Usage on the Docker Client

Usage Instructions

• Ascend Docker Runtime supports mounting physical and virtual chips. Before mounting virtual chips, refer to Creating vNPUs to virtualize physical chips. Both static virtualization and dynamic virtualization of physical chips are supported.
• You can query the currently available physical chip IDs by running the ls /dev/davinci* command, and query the currently available virtual chip IDs by running the ls /dev/vdavinci* command.
• If you do not need to mount all the content in the default configuration file /etc/ascend-docker-runtime.d/base.list of Ascend Docker Runtime, create a custom configuration file (for example, hostlog.list) to reduce the mounted content. For details, see (Optional) Configuring Custom Mounted Content.

Mounting Chips Using Ascend Docker Runtime

In the examples, image-name:tag represents the image name and tag. For details about other parameters, see Table 1.

• Example 1: Mount the physical chip with chip ID 0 when starting the container.

  docker run -it -e ASCEND_VISIBLE_DEVICES=0 {image-name:tag} /bin/bash

• Example 2: Mount only the NPU device and management device when starting the container, without mounting driver-related directories.

  docker run --rm -it -e ASCEND_VISIBLE_DEVICES=0 -e ASCEND_RUNTIME_OPTIONS=NODRV {image-name:tag} /bin/bash

• Example 3: Mount the physical chip with chip ID 0 when starting the container, and read the mount content from the custom configuration file hostlog.list.

  docker run --rm -it -e ASCEND_VISIBLE_DEVICES=0 -e ASCEND_RUNTIME_MOUNTS=hostlog {image-name:tag} /bin/bash

• Example 4: Mount the chip with virtual chip ID 100 when starting the container.

  docker run -it -e ASCEND_VISIBLE_DEVICES=100 -e ASCEND_RUNTIME_OPTIONS=VIRTUAL {image-name:tag} /bin/bash

• Example 5: When starting the container, slice 4 AICores from the chip with physical chip ID 0 as virtual devices and mount them to the container.

  docker run -it --rm -e ASCEND_VISIBLE_DEVICES=0 -e ASCEND_VNPU_SPECS=vir04 {image-name:tag} /bin/bash

• Example 6: When starting the container, mount the chip with physical chip ID 0, and allow soft links in the mounted driver files (applicable only to Atlas 500 A2 intelligent station, Atlas 200I A2 accelerator module, and Atlas 200I DK A2):

  docker run --rm -it -e ASCEND_VISIBLE_DEVICES=0 -e ASCEND_ALLOW_LINK=True {image-name:tag} /bin/bash

After the container is started, run the following commands inside and outside the container to check whether the corresponding devices and drivers are mounted successfully. For the specific mount directory of each model, refer to Content Mounted by Ascend Docker Runtime. Example commands are as follows:

ls /dev | grep davinci* && ls /dev | grep devmm_svm && ls /dev | grep hisi_hdc && ls /usr/local/Ascend/driver && ls /usr/local/ |grep dcmi && ls /usr/local/bin

Possible outputs:

davinci0
davinci_manager
devmm_svm
hisi_hdc
include lib64
dcmi
npu-smi

Using Ascend Docker Runtime to Mount Chips and Other Devices

Use Ascend Docker Runtime to support containers running training, inference, or other tasks.

• Taking the Atlas 200I SoC A1 core board running an inference container as an example, modify the command according to the actual situation. The example is as follows, and the related parameters are shown in Table 1 and Table 2.

  docker run -it -e ASCEND_VISIBLE_DEVICES=0 --device=/dev/xsmem_dev:rwm --device=/dev/event_sched:rwm --device=/dev/svm0:rwm --device=/dev/sys:rwm --device=/dev/vdec:rwm --device=/dev/vpc:rwm --device=/dev/log_drv:rwm --device=/dev/spi_smbus:rwm --device=/dev/upgrade:rwm --device=/dev/user_config:rwm --device=/dev/ts_aisle:rwm --device=/dev/memory_bandwidth:rwm -v /var/dmp_daemon:/var/dmp_daemon:ro -v /var/slogd:/var/slogd:ro -v /var/log/npu/conf/slog/slog.conf:/var/log/npu/conf/slog/slog.conf:ro -v /usr/local/Ascend/driver/tools:/usr/local/Ascend/driver/tools -v /usr/local/Ascend/driver/lib64:/usr/local/Ascend/driver/lib64 -v /usr/lib64/aicpu_kernels:/usr/lib64/aicpu_kernels:ro -v /sys/fs/cgroup/memory:/sys/fs/cgroup/memory:ro -v /usr/lib64/libyaml-0.so.2:/usr/lib64/libyaml-0.so.2:ro -v /etc/ascend_install.info:/etc/ascend_install.info -v /usr/local/Ascend/driver/version.info:/usr/local/Ascend/driver/version.info workload-image:v1.0 /bin/bash

