使用gpu从Ubuntu 16上的源安装tensorflow出现问题
[英]problems with installation tensorflow from source on ubuntu 16 with gpu
问题描述
I have ubuntu 16.04 I have installed CUDA 7.5 from Ubuntu repo and cuDNN 5.1.3 for CUDA 7.5 and ran CUDA examples that works, same for pycuda and I want to install tensorflow from source with gpu support, tensorflow configuration is stuck on nvvm, and I can't find it in system also find isn't any helpful 
我有ubuntu 16.04，我已经从Ubuntu repo安装了CUDA 7.5，并为CUDA 7.5安装了cuDNN 5.1.3，并运行了适用于pycuda的CUDA示例，我想从源代码安装具有gpu支持的tensorflow，tensorflow配置卡在nvvm上，并且我在系统中找不到它，也找不到任何帮助
$ find / -name nvvm*
/usr/include/nvvm.h
/usr/share/doc/nvidia-cuda-doc/html/nvvm-ir-spec
where can I find nvvm? 在哪里可以找到nvvm？
1 个解决方案
解决方案1
0 2017-01-20 21:16:38
Your find command is wrong. 您的find命令是错误的。 You need to quote the -name argument to prevent shell globbing: 您需要用引号-name来防止shell阻塞：
$ find / -name 'nvvm*'
声明:本站的技术帖子网页，遵循CC BY-SA 4.0协议，如果您需要转载，请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.