cudnn

CUDA linking error on OS X El Capitan

时间秒杀一切 提交于 2019-12-04 06:32:19
问题 I've been trying to install CUDA and cuDNN for TensorFlow as instructed on the TensorFlow page. Everything works until I do the following: $ cp -r /usr/local/cuda/samples ~/cuda-samples $ pushd ~/cuda-samples $ make I then get the following error: clang++ -rpath /Developer/NVIDIA/CUDA-7.5/lib -L/Developer/NVIDIA/CUDA-7.5/lib -framework CUDA -o clock_nvrtc clock.o -lnvrtc ld: framework not found CUDA clang: error: linker command failed with exit code 1 (use -v to see invocation) make[1]: ***

How to enable Keras with Theano to utilize multiple GPUs

三世轮回 提交于 2019-12-03 17:26:39
问题 Setup: Using a Amazon Linux system with a Nvidia GPU I'm using Keras 1.0.1 Running Theano v0.8.2 backend Using CUDA and CuDNN THEANO_FLAGS="device=gpu,floatX=float32,lib.cnmem=1" Everything works fine, but I run out of video memory on large models when I increase the batch size to speed up training. I figure moving to a 4 GPU system would in theory either improve total memory available or allow smaller batches to build faster, but observing the the nvidia stats, I can see only one GPU is used

在docker容器中python3.5环境下使用DIGITS训练caffe模型

旧街凉风 提交于 2019-12-03 17:02:54
********* 此处使用的基础镜像为 nvcr.io/nvidia/digits:18.06 ,镜像大小为6.04GB,可从nvidia官方pull此镜像; 容器配置:   CUDA:9.0   CUDNN:7.0 注:此文档建立在已会使用python2.7版本的DIGITS基础之上 镜像中含有python3.5与python2.7两个版本,直接使用python3.5 修改系统python默认值,使用python3为默认启动:   sudo update-alternatives --install /usr/bin/python python /usr/bin/python2 100   sudo update-alternatives --install /usr/bin/python python /usr/bin/python3 150 ********* 一、编译安装caffe 从github下载caffe源码,准备编译,下载地址: https://github.com/BVLC/caffe.git 【CUDA与CUDNN请查找对应的安装教程,此处忽略】 进入caffe目录 1、安装依赖 :   sudo apt-get install libprotobuf-dev libleveldb-dev libsnappy-dev libopencv-dev libhdf5

Would cuDNN v6.0 work with TensorFlow currently?

|▌冷眼眸甩不掉的悲伤 提交于 2019-12-03 14:22:52
Do we specifically need cuDNN v5.1 (as suggested) for TensorFlow, or would the latest version (v6.0) work as well? Is there backward compatibility in cuDNN versions? No, cuDNN 6.0 is not supported in the latest 1.2 version. But there is a hope: the official release notes tell the following: TensorFlow 1.2 may be the last time we build with cuDNN 5.1. Starting with TensorFlow 1.3, we will try to build all our prebuilt binaries with cuDNN 6.0. While we will try to keep our source code compatible with cuDNN 5.1, it will be best effort. So hopefully the next 1.3 version will use cuDNN 6.0.

On windows, how do you verify the version number of CuDNN installed?

百般思念 提交于 2019-12-03 12:20:20
On windows, how do you verify the version number of CuDNN installed? I'm finding a lot of results when I search for the answer for Linux machines. It doesn't seem like there's a clear way to find out what version is installed for Windows. Go to C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v8.0\include\ open cudnn.h define CUDNN_MAJOR 5 define CUDNN_MINOR 1 define CUDNN_PATCHLEVEL 10 in my case its 5.1.10 来源: https://stackoverflow.com/questions/45641087/on-windows-how-do-you-verify-the-version-number-of-cudnn-installed

Use of cuDNN RNN

匿名 (未验证) 提交于 2019-12-03 09:06:55
可以将文章内容翻译成中文,广告屏蔽插件可能会导致该功能失效(如失效,请关闭广告屏蔽插件后再试): 问题: I will first summarize what I think I understood about cuDNN 5.1 rnn functions: Tensor dimensions x = [seq_length, batch_size, vocab_size] # input y = [seq_length, batch_size, hiddenSize] # output dx = [seq_length, batch_size, vocab_size] # input gradient dy = [seq_length, batch_size, hiddenSize] # output gradient hx = [num_layer, batch_size, hiddenSize] # input hidden state hy = [num_layer, batch_size, hiddenSize] # output hidden state cx = [num_layer, batch_size, hiddenSize] # input cell state cy = [num_layer, batch_size, hiddenSize] #

