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.

AM68A: Sitara AM68A and AM69A vs Jacinto TDA4x

Part Number: AM68A
Other Parts Discussed in Thread: AM69A, , TDA4VM

Dear TI Community,

Looking at the following overview of TI's ARM processors :, the Sitara AM68A and AM69A are in the same column with Jacinto TDA4x, but it seems that this later is reserved for automotive industry thus is a little bit difficult to access its resources for companies who expect to make products different than automotive (such as smart cameras for surveillance, video analytics or autonomous robotics, ... with smaller quantities).

Could AM68A or AM69A be considered as a more "general purpose" MPU compared to TDA4x, please ?

Thanks in advance and best regards,

Khang Le Truong

Engineering Manager,


  • Hello Khang,

    To begin the AM68A and AM69A are based off the Jacinto 7 architecture and thus do not full under the Sitara category of devices. 

    The design of the AM68A and AM69A processor family is focused primarily on smart vision camera applications in general environments. The focus of the resources in this group are geared to optimize for vision processing and AI applications. This leads to the the ability to greatly leverage functionalities such as the deep learning accelerator and vision processing accelerator over the TDA4x devices.

    The TDA4x devices are instead built to optimize for ADAS and autonomous vehicle applications. This means that other considerations such as security, connectivity and specialized I/O ports are also necessary to optimize for such purposes.

    The two processor families have great overlap in their architecture as they were both built off the Jacinto 7 architecture however there target applications are different even though there are similar capabilities in both line of products. These products are built in mind to differentiate themselves in key areas and thus are difficult to be considered "general purpose". More details would need to be provided on the intended use case as TI offers a broad range of products that are designed from its' inception to be general purpose such as the MSP line of products as an example.


  • Hello Khang,

    Please read my previous response and let me know should you need more direction.


  • Hi ,

    Thanks very much for your clarification. In fact, the requirements of our stereo camera for indoor/outdoor autonomous robots is below :

    - On-chip ISP for RAW Bayer sensors (2 or more),

    - On-board/on-chip hardware accelerated depth mapping,

    -  Multiple H264/H265 4K/1080 encoding streams,

    - Visual SLAM capability,

    - NPU for vision based deep learning features such as situational awareness, obstacle/object avoidance, ...

    TDA4VM seems to be a great match with its interesting with its hardware accelerated Stereo depth estimation, key feature of stereo camera. However, as your colleague explained, the DMPAC hardware accelerator is only available in TDA4VM but not in any other AM6xA. Therefore, AM68A and AM69A seem not to fulfill the above requirements even-though they are found in the same column with TDA4VM, except that there would be another way to accelerate the depth calculation which is heavily computational.

  • Hello Khang,

    Given the complexity, revisions and scale of the processor product family , these columns/figures are geared to give an overall summary at a glance rather than get into specific case-by-case details. The datasheets and app notes found for each product offering serve as a better resource to understand the best fit for product use cases. Granted these could be long at times, even the first key feature description pages found on the datasheet would grant a better understanding beyond the column/figures. 

    I will review these product specifications further along with my colleagues thread and return with further insight. 


  • Hello Khang,

    The C7x DSP can be used for acceleration purposes other than deep learning however currently there is no existing documentation regarding other applications.

    As you concluded , yes the TDA4VM appears to be the best fit product for your needs.