  NOTE

  • If the driver of the Atlas 200I SoC A1 core board is version 1.0.0 (Ascend HDK 22.0.0) or earlier, you need to mount /dev/xsmem_dev and /dev/event_sched.
  • If the driver of the Atlas 200I SoC A1 core board is later than version 1.0.0 (Ascend HDK 22.0.0), you do not need to mount /dev/xsmem_dev and /dev/event_sched.

• Taking the Atlas 500 A2 intelligent station running an inference container as an example, modify the command according to the actual situation. The example is as follows, and the related parameters are shown in Table 1 and Table 2.

  docker run --rm -it -e ASCEND_VISIBLE_DEVICES=0 -e ASCEND_ALLOW_LINK=True workload-image:v1.0 /bin/bash

Mounting Chips and Other Devices Without Ascend Docker Runtime

• Taking the Atlas 200I SoC A1 core board running an inference container as an example, modify the command according to the actual situation. The example is as follows, and the related parameters are shown in Table 2.

  docker run -it --device=/dev/davinci0:rwm --device=/dev/xsmem_dev:rwm --device=/dev/event_sched:rwm --device=/dev/svm0:rwm --device=/dev/sys:rwm --device=/dev/vdec:rwm --device=/dev/venc:rwm --device=/dev/vpc:rwm --device=/dev/davinci_manager:rwm --device=/dev/spi_smbus:rwm --device=/dev/upgrade:rwm --device=/dev/user_config:rwm --device=/dev/ts_aisle:rwm --device=/dev/memory_bandwidth:rwm -v /etc/sys_version.conf:/etc/sys_version.conf:ro -v /usr/local/bin/npu-smi:/usr/local/bin/npu-smi:ro -v /var/dmp_daemon:/var/dmp_daemon:ro -v /var/slogd:/var/slogd:ro -v /var/log/npu/conf/slog/slog.conf:/var/log/npu/conf/slog/slog.conf:ro -v /etc/hdcBasic.cfg:/etc/hdcBasic.cfg:ro -v /usr/local/Ascend/driver/tools:/usr/local/Ascend/driver/tools -v /usr/local/Ascend/driver/lib64:/usr/local/Ascend/driver/lib64 -v /usr/lib64/aicpu_kernels:/usr/lib64/aicpu_kernels:ro -v /sys/fs/cgroup/memory:/sys/fs/cgroup/memory:ro -v /usr/lib64/libyaml-0.so.2:/usr/lib64/libyaml-0.so.2:ro -v /etc/ascend_install.info:/etc/ascend_install.info -v /usr/local/Ascend/driver/version.info:/usr/local/Ascend/driver/version.info workload-image:v1.0 /bin/bash

  • If the driver of the Atlas 200I SoC A1 core board is version 1.0.0 (Ascend HDK 22.0.0) or earlier, you need to mount /dev/xsmem_dev and /dev/event_sched.
  • If the driver of the Atlas 200I SoC A1 core board is a version later than 1.0.0 (Ascend HDK 22.0.0), you do not need to mount /dev/xsmem_dev and /dev/event_sched.

• To run inference tasks on the Atlas 500 A2 intelligent station without using Ascend Docker Runtime, refer to the "Starting the Container" section in "Deploying Ascend Software (Customized System Scenario) > Container Deployment > Creating a Container Image" in the Atlas 500 A2 Intelligent Station Ascend Software Installation Guide.

Parameter Description

Table 1 Ascend Docker Runtime running parameters

Table 2 Explanation of other parameters