TF报错:CUDA_ERROE_OUT_OF_MEMORY
来源:互联网 发布:鸡尾酒工具推荐 知乎 编辑:程序博客网 时间:2024/06/06 03:59
I tensorflow/core/common_runtime/gpu/gpu_device.cc:975] Creating TensorFlow device (/gpu:0) -> (device: 0, name: GeForce GTX 860M, pci bus id: 0000:01:00.0)E tensorflow/stream_executor/cuda/cuda_dnn.cc:397] could not create cudnn handle: CUDNN_STATUS_INTERNAL_ERRORE tensorflow/stream_executor/cuda/cuda_dnn.cc:364] could not destroy cudnn handle: CUDNN_STATUS_BAD_PARAMF tensorflow/core/kernels/conv_ops.cc:605] Check failed: stream->parent()->GetConvolveAlgorithms(&algorithms) 已放弃 (核心已转储)
找到如下为tf分配GPU资源代码:
gpu_options = tf.GPUOptions(per_process_gpu_memory_fraction=0.6) #gpu资源的占比,显存×0.6with tf.Session(config=tf.ConfigProto(log_device_placement=False,gpu_options=gpu_options)) as sess:
也可以设置
config.gpu_options.allow_growth = True #开始不会给tensorflow全部gpu资源 而是按需增加
不够不干活,就是这么牛逼。。。。。。。。。。。。
阅读全文
0 0
- TF报错:CUDA_ERROE_OUT_OF_MEMORY
- 【TensorFlow报错】tf.pack改为tf.stack
- initialize_all_variables报错,改为tf.global_variables_initializer
- SummaryWriter报错,改为tf.summary.FileWriter
- SummaryWriter报错,改为tf.summary.FileWriter
- initialize_all_variables报错,改为tf.global_variables_initializer
- initialize_all_variables报错,改为tf.global_variables_initializer
- 【TensorFlow报错】initialize_all_variables改为tf.global_variables_initializer
- 解决PyCharm [import tensorflow as tf]报错
- 解决Jupyter notebook[import tensorflow as tf]报错
- tf 报错提示ValueError: setting an array element with a sequence.
- TF
- tf
- 使用 tf.gfile.DeleteRecursively 时,报 FailedPreconditionError 错误
- tf.scatter_update tf.scatter_sub
- tf.get_variable tf.variable_scope tf.name_scope
- TF/IDF
- TF-IDF
- 实模式,保护模式
- Python如何使用gedit和powershell
- c语言 模拟骰子实验 产生随机数
- SQL数据操作
- CentOS7 安装svn
- TF报错:CUDA_ERROE_OUT_OF_MEMORY
- chrome中ul的padding-left: 40px
- test
- c语言:判断一个年份是不是闰年和鸡兔同笼问题
- React DOM click focus textbox
- 用python3 opencv3 实现相机学生证的实时跟踪
- svn 批量移除版本库
- linux学习
- LeetCode刷题(35)--Permutation Sequence