多 GPU 训练比 Tensorflow 上的单 GPU 慢

Question

我创建了 3 个虚拟 GPU（有 1 个 GPU）并尝试加速图像的矢量化。但是，使用下面提供的代码并从文档（这里）中手动放置我得到了奇怪的结果：在所有 GPU 上的训练比在单个 GPU 上慢两倍。还要在具有 3 个物理 GPU 的机器上检查此代码（并删除虚拟设备初始化） - 工作方式相同。

环境：Python 3.6、Ubuntu 18.04.3、tensorflow-gpu 1.14.0。

代码（此示例创建 3 个虚拟设备，您可以在具有一个 GPU 的 PC 上对其进行测试）：

import os
import time
import numpy as np
import tensorflow as tf

start = time.time()

def load_graph(frozen_graph_filename):
    # We load the protobuf file from the disk and parse it to retrieve the
    # unserialized graph_def
    with tf.gfile.GFile(frozen_graph_filename, "rb") as f:
        graph_def = tf.GraphDef()
        graph_def.ParseFromString(f.read())

    # Then, we import the graph_def into a new Graph and returns it
    with tf.Graph().as_default() as graph:
        # The name var will prefix every op/nodes in your graph
        # Since we load everything in a new graph, this is not needed
        tf.import_graph_def(graph_def, name="")
    return graph

path_to_graph = '/imagenet/'  # Path to imagenet folder where graph file is placed
GRAPH = load_graph(os.path.join(path_to_graph, 'classify_image_graph_def.pb'))

# Create Session
config = tf.ConfigProto()
config.gpu_options.per_process_gpu_memory_fraction = 0.9
config.gpu_options.allow_growth = True
session = tf.Session(graph=GRAPH, config=config)

output_dir = '/vectors/'  # where to saved vectors from images

# Single GPU vectorization
for image_index, image in enumerate(selected_list):
    with Image.open(image) as f:
        image_data = f.convert('RGB')
        feature_tensor = session.graph.get_tensor_by_name('pool_3:0')
        feature_vector = session.run(feature_tensor, {'DecodeJpeg:0': image_data})
        feature_vector = np.squeeze(feature_vector)
        outfile_name = os.path.basename(image) + ".vc"
        out_path = os.path.join(output_dir, outfile_name)
        # Save vector
        np.savetxt(out_path, feature_vector, delimiter=',')

print(f"Single GPU: {time.time() - start}")
start = time.time()

print("Start calculation on multiple GPU")
gpus = tf.config.experimental.list_physical_devices('GPU')
if gpus:
  # Create 3 virtual GPUs with 1GB memory each
  try:
    tf.config.experimental.set_virtual_device_configuration(
        gpus[0],
        [tf.config.experimental.VirtualDeviceConfiguration(memory_limit=1024),
         tf.config.experimental.VirtualDeviceConfiguration(memory_limit=1024),
         tf.config.experimental.VirtualDeviceConfiguration(memory_limit=1024)])
    logical_gpus = tf.config.experimental.list_logical_devices('GPU')
    print(len(gpus), "Physical GPU,", len(logical_gpus), "Logical GPUs")
  except RuntimeError as e:
    # Virtual devices must be set before GPUs have been initialized
    print(e)

print("Create prepared ops")
start1 = time.time()
gpus = logical_gpus  # comment this line to use physical GPU devices for calculations

