Open minwim opened 1 year ago
I've same problem here :( . Did you found a solution to fix it?
I've same problem with WSL2
I've found an explanation and how to get around the problem. First you need to install Nvidia Container Kit, if you haven't already. The problem is that it doesn't work with docker desktop. If you use it, you need to create a new context in order to use docker engine only (issue #1717).
Hoping to help.
I've found an explanation and how to get around the problem. First you need to install Nvidia Container Kit, if you haven't already. The problem is that it doesn't work with docker desktop. If you use it, you need to create a new context in order to use docker engine only (issue #1717).
Hoping to help.
It occurs to me too, cloud you plz explain how to create a new context with detail?
I also experienced the same problem
docker run --rm -it --gpus=all nvcr.io/nvidia/k8s/cuda-sample:nbody nbody -gpu -benchmark
docker: error during connect: in the default daemon configuration on Windows, the docker client must be run with elevated privileges to connect: Head "http://%2F%2F.%2Fpipe%2Fdocker_engine/_ping": open //./pipe/docker_engine: The system cannot find the file specified.
See 'docker run --help'.
PS C:\Users\Indo Online> docker run --rm -it --gpus=all nvcr.io/nvidia/k8s/cuda-sample:nbody nbody -gpu -benchmark
Run "nbody -benchmark [-numbodies=
NOTE: The CUDA Samples are not meant for performance measurements. Results may vary when GPU Boost is enabled.
Error: only 0 Devices available, 1 requested. Exiting.
I've found an explanation and how to get around the problem. First you need to install Nvidia Container Kit, if you haven't already. The problem is that it doesn't work with docker desktop. If you use it, you need to create a new context in order to use docker engine only (issue #1717). Hoping to help.
It occurs to me too, cloud you plz explain how to create a new context with detail?
So I don't know if the problem is still the same, because it seems to me that maj docker has changed its behavior with WSL (https://docs.docker.com/desktop/gpu/). Normally, it's taken care of now.
Before, Docker desktop didn't support GPU support. So you had to change the execution context to dockerd docker context use default
.
I also experienced the same problem docker run --rm -it --gpus=all nvcr.io/nvidia/k8s/cuda-sample:nbody nbody -gpu -benchmark docker: error during connect: in the default daemon configuration on Windows, the docker client must be run with elevated privileges to connect: Head "http://%2F%2F.%2Fpipe%2Fdocker_engine/_ping": open //./pipe/docker_engine: The system cannot find the file specified. See 'docker run --help'. PS C:\Users\Indo Online> docker run --rm -it --gpus=all nvcr.io/nvidia/k8s/cuda-sample:nbody nbody -gpu -benchmark Run "nbody -benchmark [-numbodies=]" to measure performance. -fullscreen (run n-body simulation in fullscreen mode) -fp64 (use double precision floating point values for simulation) -hostmem (stores simulation data in host memory) -benchmark (run benchmark to measure performance) -numbodies= (number of bodies (>= 1) to run in simulation) -device= (where d=0,1,2.... for the CUDA device to use) -numdevices= (where i=(number of CUDA devices > 0) to use for simulation) -compare (compares simulation results running once on the default GPU and once on the CPU) -cpu (run n-body simulation on the CPU) -tipsy=
(load a tipsy model file for simulation) NOTE: The CUDA Samples are not meant for performance measurements. Results may vary when GPU Boost is enabled.
Error: only 0 Devices available, 1 requested. Exiting.
I had the same problem, I updated the packages on wsl as well as docker-desktop. Then I had a problem initializing NVML. I followed this solution and it works: https://forums.developer.nvidia.com/t/nvida-container-toolkit-failed-to-initialize-nvml-unknown-error/286219/2
running on the same issue, if anybody needs a log or something please ask!
No matter what I do, it does not work. Tried every single solution available
Running latest WSL, Docker Desktop and NVIDIA driver fixes the issue on my end. I can provide versions if someone’s interested.
Running latest WSL, Docker Desktop and NVIDIA driver fixes the issue on my end. I can provide versions if someone’s interested.
Yeah also provide what things you are following if possible.
I'm stuck for 2 days on this.
