Device Information Acquisition

Tool Introduction

After profile data collection is complete, you can use the get_msprof_info.py script tool to obtain device information from the device_{id} or host directory under the PROF_XXX directory. The following table describes the functionality and path get_msprof_info.py.

Procedure

1. Log in to the environment where the Ascend-CANN-Toolkit is located as the running user.
2. Switch to the directory where the msprof.py script is located.

   ${INSTALL_DIR}/tools/profiler/profiler_tool/analysis/msprof. Replace ${INSTALL_DIR} with the actual CANN component directory. If the Ascend-CANN-Toolkit package is installed as the root user, the CANN component directory is /usr/local/Ascend/ascend-toolkit/latest.

   

   Quick tip: Create an alias for the msprof.py script with the command alias msprof_analysis='python3 msprof.py_script_directory' as the running user. Then, you can start profiling with the shortcut msprof_analysis in any directory. This operation takes effect only in the current window.

3. Run the get_msprof_info.py script. An example command is as follows:
   • Non-cluster scenario

     python3 get_msprof_info.py -dir /home/1/PROF_000001_20220129014731273_KEDKPORHMAGPGD/device_0

   • Cluster scenario

     python3 get_msprof_info.py -dir /home/1/

   Table 2 Command-line options

   Option	Description	Required/Optional
   -dir, --collection-dir	Directory of collected profile data. In a non-cluster scenario, set this parameter to the host or device_{id} directory under PROF_XXX. In a cluster scenario, set this parameter to the parent directory of PROF_XXX.	Required
   -h, --help	Help information.	Optional

4. View the output.

   In a non-cluster scenario, the output is displayed on a screen, as shown in Figure 1. Table 3 describes corresponding fields. In a cluster scenario, the /query/cluster_info.json file is generated in the directory specified by -dir to store information about each node, as shown in Figure 2. Table 4 describes corresponding fields.

   Figure 1 Device information (non-cluster scenario)
   

   Table 3 Field description (non-cluster scenario)

   Field	Description
   collection_info	Information collection.
   Collection end time	End time of information collection.
   Collection start time	Start time of information collection.
   Result Size	Size of the information data, in MB.
   device_info	Device information.
   AI Core Number	Number of AI Cores.
   AI CPU Number	Number of AI CPUs.
   Control CPU Number	Number of Ctrl CPUs.
   Control CPU Type	Ctrl CPU Type.
   Device Id	Device ID.
   TS CPU Number	Number of TS CPUs.
   host_info	Host information.
   cpu_info	Host CPU information.
   CPU ID	Host CPU ID.
   Name	Host CPU name.
   Type	Host CPU type.
   Frequency	Host CPU frequency.
   Logical_CPU_Count	Number of logical CPUs on the host.
   cpu_num	Number of Host CPUs.
   Host Computer Name	Host device name.
   Host Operating System	Host OS.
   model_info	Model information.
   Device Id	Device ID.
   iterations	Iteration statistics.
   Iteration Number	Number of iterations.
   Model Id	Model ID, which is displayed based on the number of models.
   version_info	Version information.
   analysis_version	Parsing version information.
   collection_version	Collection version information.
   drv_version	Driver version information.

   Figure 2 Device information (cluster scenario)
   

   Table 4 Field description (cluster scenario)

   Field	Description
   Rank Id	Node ID that uniquely identifies a device in the cluster scenario.
   Device Id	Device ID, which is not the unique identifier of a device in the cluster scenario.
   Prof Dir	In the cluster scenario, the current rank ID corresponds to the PROF_XXX directory on the device.
   Device Dir	device_{id} directory in the PROF_XXX directory in the cluster scenario.
   Models	Model information, including all model IDs of the device corresponding to the current rank ID and the number of iterations in the model.