image_list = ['1.jpg', '2.jpg', '3.jpg']  # list with images to vectorize (tested on 100 and 1000 examples)
# Assign chunk of list to each GPU
# image_list1, image_list2, image_list3 = image_list[:len(image_list)],\
#                                         image_list[len(image_list):2*len(image_list)],\
#                                         image_list[2*len(image_list):]
selected_list = image_list # commit this line if you want to try to assign chunk of list manually to each GPU
output_vectors = []
if gpus:
  # Replicate your computation on multiple GPUs
  feature_vectors = []
  for gpu in gpus:  # iterating on a virtual GPU devices, not physical
    with tf.device(gpu.name):
      print(f"Assign list of images to {gpu.name.split(':', 4)[-1]}")
      # Try to assign chunk of list with images to each GPU - work the same time as single GPU
      # if gpu.name.split(':', 4)[-1] == "GPU:0":
      #     selected_list = image_list1
      # if gpu.name.split(':', 4)[-1] == "GPU:1":
      #     selected_list = image_list2
      # if gpu.name.split(':', 4)[-1] == "GPU:2":
      #     selected_list = image_list3
      for image_index, image in enumerate(selected_list):
          with Image.open(image) as f:
            image_data = f.convert('RGB')
            feature_tensor = session.graph.get_tensor_by_name('pool_3:0')
            feature_vector = session.run(feature_tensor, {'DecodeJpeg:0': image_data})
            feature_vectors.append(feature_vector)

print("All images has been assigned to GPU's")
print(f"Time spend on prep ops: {time.time() - start1}")
print("Start calculation on multiple GPU")
start1 = time.time()
for image_index, image in enumerate(image_list):
  feature_vector = np.squeeze(feature_vectors[image_index])
  outfile_name = os.path.basename(image) + ".vc"
  out_path = os.path.join(output_dir, outfile_name)
  # Save vector
  np.savetxt(out_path, feature_vector, delimiter=',')

# Close session
session.close()
print(f"Calc on GPU's spend: {time.time() - start1}")
print(f"All time, spend on multiple GPU: {time.time() - start}")

提供输出视图（来自包含 100 个图像的列表）：

1 Physical GPU, 3 Logical GPUs
Single GPU: 18.76301646232605
Start calculation on multiple GPU
Create prepared ops
Assign list of images to GPU:0
Assign list of images to GPU:1
Assign list of images to GPU:2
All images has been assigned to GPU's
Time spend on prep ops: 18.263537883758545
Start calculation on multiple GPU
Calc on GPU's spend: 11.697082042694092
All time, spend on multiple GPU: 29.960679531097412

我尝试过的：将带有图像的列表分成 3 个块并将每个块分配给 GPU（请参阅提交的代码行）。这将多 GPU 时间减少到 17 秒，这比单 GPU 运行 18 秒（~5%）快一点。

预期结果：MultiGPU 版本比 singleGPU 版本更快（至少 1.5 倍加速）。

想法，为什么会发生：我以错误的方式编写了计算

Answer 1

有两个基本的误解会导致您遇到麻烦：

1. with tf.device(...):适用于在范围内创建的图节点，而不是Session.run调用。

2. Session.run是阻塞调用。它们不是并行运行的。TensorFlow 只能并行化单个Session.run.

现代 TF (>= 2.0) 可以使这更容易。

主要是你可以停止使用tf.Session和tf.Graph。使用@tf.function相反，我相信这个基本结构将工作：

@tf.function
def my_function(inputs, gpus, model):
  results = []
  for input, gpu in zip(inputs, gpus):
    with tf.device(gpu):
      results.append(model(input))    
  return results

但是你会想要尝试一个更现实的测试。仅使用 3 张图像，您根本无法衡量实际性能。

另请注意：

1. 该tf.distribute.Strategy班可以帮助简化一些这方面，通过分离从设备规格@tf.function多数民众赞成正在运行。strategy.experimental_run_v2(my_function, args=(dataset_inputs,))
2. 使用tf.data.Dataset输入管道将帮助您将加载/预处理与模型执行重叠。

但是，如果您真的打算使用它来执行此操作，tf.Graph并且tf.Session我认为您基本上需要从此重新组织您的代码：

# Your code:
# Builds a graph
graph = build_graph()

for gpu in gpus():
  with tf.device(gpu):
    # Calls `gpu` in each device scope.
    session.run(...)

对此：

g = tf.Graph()
with g.as_default():
  results = []
  for gpu in gpus:
    # Build the graph, on each device
    input = iterator.get_next()
    with tf.device(gpu):    
      results.append(my_function(input))       

# Use a single `Session.run` call
np_result = session.run(results, feed_dict={inputs: my_inputs})