用模型
#================================================================================
model.predict(dl_val)[0:10]
#================================================================================
# 六,保存模型
#================================================================================
# The best net parameters has been saved at ckpt_path='checkpoint.pt' during training.
net_clone = create_net()
net_clone.load_state_dict(torch.load("checkpoint.pt"))
五,M1芯片与CPU和Nvidia GPU速度对比
使用以上代码作为范例,分别在CPU, mac m1芯片,以及Nvidia GPU上 运行。
得到的运行速度截图如下:
纯CPU跑效果
Mac M1 芯片加速效果
Tesla P100 GPU加速效果
纯CPU跑一个epoch大约是3min 18s。
使用mac m1芯片加速,一个epoch大约是33 s,相比CPU跑,加速约6倍。
这和pytorch官网显示的训练过程平均加速7倍相当。
使用Nvidia Tesla P100 GPU加速,一个epoch大约是 8s,相比CPU跑,加速约25倍。
整体来说Mac M1芯片对 深度学习训练过程的加速还是非常显著的,通常达到5到7倍左右。
不过目前看和企业中最常使用的高端的Tesla P100 GPU相比,还是有2到4倍的训练速度差异,可以视做一个mini版的GPU吧。
因此Mac M1芯片比较适合本地训练一些中小规模的模型,快速迭代idea,使用起来还是蛮香的。
尤其是本来就打算想换个电脑的,用mac做开发本来比windows好使多了。