• State TI Thinks Resolved
  • Locked Locked
  • Replies 1 reply
  • Answers 1 answer
  • Subscribers 101 subscribers
  • Views 225 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

SK-TDA4VM: EDGEAI: A72 MPU loading

Gunter Schmer
Genius 14067 points
Part Number: SK-TDA4VM

Hi,

looking through the EDGEAI SDK docs and specifically the Flows and Datasheet sections

http://software-dl.ti.com/jacinto7/esd/processor-sdk-linux-sk-tda4vm/08_00_01_10/exports/docs/data_flows.html

http://software-dl.ti.com/jacinto7/esd/processor-sdk-linux-sk-tda4vm/08_00_01_10/exports/docs/datasheet.html

For some of those examples the A72 comes in at about 20-30% loading. Is that loading mainly due to post-processing and SW mosaicing?

Or does the A72 get it's loading from gstreamer activity or any TIDL layer compute?

Thanks!

--Gunter

  • 0
    TI__Mastermind 36452 points

    Hello Gunter,

    In the first data flows page that you linked, the blue boxes and the gray boxes which denotes that they are "GStreamer plugins on ARM" and "Compute on ARM" respectively are the things that are loaded to the A72.

    For example, on the image classification dataflow below the following are loaded on the A72:

    • GStreamer plugins: v4l2src, jpegdec, two instances of queue, videobox, two instances of appsink, and appsrc.
    • Post-processing and sw-mosaic, which is any computing done after the inferencing occurs to add onto each frame a bounding box around detected objects, display the top N classified objects, or a mask on regions depending on the use case and position the different components of the output into a frame.

    Everything else that is green or red are processes that are offloaded to our hardware accelerators or DSP so that they are not loaded onto the A72.

    Regards,

    Takuma