Shulou(Shulou.com)06/01 Report--

This article mainly shows you "how to install Ubuntu14.04 CuDNN6.5", the content is easy to understand, clear, hope to help you solve your doubts, the following let the editor lead you to study and learn "how to install Ubuntu14.04 CuDNN6.5" this article.

Environment

System: Ubuntu 14.04 x64

CUDA:cuda7 (cuda must be installed before installing cudnn. For more information, please see the related article http://blog.csdn.net/honyniu/article/details/46387429 of this blog)

Brief introduction

CuDNN is a set of GPU computing acceleration scheme specially designed for Deep Learning framework. At present, the DL libraries supported include Caffe,ConvNet, Torch7 and so on. The installation here is mainly for later caffe to use cudnn acceleration, which is not necessary in the process of using caffe.

CuDNN can be obtained free of charge on the official website and can be downloaded after registering an account.

Download link: https://developer.nvidia.com/cuDNN

Installation package

Here I have downloaded two versions of the installation package, both of which can be downloaded after registration, as follows

Cudnn-6.5-linux-R1cudnn-6.5-linux-x64-v2

The file names and structure under the two installation packages are roughly the same, except that the interfaces provided by the two version libraries of cudnn are completely different (see the CUDNN_Library.pdf interface documentation provided), and caffe used the R1 interface in the early version, but then the latest one became the v2 interface, which can not be used at all. There are still many projects based on the early caffe, if you want to use these projects, you need to install R1, and there are also requirements on the latest caffe, so you need to use both, just switch between different compilations.

Installation

The basic principle is to add the lib file to the lib folder that the system can find, and add the header file to the include folder where the system can find it, and put it here in the corresponding installation directory of cuda (provided that cuda has been installed successfully)

The installation process is exactly the same for R1 and v2

Cd cudnn_dirsudo cp lib* / usr/local/cuda/lib64/sudo cp cudnn.h / usr/local/cuda/include/

When you want to switch from R1 to v2, you only need to install v2 again, similarly, from v2 to R1

Error

If 'USE_CUDNN: = 1' is turned on when compiling caffe, the relevant files will not be copied as above, and an error will be reported at compile time

. / include/caffe/util/cudnn.hpp:5:19: fatal error: cudnn.h: there is no file or directory. # include is all the contents of the article "how to install Ubuntu14.04 CuDNN6.5". Thank you for reading! I believe we all have a certain understanding, hope to share the content to help you, if you want to learn more knowledge, welcome to follow the industry information channel!

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.