× We want your feedback about Neuron SDK! Let us know by taking the Neuron survey

Neuron Top User Guide

Overview

neuron-top provides useful information about NeuronCore and vCPU utilization, memory usage, loaded models, and Neuron applications.

Note

If you are parsing neuron-top output in your automation environment, you can now replace it with neuron-monitor (Neuron Monitor User Guide) which outputs data in a standardized, easier to parse JSON format.

Using neuron-top

Command line arguments

Launching neuron-top with no arguments will show data for all ML Applications running with Neuron Runtime 2.x. neuron-top will also show data for any Neuron Runtime 1.x (neuron-rtd daemon) running on the default GRPC address - either $NEURON_RTD_ADDRESS or unix:/run/neuron.sock if that environment variable is not defined.

If more than one neuron-rtd daemons are running, each of their GRPC addresses needs to be specified in the command line:

neuron-top unix:/run/neuron2.sock unix:/run/neuron3.sock

User interface

The user interface is divided in 4 sections. The data shown in these sections only applies to the currently selected Neuron application:

image0

  • The Neuroncore Utilization section shows the NeuronCore utilization for the currently selected Neuron application.

  • The VCPU and Memory Info section shows:

    • Total system vCPU usage - the two percentages are user% and system%

    • Runtime vCPU usage - same breakdown

    • Runtime Memory Host - amount of host memory used by the Application and total available

    • Runtime Memory Device - amount of device memory used by the Application

  • Loaded Models is a tree view which can be expanded/collapsed. The columns are:

    • Model ID - an Application-level identifier for this model instance

    • Host Memory - amount of host memory used by the model, displayed hierarchically, where the ‘parent’ value is the sum of its ‘children’

    • Device Memory - amount of device memory used by the model, displayed just like the Host Memory

Note

The up/down/left/right keys can be used to navigate the tree view. The ‘x’ key expands/collapses the entire tree.

The bottom bar shows which application is currently displayed by highlighting its tag using a yellow font and marking it using a pair of ‘>’, ‘<’ characters. For neuron-rtd daemons, the GRPC address will be shown here instead of the tag:

image1

Note

The ‘1’-‘9’ keys select the current application.’a’/’d’ selects the previous/next application on the bar.