Running latest WSL, Docker Desktop and NVIDIA driver fixes the issue on my end. I can provide versions if someone’s interested.
Yeah also provide what things you are following if possible.
I'm stuck for 2 days on this.
Driver Version: 555.99
CUDA Version: 12.5
WSL Version: 2.2.4.0
Docker Desktop Version: 4.31.0
After running docker run --rm -it --gpus=all nvcr.io/nvidia/k8s/cuda-sample:nbody nbody -gpu -benchmark
> Windowed mode
> Simulation data stored in video memory
> Single precision floating point simulation
> 1 Devices used for simulation
GPU Device 0: "Ampere" with compute capability 8.6
> Compute 8.6 CUDA device: [NVIDIA GeForce RTX 3060]
28672 bodies, total time for 10 iterations: 22.861 ms
= 359.604 billion interactions per second
= 7192.081 single-precision GFLOP/s at 20 flops per interaction
Running latest WSL, Docker Desktop and NVIDIA driver fixes the issue on my end. I can provide versions if someone’s interested.
Yeah also provide what things you are following if possible. I'm stuck for 2 days on this.
Driver Version: 555.99
CUDA Version: 12.5
WSL Version: 2.2.4.0
Docker Desktop Version: 4.31.0
After running
docker run --rm -it --gpus=all nvcr.io/nvidia/k8s/cuda-sample:nbody nbody -gpu -benchmark
> Windowed mode > Simulation data stored in video memory > Single precision floating point simulation > 1 Devices used for simulation GPU Device 0: "Ampere" with compute capability 8.6 > Compute 8.6 CUDA device: [NVIDIA GeForce RTX 3060] 28672 bodies, total time for 10 iterations: 22.861 ms = 359.604 billion interactions per second = 7192.081 single-precision GFLOP/s at 20 flops per interaction
Can you also say what you installed on both inside ubuntu and in your windows for the cuda to work
Also what is the ubuntu version on your WSL, what is the default distro of wsl and anything else that might be important to have a look at
Running latest WSL, Docker Desktop and NVIDIA driver fixes the issue on my end. I can provide versions if someone’s interested.
Yeah also provide what things you are following if possible. I'm stuck for 2 days on this.
Driver Version: 555.99 CUDA Version: 12.5 WSL Version: 2.2.4.0 Docker Desktop Version: 4.31.0 After running
docker run --rm -it --gpus=all nvcr.io/nvidia/k8s/cuda-sample:nbody nbody -gpu -benchmark
> Windowed mode > Simulation data stored in video memory > Single precision floating point simulation > 1 Devices used for simulation GPU Device 0: "Ampere" with compute capability 8.6 > Compute 8.6 CUDA device: [NVIDIA GeForce RTX 3060] 28672 bodies, total time for 10 iterations: 22.861 ms = 359.604 billion interactions per second = 7192.081 single-precision GFLOP/s at 20 flops per interaction
Can you also say what you installed on both inside ubuntu and in your windows for the cuda to work
Also what is the ubuntu version on your WSL, what is the default distro of wsl and anything else that might be important to have a look at
I haven't installed the CUDA Toolkit inside of Ubuntu. Just the NVIDIA Studio Driver on my Windows machine. I am using the default WSL distro Ubuntu 22.04.
Running latest WSL, Docker Desktop and NVIDIA driver fixes the issue on my end. I can provide versions if someone’s interested.
Yeah also provide what things you are following if possible. I'm stuck for 2 days on this.
Driver Version: 555.99
CUDA Version: 12.5
WSL Version: 2.2.4.0
Docker Desktop Version: 4.31.0
After running
docker run --rm -it --gpus=all nvcr.io/nvidia/k8s/cuda-sample:nbody nbody -gpu -benchmark
> Windowed mode > Simulation data stored in video memory > Single precision floating point simulation > 1 Devices used for simulation GPU Device 0: "Ampere" with compute capability 8.6 > Compute 8.6 CUDA device: [NVIDIA GeForce RTX 3060] 28672 bodies, total time for 10 iterations: 22.861 ms = 359.604 billion interactions per second = 7192.081 single-precision GFLOP/s at 20 flops per interaction
Problem Solved!!! Thanks Bro
I had Docker Desktop Version: 4.30.0
and got the: Error: only 0 Devices available, 1 requested. Exiting.
