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.

Edge AI demos on AM6xA family

Other Parts Discussed in Thread: TDA4VM, AM68A

Dear TI processor experts,

Looking at the TI Edge AI demos training, I saw that there's examples mainly for TDA4VM platform:

Smart Cameras and AI Boxes

Autonomous Machines and Robots

As a camera maker whose products include stereo camera, we are really interested in the hardware accelerated Stereo depth estimation feature. We would like to know if it would be feasible to have same acceleration for that feature on any of your new SoCs such as AM62A potentially, AM68A preferably, please ?

Thanks in advance and best regards,

Khang Le Truong,

Engineering Manager,
Dynimlabs

  • Hello,

    The AM62A processor does not have hardware accelerated depth estimation. This requires the DMPAC hardware accelerator. I do not believe the AM68A has this accelerator on chip either, according to the datasheet.

    The TDA4VM is a more mature processor and has more example applications as a result of this. We are currently growing our set of examples and reference designs for new processors.

    Best,
    Reese

  • Dear ,

    Thanks for confirming my doubt about the necessity and the availability of DMPAC on the SoCs other than TDA4VM. I wonder if the C7x DSP could be used for acceleration purpose(s) other than Deep Learning acceleration ?

    Khang

  • Hello Khang,

    The C7xMMA is currently used purely for Deep Learning acceleration on AM62A. However, the AM68A includes a C7x DSP without MMA. I do not have a timeline to share on general-purpose programming support for that DSP. I'm looping in from the support team on that processor to comment.

    Best,
    Reese