Recording a very low-level errors: command 'usr / local / cuda / bin / vncc' failed with exit status 2

Recording a very low-level errors: command 'usr / local / cuda / bin / vncc' failed with exit status 2

Silent, and today is a very low-level errors stuck

The main development environment is now really a bit messy: laptop or desktop computer or server or a big guy, each has a slightly different configuration, so often a problem in the environment.

报错:command 'usr/local/cuda/bin/vncc' failed with exit status 2

The reason: this environment variable is not configured CUDA_HOME because there are several versions of cuda on the server, so this automatically selects the cuda soft link points to version

Solution:

  1. The use of soft links to their own version

  2. In ~ / .bashrc configuration:

    export CUDA_HOME="<要使用的cuda的目录>"

Note that this CUDA_HOME and the path is not the same, I was there to see their PATH cuda / bin, and no attention, in fact, this is two different things: PATH command input is the default location to find the corresponding command file is located in the shell, the CUDA_HOME is to provide for calls to the software, this machine represents the configuration directory where cuda. At that time thought to understand this, for a long time in setup.py, in fact, people are called directly to your environment variables

Guess you like

Origin www.cnblogs.com/jiading/p/12101680.html