1、使用多卡make runtest时,若直接make,会出现如下编译问题,修改方法为:使用export CUDA_VISIBLE_DEVICES=?指定某一张卡。
2、使用export CUDA_VISIBLE_DEVICES=?指定第?卡进行make runtest时(在Docker中编译),Makefile.config文件中应设置GPUID为0,因为Docker中只显示一张显卡。采用默认第0张卡即可。
# The ID of the GPU that 'make runtest' will use to run unit tests.
TEST_GPUID := 0
3、libopencv_core.so.3.2: cannot open shared object file: No such file or directory
解决方法:https://github.com/GaoHongchen/DIPDemo/issues/1
- Find the folder containing the shared library libopencv_core.so.3.2 using the following command line.
sudo find / -name libopencv_core.so.3.2
Then I got the result: /usr/local/lib/libopencv_core.so.3.2.
2. Create a file called /etc/ld.so.conf.d/opencv.conf and write to it the path to the folder where the binary is stored.For example, I wrote /usr/local/lib/ to my opencv.conf file.
3. Run the command line as follows.
ldconfig -v