RedHat安装CUDA-9.1以及gcc-6过程及问题

一、安装CUDA:

安装cuda时,跳过了驱动安装:

http://blog.sina.com.cn/s/blog_15d0344a30102xe0n.html

https://www.cnblogs.com/cxf-zzj/p/9888389.html

升级了驱动后:

二、测试cuda遇到的问题

GPU is lost

解决:reboot

三、安装gcc

扫描二维码关注公众号,回复: 5305011 查看本文章

参考:

https://blog.csdn.net/stormbjm/article/details/9107831

https://blog.csdn.net/muziwenrong/article/details/78672059

https://www.cnblogs.com/tankblog/p/6096681.html

下载gcc:http://ftp.gnu.org/gnu/gcc/gcc-4.4.7/

各种软件:https://mirrors.ustc.edu.cn/gnu/

1.mpfr make check时出错:

https://blog.csdn.net/ai297313/article/details/44708373

https://blog.csdn.net/ytang_/article/details/79417979

结合第二个的第二种情况解决,make check通过

2.error while loading shared libraries: libmpfr.so.4:

$ echo $LD_LIBRARY_PATH

如果没有的话,手动添加即可

$ export LD_LIBRARY_PATH="$LD_LIBRARY_PATH:/usr/local/lib"

3.安装gcc报I suspect your system does not have 32-bit developement libraries (libc and headers). If you have them, rerun configure with --enable-multilib. If you do not have them,  and want to build a 64-bit-only compiler, rerun configure with --disable-multilib.

加上--disable-multilib:

../configure --prefix=/usr/local/gcc-6.1.0 --enable-threads=posix --disable-checking --enable-languages=c,c++,objc --disable-multilib

4.安装gcc时make check出现autogen: Command not found

(忽略,centos下太难找了)

 

猜你喜欢

转载自blog.csdn.net/m0_37705898/article/details/87902012
今日推荐