tensorflow:Multiple GPUs
来源:互联网 发布:网络作家的故事 编辑:程序博客网 时间:2024/05/30 04:29
深度学习theano/tensorflow多显卡多人使用问题集
tensorflow中使用指定的GPU及GPU显存
Using GPUs
petewarden/tensorflow_makefile
tf_gpu_manager/manager.py
多GPU运行Deep Learning 和 并行Deep Learning(待续)
Multiple GPUs
1. 终端执行程序时设置使用的GPU
如果电脑有多个GPU,tensorflow默认全部使用。如果想只使用部分GPU,可以设置CUDA_VISIBLE_DEVICES。在调用python程序时,可以使用
CUDA_VISIBLE_DEVICES=1 python my_script.py #只使用GPU1CUDA_VISIBLE_DEVICES=0,1 python my_script.py #使用GPU0,GPU1
Environment Variable Syntax ResultsCUDA_VISIBLE_DEVICES=1 Only device 1 will be seenCUDA_VISIBLE_DEVICES=0,1 Devices 0 and 1 will be visibleCUDA_VISIBLE_DEVICES="0,1" Same as above, quotation marks are optionalCUDA_VISIBLE_DEVICES=0,2,3 Devices 0, 2, 3 will be visible; device 1 is maskedCUDA_VISIBLE_DEVICES="" No GPU will be visible
2. python代码中设置使用的GPU
如果要在python代码中设置使用的GPU,可以使用下面的代码
import osos.environ["CUDA_VISIBLE_DEVICES"] = "2"
3. 设置tensorflow使用的显存大小
定量设置显存
默认tensorflow是使用GPU尽可能多的显存。可以通过下面的方式,来设置使用的GPU显存:
gpu_options = tf.GPUOptions(per_process_gpu_memory_fraction=0.7)sess = tf.Session(config=tf.ConfigProto(gpu_options=gpu_options))
上面分配给tensorflow的GPU显存大小为:GPU实际显存*0.7。
可以按照需要,设置不同的值,来分配显存。
按需设置显存
上面的只能设置固定的大小。如果想按需分配,可以使用allow_growth参数(参考网址:http://blog.csdn.net/cq361106306/article/details/52950081):
gpu_options = tf.GPUOptions(allow_growth=True)sess = tf.Session(config=tf.ConfigProto(gpu_options=gpu_options))
4. 使用多个 GPU
如果你想让 TensorFlow 在多个 GPU 上运行, 你可以建立 multi-tower 结构, 在这个结构 里每个 tower 分别被指配给不同的 GPU 运行. 比如:
# 新建一个 graph.c = []for d in ['/gpu:2', '/gpu:3']: with tf.device(d): a = tf.constant([1.0, 2.0, 3.0, 4.0, 5.0, 6.0], shape=[2, 3]) b = tf.constant([1.0, 2.0, 3.0, 4.0, 5.0, 6.0], shape=[3, 2]) c.append(tf.matmul(a, b))with tf.device('/cpu:0'): sum = tf.add_n(c)# 新建session with log_device_placement并设置为True.sess = tf.Session(config=tf.ConfigProto(log_device_placement=True))# 运行这个op.print sess.run(sum)
你会看到如下输出:
Device mapping:/job:localhost/replica:0/task:0/gpu:0 -> device: 0, name: Tesla K20m, pci busid: 0000:02:00.0/job:localhost/replica:0/task:0/gpu:1 -> device: 1, name: Tesla K20m, pci busid: 0000:03:00.0/job:localhost/replica:0/task:0/gpu:2 -> device: 2, name: Tesla K20m, pci busid: 0000:83:00.0/job:localhost/replica:0/task:0/gpu:3 -> device: 3, name: Tesla K20m, pci busid: 0000:84:00.0Const_3: /job:localhost/replica:0/task:0/gpu:3Const_2: /job:localhost/replica:0/task:0/gpu:3MatMul_1: /job:localhost/replica:0/task:0/gpu:3Const_1: /job:localhost/replica:0/task:0/gpu:2Const: /job:localhost/replica:0/task:0/gpu:2MatMul: /job:localhost/replica:0/task:0/gpu:2AddN: /job:localhost/replica:0/task:0/cpu:0[[ 44. 56.] [ 98. 128.]]