Updated to 4.31.1 and got the:
> Windowed mode
> Simulation data stored in video memory
> Single precision floating point simulation
> 1 Devices used for simulation
GPU Device 0: "Pascal" with compute capability 6.1
> Compute 6.1 CUDA device: [NVIDIA GeForce GTX 1080 Ti]
28672 bodies, total time for 10 iterations: 24.321 ms
= 338.013 billion interactions per second
= 6760.262 single-precision GFLOP/s at 20 flops per interaction
Running latest WSL, Docker Desktop and NVIDIA driver fixes the issue on my end. I can provide versions if someone’s interested.
thank you solved my issue.
Running latest WSL, Docker Desktop and NVIDIA driver fixes the issue on my end. I can provide versions if someone’s interested.
Yeah also provide what things you are following if possible. I'm stuck for 2 days on this.
Driver Version: 555.99
CUDA Version: 12.5
WSL Version: 2.2.4.0
Docker Desktop Version: 4.31.0
After running
docker run --rm -it --gpus=all nvcr.io/nvidia/k8s/cuda-sample:nbody nbody -gpu -benchmark
> Windowed mode > Simulation data stored in video memory > Single precision floating point simulation > 1 Devices used for simulation GPU Device 0: "Ampere" with compute capability 8.6 > Compute 8.6 CUDA device: [NVIDIA GeForce RTX 3060] 28672 bodies, total time for 10 iterations: 22.861 ms = 359.604 billion interactions per second = 7192.081 single-precision GFLOP/s at 20 flops per interaction
did exactly the same and worked for me
Yes! just updated Docker Desktop Version to 4.31.1
Worked for me as well!
Windows Version
Microsoft Windows [version 10.0.22621.1848]
WSL Version
1.2.5.0
Are you using WSL 1 or WSL 2?
Kernel Version
5.15.90
Distro Version
Ubuntu 22.04
Other Software
Docker version 24.0.2, build cb74dfc
Repro Steps
when I run docker run --gpus all nvcr.io/nvidia/k8s/cuda-sample:nbody nbody -gpu -benchmark
Expected Behavior
GPU can be used
Actual Behavior
root@DESKTOP-UOKSUTD:~# sudo docker run --gpus all nvcr.io/nvidia/k8s/cuda-sample:nbody nbody -gpu -benchmark Run "nbody -benchmark [-numbodies=]" to measure performance.
-fullscreen (run n-body simulation in fullscreen mode)
-fp64 (use double precision floating point values for simulation)
-hostmem (stores simulation data in host memory)
-benchmark (run benchmark to measure performance)
-numbodies= (number of bodies (>= 1) to run in simulation)
-device= (where d=0,1,2.... for the CUDA device to use)
-numdevices= (where i=(number of CUDA devices > 0) to use for simulation)
-compare (compares simulation results running once on the default GPU and once on the CPU)
-cpu (run n-body simulation on the CPU)
-tipsy= (load a tipsy model file for simulation)
NOTE: The CUDA Samples are not meant for performance measurements. Results may vary when GPU Boost is enabled.
Error: only 0 Devices available, 1 requested. Exiting.
Diagnostic Logs
root@DESKTOP-UOKSUTD:~# sudo docker run --gpus all nvcr.io/nvidia/k8s/cuda-sample:nbody nbody -gpu -benchmark Run "nbody -benchmark [-numbodies=]" to measure performance.
-fullscreen (run n-body simulation in fullscreen mode)
-fp64 (use double precision floating point values for simulation)
-hostmem (stores simulation data in host memory)
-benchmark (run benchmark to measure performance)
-numbodies= (number of bodies (>= 1) to run in simulation)
-device= (where d=0,1,2.... for the CUDA device to use)
-numdevices= (where i=(number of CUDA devices > 0) to use for simulation)
-compare (compares simulation results running once on the default GPU and once on the CPU)
-cpu (run n-body simulation on the CPU)
-tipsy= (load a tipsy model file for simulation)
NOTE: The CUDA Samples are not meant for performance measurements. Results may vary when GPU Boost is enabled.
Error: only 0 Devices available, 1 requested. Exiting.