This project is deprecated. xmr-stak is the successor to xmrMiner and Dockerfiles now exist in the codebase.

docker-xmrminer

Docker images for running psychocrypt's Monero miner: xmrMiner inside a NVIDIA CUDA enabled Docker container. Operating system targets for this image are limited to Linux kernels at this point since nvidia-docker is only available for these architectures so far.

By using the nvidia-docker wrapper for the Docker CLI, we can achieve true containerisation of a CUDA application and avoid the following:

1. The host machine requiring CUDA installed
2. Passing the NVIDIA devices as command-line arguments to the Docker engine

See the nvidia-docker "Motivation" Wiki entry for more detail.

Prerequisites

1. Verify you have a CUDA-enabled GPU
2. Install Docker
3. Install nvidia-docker

Selecting a Docker image

Running a CUDA container requires a machine with at least one CUDA-capable GPU and a driver compatible with the CUDA toolkit version you are using.

The machine running the CUDA container only requires the NVIDIA driver, the CUDA toolkit doesn't have to be installed on the host.

NVIDIA drivers are backward-compatible with CUDA toolkits versions so please update your NVIDIA drivers and use the cuda-8.0 image. The 2.0 and 2.1 GPUs will soon be deprecated in CUDA, and those cards are unlikely to be profitable mining hardware at this point anyway.

CUDA toolkit version	Driver version	GPU architecture	Docker image
6.5	>= 340.29	>= 2.0 (Fermi)	Unsupported - please use a later driver
7.0	>= 346.46	>= 2.0 (Fermi)	May be supported - check back later
7.5	>= 352.39	>= 2.0 (Fermi)	May be supported - check back later
8.0	== 361.93 or >= 375.51	== 6.0 (P100)	aaronmboyd/xmrminer:latest
8.0	>= 367.48	>= 2.0 (Fermi)	aaronmboyd/xmrminer:latest

Quickstart

1. Pull from the Docker repository:

docker pull aaronmboyd/xmrminer

2. Run the container using nvidia-docker run instead of docker run, and pass in all the same arguments for your pool and username to xmrminer as normal.

• Note - you can isolate the GPUs you want to use in this container by passing in the attribute NV_GPU=0 for the first GPU, NV_GPU=0,1 for the first two, etc. This may be useful where you have different GPUs and need to adjust the thread/block --launch parameter for each GPU. This command is not required though, nvidia-docker will probe the available devices for you if you omit this argument. See nvidia-docker documentation for more on this syntax.

Example run command

nvidia-docker run aaronmboyd/xmrminer --url=stratum+tcp://mine.xmrpool.net:5555 --launch=10x24 --bfactor=4 --bsleep=100 --user=44NxHdzAJPVZkfXGnRd7kiGc1xCrg3GPncMwECKCmfbXRhVqhTreT7a2DGWcwCD3f7FnDsu1eCYusaTJoaETPajD3dPTdpQ -p docker_worker_1 --donate=1

Example run command isolating the first GPU only

  NV_GPU=0 nvidia-docker run aaronmboyd/xmrminer --url=stratum+tcp://mine.xmrpool.net:5555 --launch=10x24 --bfactor=4 --bsleep=100 --user=44NxHdzAJPVZkfXGnRd7kiGc1xCrg3GPncMwECKCmfbXRhVqhTreT7a2DGWcwCD3f7FnDsu1eCYusaTJoaETPajD3dPTdpQ -p docker_worker_1 --donate=1

Troubleshooting installation of nvidia-docker (Debian/Ubuntu)

nvidia-modprobe missing? If you see something similar to the following message on install;

Selecting previously unselected package nvidia-docker.
(Reading database ... 308844 files and directories currently installed.)
Preparing to unpack .../nvidia-docker_1.0.1-1_amd64.deb ...
Unpacking nvidia-docker (1.0.1-1) ...
Setting up nvidia-docker (1.0.1-1) ...
Setting up permissions
Job for nvidia-docker.service failed because the control process exited with error code. See "systemctl status nvidia-docker.service" and "journalctl -xe" for details.
nvidia-docker.service couldn't start.

Now try systemctl status nvidia-docker.service. If you now see:

...systemd[1]: Failed to start NVIDIA Docker plugin.
...systemd[1]: nvidia-docker.service: Unit entered failed state.
...systemd[1]: nvidia-docker.service: Failed with result 'exit-code'.
...systemd[1]: nvidia-docker.service: Service hold-off time over, scheduling restart.
...systemd[1]: Stopped NVIDIA Docker plugin.
...systemd[1]: nvidia-docker.service: Start request repeated too quickly.
...systemd[1]: Failed to start NVIDIA Docker plugin.

Try to start the service manually: sudo nvidia-docker-plugin . If you now see:

nvidia-docker-plugin | <timestamp> Loading NVIDIA unified memory
nvidia-docker-plugin | <timestamp> Error: Could not load UVM kernel module. Is nvidia-modprobe installed?

Verify you have nvidia-modprobe installed. If not, install with sudo apt-get install nvidia-modprobe and retry the nvidia-docker install.

Enjoy hashing!

Further reading

• nvidia-docker
• xmrMiner GitHub
• xmrMiner on Reddit
• CUDA Home

Donations

Thanks to psychocrypt for the work on xmrMiner:

• XMR: 43NoJVEXo21hGZ6tDG6Z3g4qimiGdJPE6GRxAmiWwm26gwr62Lqo7zRiCJFSBmbkwTGNuuES9ES5TgaVHceuYc4Y75txCTU

Donations for work on dockerizing are accepted at:

• XMR: 44NxHdzAJPVZkfXGnRd7kiGc1xCrg3GPncMwECKCmfbXRhVqhTreT7a2DGWcwCD3f7FnDsu1eCYusaTJoaETPajD3dPTdpQ