5. 如何实现multi_gpu_model函数
def multi_gpu_model(num_gpus=1): grads = [] for i in range(num_gpus): with tf.device("/gpu:%d"%i): with tf.name_scope("tower_%d"%i): model = Model(is_training, config, scope) # 放到collection中,方便feed的时候取 tf.add_to_collection("train_model", model) grads.append(model.grad) #grad 是通过tf.gradients(loss, vars)求得 #以下这些add_to_collection可以直接在模型内部完成。 # 将loss放到 collection中, 方便以后操作 tf.add_to_collection("loss",model.loss) #将predict放到collection中,方便操作 tf.add_to_collection("predict", model.predict) #将 summary.merge op放到collection中,方便操作 tf.add_to_collection("merge_summary", model.merge_summary) # ... with tf.device("cpu:0"): averaged_gradients = average_gradients(grads)# average_gradients后面说明 opt = tf.train.GradientDescentOptimizer(learning_rate) train_op=opt.apply_gradients(zip(average_gradients,tf.trainable_variables())) return train_op
6. cifar10 tutorial-cifar10_multi_gpu_train.py
code 见 models/tutorials/image/cifar10/
# Copyright 2015 The TensorFlow Authors. All Rights Reserved.## Licensed under the Apache License, Version 2.0 (the "License");# you may not use this file except in compliance with the License.# You may obtain a copy of the License at## http://www.apache.org/licenses/LICENSE-2.0## Unless required by applicable law or agreed to in writing, software# distributed under the License is distributed on an "AS IS" BASIS,# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.# See the License for the specific language governing permissions and# limitations under the License.# =============================================================================="""A binary to train CIFAR-10 using multiple GPUs with synchronous updates.Accuracy:cifar10_multi_gpu_train.py achieves ~86% accuracy after 100K steps (256epochs of data) as judged by cifar10_eval.py.Speed: With batch_size 128.System | Step Time (sec/batch) | Accuracy--------------------------------------------------------------------1 Tesla K20m | 0.35-0.60 | ~86% at 60K steps (5 hours)1 Tesla K40m | 0.25-0.35 | ~86% at 100K steps (4 hours)2 Tesla K20m | 0.13-0.20 | ~84% at 30K steps (2.5 hours)3 Tesla K20m | 0.13-0.18 | ~84% at 30K steps4 Tesla K20m | ~0.10 | ~84% at 30K stepsUsage:Please see the tutorial and website for how to download the CIFAR-10data set, compile the program and train the model.http://tensorflow.org/tutorials/deep_cnn/"""from __future__ import absolute_importfrom __future__ import divisionfrom __future__ import print_functionfrom datetime import datetimeimport os.pathimport reimport timeimport numpy as npfrom six.moves import xrange # pylint: disable=redefined-builtinimport tensorflow as tfimport cifar10FLAGS = tf.app.flags.FLAGStf.app.flags.DEFINE_string('train_dir', '/tmp/cifar10_train', """Directory where to write event logs """ """and checkpoint.""")tf.app.flags.DEFINE_integer('max_steps', 1000000, """Number of batches to run.""")tf.app.flags.DEFINE_integer('num_gpus', 1, """How many GPUs to use.""")tf.app.flags.DEFINE_boolean('log_device_placement', False, """Whether to log device placement.""")def tower_loss(scope, images, labels): """Calculate the total loss on a single tower running the CIFAR model. Args: scope: unique prefix string identifying the CIFAR tower, e.g. 'tower_0' images: Images. 4D tensor of shape [batch_size, height, width, 3]. labels: Labels. 1D tensor of shape [batch_size]. Returns: Tensor of shape [] containing the total loss for a batch of data """ # Build inference Graph. logits = cifar10.inference(images) # Build the portion of the Graph calculating the losses. Note that we will # assemble the total_loss using a custom function below. _ = cifar10.loss(logits, labels) # Assemble all of the losses for the current tower only. losses = tf.get_collection('losses', scope) # Calculate the total loss for the current tower. total_loss = tf.add_n(losses, name='total_loss') # Attach a scalar summary to all individual losses and the total loss; do the # same for the averaged version of the losses. for l in losses + [total_loss]: # Remove 'tower_[0-9]/' from the name in case this is a multi-GPU training # session. This helps the clarity of presentation on tensorboard. loss_name = re.sub('%s_[0-9]*/' % cifar10.TOWER_NAME, '', l.op.name) tf.summary.scalar(loss_name, l) return total_lossdef average_gradients(tower_grads): """Calculate the average gradient for each shared variable across all towers. Note that this function provides a synchronization point across all towers. Args: tower_grads: List of lists of (gradient, variable) tuples. The outer list is over individual gradients. The inner list is over the gradient calculation for each tower. Returns: List of pairs of (gradient, variable) where the gradient has been averaged across all towers. """ average_grads = [] for grad_and_vars in zip(*tower_grads): # Note that each grad_and_vars looks like the following: # ((grad0_gpu0, var0_gpu0), ... , (grad0_gpuN, var0_gpuN)) grads = [] for g, _ in grad_and_vars: # Add 0 dimension to the gradients to represent the tower. expanded_g = tf.expand_dims(g, 0) # Append on a 'tower' dimension which we will average over below. grads.append(expanded_g) # Average over the 'tower' dimension. grad = tf.concat(axis=0, values=grads) grad = tf.reduce_mean(grad, 0) # Keep in mind that the Variables are redundant because they are shared # across towers. So .. we will just return the first tower's pointer to # the Variable. v = grad_and_vars[0][1] grad_and_var = (grad, v) average_grads.append(grad_and_var) return average_gradsdef train(): """Train CIFAR-10 for a number of steps.""" with tf.Graph().as_default(), tf.device('/cpu:0'): # Create a variable to count the number of train() calls. This equals the # number of batches processed * FLAGS.num_gpus. global_step = tf.get_variable( 'global_step', [], initializer=tf.constant_initializer(0), trainable=False) # Calculate the learning rate schedule. num_batches_per_epoch = (cifar10.NUM_EXAMPLES_PER_EPOCH_FOR_TRAIN / FLAGS.batch_size) decay_steps = int(num_batches_per_epoch * cifar10.NUM_EPOCHS_PER_DECAY) # Decay the learning rate exponentially based on the number of steps. lr = tf.train.exponential_decay(cifar10.INITIAL_LEARNING_RATE, global_step, decay_steps, cifar10.LEARNING_RATE_DECAY_FACTOR, staircase=True) # Create an optimizer that performs gradient descent. opt = tf.train.GradientDescentOptimizer(lr) # Get images and labels for CIFAR-10. images, labels = cifar10.distorted_inputs() batch_queue = tf.contrib.slim.prefetch_queue.prefetch_queue( [images, labels], capacity=2 * FLAGS.num_gpus) # Calculate the gradients for each model tower. tower_grads = [] with tf.variable_scope(tf.get_variable_scope()): for i in xrange(FLAGS.num_gpus): with tf.device('/gpu:%d' % i): with tf.name_scope('%s_%d' % (cifar10.TOWER_NAME, i)) as scope: # Dequeues one batch for the GPU image_batch, label_batch = batch_queue.dequeue() # Calculate the loss for one tower of the CIFAR model. This function # constructs the entire CIFAR model but shares the variables across # all towers. loss = tower_loss(scope, image_batch, label_batch) # Reuse variables for the next tower. tf.get_variable_scope().reuse_variables() # Retain the summaries from the final tower. summaries = tf.get_collection(tf.GraphKeys.SUMMARIES, scope) # Calculate the gradients for the batch of data on this CIFAR tower. grads = opt.compute_gradients(loss) # Keep track of the gradients across all towers. tower_grads.append(grads) # We must calculate the mean of each gradient. Note that this is the # synchronization point across all towers. grads = average_gradients(tower_grads) # Add a summary to track the learning rate. summaries.append(tf.summary.scalar('learning_rate', lr)) # Add