A maneira correta de fazer isso é descrita na documentação do CUDA:
link
Script do link acima:
#!/bin/bash
/sbin/modprobe nvidia
if [ "$?" -eq 0 ]; then
# Count the number of NVIDIA controllers found.
NVDEVS='lspci | grep -i NVIDIA'
N3D='echo "$NVDEVS" | grep "3D controller" | wc -l'
NVGA='echo "$NVDEVS" | grep "VGA compatible controller" | wc -l'
N='expr $N3D + $NVGA - 1'
for i in 'seq 0 $N'; do
mknod -m 666 /dev/nvidia$i c 195 $i
done
mknod -m 666 /dev/nvidiactl c 195 255
else
exit 1
fi
/sbin/modprobe nvidia-uvm
if [ "$?" -eq 0 ]; then
# Find out the major device number used by the nvidia-uvm driver
D='grep nvidia-uvm /proc/devices | awk '{print }''
mknod -m 666 /dev/nvidia-uvm c $D 0
else
exit 1
fi