I have four GTX1080, installation with cuda 7.5 was successful. But when I run example data, it requires sm6_1 from cuda 8.0. So I uninstall cuda 7.5 and install 8.0, but the pycuda keep looking for libcurand.so.7.5 instead.
echo $LD_LIBRARY_PATH
/home/jianshi/bin/relion2/build/lib:/usr/local/IMOD/lib::/usr/lib64/openmpi/lib:/usr/local/cuda/lib64
And /usr/local/cuda is linked to cuda-8.0 already.
I am wondering what wrong?
--------------------- GPUs -----------------------------------------
CryoSPARC includes a scheduler and queuing system for jobs. This
can be configured to run on only some of the GPUs in your system,
in case you want to reserve some for other uses.
Right now, all GPUs will be enabled…
Traceback (most recent call last):
File “/home/jianshi/cryosparc/bin/configure.py”, line 337, in
import pycuda.driver as cudrv
File “/home/jianshi/cryosparc/anaconda2/lib/python2.7/site-packages/pycuda/driver.py”, line 5, in
from pycuda._driver import * # noqa
ImportError: libcurand.so.7.5: cannot open shared object file: No such file or directory
ERROR: