demo in readme not work
idreamerhx opened this issue · comments
My os: archlinux latest
extra/egl-wayland 2:1.1.11-2 [installed]
extra/libvdpau 1.5-1 [installed]
extra/libxnvctrl 520.56.06-1 [installed]
extra/nvidia-dkms 520.56.06-2 [installed]
extra/nvidia-prime 1.0-4 [installed]
extra/nvidia-settings 520.56.06-1 [installed]
extra/nvidia-utils 520.56.06-2 [installed]
extra/opencl-nvidia 520.56.06-2 [installed]
community/cuda 11.8.0-1 [installed]
community/cuda-tools 11.8.0-1 [installed]
community/cudnn 8.5.0.96-1 [installed]
community/nccl 2.14.3-1 [installed]
community/nvtop 3.0.0-1 [installed]
archlinuxcn/libnvidia-container 1.9.0-1 [installed]
archlinuxcn/libnvidia-container-tools 1.9.0-1 [installed]
archlinuxcn/nvidia-container-toolkit 1.9.0-1 [installed]
NVIDIA-SMI 520.56.06 Driver Version: 520.56.06 CUDA Version: 11.8
run the demo in readme.md failed:
Setup a new set of namespaces
cd $(mktemp -d) && mkdir rootfs
sudo unshare --mount --pid --fork
Setup a rootfs based on Ubuntu 16.04 inside the new namespaces
curl http://cdimage.ubuntu.com/ubuntu-base/releases/16.04/release/ubuntu-base-16.04.6-base-amd64.tar.gz | tar -C rootfs -xz
useradd -R $(realpath rootfs) -U -u 1000 -s /bin/bash nvidia
mount --bind rootfs rootfs
mount --make-private rootfs
cd rootfs
Mount standard filesystems
mount -t proc none proc
mount -t sysfs none sys
mount -t tmpfs none tmp
mount -t tmpfs none run
Isolate the first GPU device along with basic utilities
nvidia-container-cli --load-kmods configure --ldconfig=@/sbin/ldconfig.real --no-cgroups --utility --device 0 $(pwd)
** --ldconfig=@/sbin/ldconfig.real ** I removed this
Change into the new rootfs
pivot_root . mnt # this not work
umount -l mnt
exec chroot --userspec 1000:1000 . env -i bash
Run nvidia-smi from within the container
nvidia-smi -L
in chrooted rootfs nvidia-smi show nothing.
I tried:
arch-chroot ubu22x86-base
and in another shell cd ubu22x86-base and nvidia-container-cli --load-kmods configure --no-cgroups --utility --device 0 $(pwd)
nvidia-smi works
but a simple program cudaGetDeviceCount returns error code 35.
in chrooted rootfs
NVIDIA-SMI 520.56.06 Driver Version: 520.56.06 CUDA Version: N/A
I installed cuda toolkit cuda_11.8.0_520.61.05_linux.run
is there any document about nvidia-container-cli
in chrooted ubuntu 22
apt install nvidia-cuda-toolkit
NVIDIA-SMI 520.56.06 Driver Version: 520.56.06 CUDA Version: 11.6
hey, guys would you please add some document really simple such as :
nvidia-container-cli --load-kmods configure --no-cgroups --utility --compute --device 0 $(pwd)
this works to show cuda version。
or just add utility compute in --usage or --help
there is another issue. when exit chrooted rootfs,
umount: /mnt/data1/chroots/ubu22x86-cuda118/dev: target is busy.
umount: /mnt/data1/chroots/ubu22x86-cuda118/proc: target is busy.
should I clearup myself/?