win10下安装Tensorflow1.7+CUDA9.0+Cudnn7.0.3

风流意气都作罢 提交于 2019-12-03 08:31:03
最近看到CUDA9.0对比之前的CUDA8.0速度有了明显的提升,所以想尝尝鲜,安装之后, 训练和预测的速度均有很大的提升: 有需要安装我这个版本的不妨可以参考参考!如果在安装过程中出现什么问题,欢迎在下方评论,笔者可以帮你尝试解决一下! 1.安装前的准备: a.Visual Studio 2013, 2015 ,2017任意一个版本 b.Anaconda c.更新一下你的pip d.python版本3.5+ e. 最重要的 : Tensorflow有两个版本:GPU和CPU版本,CPU的随意啦,很好安装; GPU 版本需要 CUDA 和 cuDNN 的支持,如果你是独显+集显,那么推荐你用GPU版本的,因为CPU版本的速度是在太慢!并且CUDA是英伟达下属的程序,所以你的GPU最好是 英伟达 的,AMD的显卡没有CUDA加速!满足以上条件之后,你需要查看一下你的英伟达GPU是否支持CUDA(毕竟Xp微软也不提供更新了~~),以下是Geforce支持的目录: 你也可以 点击查看你的GPU是否支持CUDA 满足以上条件之后,你就可以安装Tensorflow了! 2.安装Tensorflow: 打开cmd窗口,输入以下命令(使用豆瓣的源,下载速度很快): GPU版本 pip install -i https://pypi.doubanio.com/simple/ tensorflow

ubuntu16.04 + CUDA8.0+cudnn5.0+tensorflow-GPU+python2.7

风流意气都作罢 提交于 2019-12-03 08:30:42
linux-CUDA8.0+tensorflow-GPU+python2.7 本Markdown编辑器使用[StackEdit][6]修改而来,用它写博客,将会带来全新的体验哦: ubuntu16.04 cuda-8.0 cudnn-v5.0 tensorflow-GPU python2.7 安装步骤 介绍一种使用预编译的tensorflow按转GPU版的简单教程。 1.下载预编译版本的tensorflow-GPU的.whl文件 # Python 2 $ sudo pip install --upgrade $TF_BINARY_URL # Python 3 $ sudo pip3 install --upgrade $TF_BINARY_URL 其中环境变量 TF_BINARY_URL 根据你的环境进行设置,典型选项如下: # Ubuntu/Linux 64-bit, CPU only, Python 2.7 $ export TF_BINARY_URL=https://storage .googleapis .com /tensorflow/linux/cpu/tensorflow- 0.12 .1 -cp27-none-linux_x86_64 .whl # Ubuntu/Linux 64-bit, GPU enabled, Python 2.7 # 需要 CUDA

After building TensorFlow from source, seeing libcudart.so and libcudnn errors

匿名 (未验证) 提交于 2019-12-03 08:28:06
可以将文章内容翻译成中文,广告屏蔽插件可能会导致该功能失效(如失效,请关闭广告屏蔽插件后再试): 由 翻译 强力驱动 问题: I'm building TensorFlow from source code. The build appears to succeed; however, when my TensorFlow program invokes import tensorflow , one or both of the following errors appear: ImportError: libcudart.so.8.0: cannot open shared object file: No such file or directory ImportError: libcudnn.5: cannot open shared object file: No such file or directory 回答1: First , for the following error: ImportError: libcudart.so.8.0: cannot open shared object file: No such file or directory make sure your LD_LIBRARY_PATH includes your lib64

win10+python3.6安装tensorflow-gpu1.8.0+CUDA9.0+cuDNN7.1

安稳与你 提交于 2019-12-03 08:24:43
以下记录为我在电脑上安装的过程(2018.4.28): 版本信息: win10 python3.6.5 tensorflow-gpu1.8.0 CUDA9.0 cuDNN7.1 说明: 1.为啥要安装这几个? tensorflow是微软家出的一个用来做深度学习的开源库,这个库有两个版本 一种是在CPU上运行,安装方法为:pip install tensorflow 一种是在GPU上运行,安装方法为 pip install tensorflow-gpu 注意,gpu版本的库安了后想要运行必须要按CUDA和cuDNN缺一不可:-) 2.cuDNN也就几十mb而已,但CUDA可是上了gb,那么为什么还要安gpu版本的呢? ......_(:з」∠)_答案是因为跑得快 根据Tensorlayor(一个针对tensorflow库做封装的开源项目)官网上的原话: 在 GPU 上训练全连接神经网络比在 CPU 上训练往往要快 10~20 倍。对于卷积神经网络,往往会快 50 倍。这需要有一个 NIVIDA 的 GPU,以及安装 CUDA 和 cuDNN。 3.那么什么是CUDA和cuDNN呢? 这是英伟达家出的两个软件,用来使GPU具备做深度学习计算的能力 安装: 首先,python3的安装直接去官网找个安装包安装就行了,但注意,就目前来说python3.6以上的版本以及所有的32位版本