Developer Guide

  • 2022.3.0
  • 09/27/2022
  • Public Content

View status, output and performance

Depending on the configuration and behavior of your container application, you can analyze your workload in multiple ways. You can use this section as a reference to prepare your container as well.

Monitor Status

The status of your project can be monitored with
Queued
,
Deploying
,
Failed
or
Error
. Depending upon the expected behaviour of your container, the status can be in
Running
state for up to
30mins
or
Completed
if the program completes execution.

Container Logs

From the
Dashboard
view of the Container Playground, expand your project run and click on the
Log Output
icon. Use the
Refresh Stream
button in the pop-up to refresh the latest console logs of your container application while in the
Running
or
Completed
state.

Performance Dashboard

Once your project reaches
Completed
status, expand your project and click on
Performance Metrics
to view detailed CPU, GPU, memory utilization and temperature metrics for the entire duration of your workload.
  • The greyed out
    Performance metrics
    icon will be enabled within few minutes of completion, without any additional project configuration.
  • For projects launched with multiple containers on the same platform, the legend indicates
    multi_container_workload
    in resource sharing scenarios.
  • If you come across
    Telemetry Error
    or
    Telemetry Data expired
    message, please Request Support ( A product or service I already own or use > Search for a product or service by name > devcloud for the edge) .
You can also enable throughput (fps) and latency (ms) in the dashboard to quickly compare multiple runs on various Intel Platforms.
Your container must write out a
FP16/performance.txt
file in below format. You must also Configure imported containers with an
Output Mount Point
before launching.
Throughput: 104 FPS Latency: 13457.780 ms
The Intel® Developer Cloud for the Edge portal recursively looks for the
performance.txt
file and
FP16
directory in your
Mount Point
or
Result Path
directories.

Data in Filesystem

You must Configure imported containers with a
Mount Point
before launching to enable storing data into the filesystem for your container. From the
Dashboard
view of the Container Playground, expand your project and click on the
FileSystem
icon next to the
Log Output
to open the FileSystem view from the
My Library
tab.

Web Service URLs

If your containers expose web-services such as grafana dashboard that can be accessed in a browser, you can
Expose Routes
while configuring your imported containers.

Container Terminal Access

When the project is in
Running
state, you can gain access to the container terminal for easy debugging.

Product and Performance Information

1

Performance varies by use, configuration and other factors. Learn more at www.Intel.com/PerformanceIndex.