histograms for gradients. for grad, var in grads: if grad is not None: summaries.append(tf.summary.histogram(var.op.name + '/gradients', grad)) # Apply the gradients to adjust the shared variables. apply_gradient_op = opt.apply_gradients(grads, global_step=global_step) # Add histograms for trainable variables. for var in tf.trainable_variables(): summaries.append(tf.summary.histogram(var.op.name, var)) # Track the moving averages of all trainable variables. variable_averages = tf.train.ExponentialMovingAverage( cifar10.MOVING_AVERAGE_DECAY, global_step) variables_averages_op = variable_averages.apply(tf.trainable_variables()) # Group all updates to into a single train op. train_op = tf.group(apply_gradient_op, variables_averages_op) # Create a saver. saver = tf.train.Saver(tf.global_variables()) # Build the summary operation from the last tower summaries. summary_op = tf.summary.merge(summaries) # Build an initialization operation to run below. init = tf.global_variables_initializer() # Start running operations on the Graph. allow_soft_placement must be set to # True to build towers on GPU, as some of the ops do not have GPU # implementations. sess = tf.Session(config=tf.ConfigProto( allow_soft_placement=True, log_device_placement=FLAGS.log_device_placement)) sess.run(init) # Start the queue runners. tf.train.start_queue_runners(sess=sess) summary_writer = tf.summary.FileWriter(FLAGS.train_dir, sess.graph) for step in xrange(FLAGS.max_steps): start_time = time.time() _, loss_value = sess.run([train_op, loss]) duration = time.time() - start_time assert not np.isnan(loss_value), 'Model diverged with loss = NaN' if step % 10 == 0: num_examples_per_step = FLAGS.batch_size * FLAGS.num_gpus examples_per_sec = num_examples_per_step / duration sec_per_batch = duration / FLAGS.num_gpus format_str = ('%s: step %d, loss = %.2f (%.1f examples/sec; %.3f ' 'sec/batch)') print (format_str % (datetime.now(), step, loss_value, examples_per_sec, sec_per_batch)) if step % 100 == 0: summary_str = sess.run(summary_op) summary_writer.add_summary(summary_str, step) # Save the model checkpoint periodically. if step % 1000 == 0 or (step + 1) == FLAGS.max_steps: checkpoint_path = os.path.join(FLAGS.train_dir, 'model.ckpt') saver.save(sess, checkpoint_path, global_step=step)def main(argv=None): # pylint: disable=unused-argument cifar10.maybe_download_and_extract() if tf.gfile.Exists(FLAGS.train_dir): tf.gfile.DeleteRecursively(FLAGS.train_dir) tf.gfile.MakeDirs(FLAGS.train_dir) train()if __name__ == '__main__':tf.app.run()
python cifar10_multi_gpu_train.py --num_gpus=2
参考文献
http://stackoverflow.com/questions/36668467/change-default-gpu-in-tensorflow
http://stackoverflow.com/questions/37893755/tensorflow-set-cuda-visible-devices-within-jupyter
(原)tensorflow中使用指定的GPU及GPU显存
Using GPUs
TensorFlow官方文档中文版 » 运作方式 » 使用gpu
tensorflow学习笔记(三十一):构建多GPU代码
cifar10 tutorial
CIFAR10 多 GPU 版本例程源码分析
tensorflow cifar_10 代码阅读与理解
- tensorflow:Multiple GPUs
- tensorflow学习笔记(十八):Multiple GPUs
- 使用 GPUs
- Using GPUs
- Linear regression of multiple features in Tensorflow
- 关于TensorFlow中的多图(Multiple Graphs)
- Tensorflow- Class path contains multiple SLF4J bindings
- 关于TensorFlow中的多图(Multiple Graphs)
- Purpose Of Multiple Graphs In Tensorflow
- How GPUs Work
- How GPUs Work
- How GPUs Work
- DL multi GPUs problem
- tensorflow TypeError: run() got multiple values for argument 'feed_dict'
- nvidia-smi: Control Your GPUs
- Multiple
- tensorflow
- TensorFlow
- Java入门教程(视频)
- [VS]在整个项目中查找
- PHP5中的this,self和parent关键字详解
- Redis主从复制的实现
- 文件定位函数 fseek
- tensorflow:Multiple GPUs
- Kubernetes服务之“运行单实例的有状态服务”
- java反射系列02 toString() toGenericString()
- JSON 入门指南
- 清华科技大讲堂(第4季) 企业商务与IT如何应对新时代的竞争大潮,实现数字化转型与创新
- 最新版windows 2012R2、Apache24、mod_wsgi、Django1.9、python3.4项目部署(生产环境)
- 剑指offer——扑克牌顺子
- react-redux初探1
- angularJs的理解之two