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.

Is TMS320 a DSP?

Other Parts Discussed in Thread: TMS320F28333

Are the TMS320F family dsp microcontrollers?  (Even though they aren't called "DSP.")

Texas Instruments used to call the TMS320 family devices DSP, but then they changed to calling them just microcontroller? Why?

Are they DSP? Do they have shift-add in single operation for sum-of-products expressions?

Is  Delfino a DSP?

(Also, can somebody fix the post form? It says TMS320F28333 is not a valid part number?)

Thanks.

  • Consider TMS320F2x family as DSPs with advanced microcontroller capabilities or as microcontrollers with advanced DSP capabilities optimized for specific real-time control purposes, which C6000 family doesn't have, but can be combined with each over to solve problems that require up to terraflops of MAC computational power or even build artificial neural networks with real-time image recognition (TDA3x).
    Each device has its own market place and was designed and optimized for specific area of usage. TMS320F2x family has pipelined CPU architecture and can execute highly complex operations per cycle. TMS320F2837x can even execute trigonometric instructions. So, would you call it then a "DSP"?
  • Hi 78sys,

    Alexey has it correct.  Debates like these often come down to the definition of terms (and sometimes semantics).

    For me, a DSP would tend to have no embedded non-volatile memory, would have specialized instructions for complex mathematical operations, and the software run on the device would tend toward being OS-based.  On the other hand, an MCU would tend toward the opposite.  The lines are blurry though - and they've gotten more blurry over time.

    The TMS320*28* series (C2000) - having embedded non-volatile memory (flash) and having specialized signal-processing instructions (and a modified Harvard bus architecture) - falls in between my definitions of DSP and MCU.  Software-wise, the majority of our customers do not use an RTOS, but we do have some that do.

    As a result, the C2000 family's marketing team lives with a conundrum.  If we call ourselves a DSP, customers who are looking for a high-end MCU may not look at us.  Similarly, if we choose to call ourselves an MCU, customers may not realize that we have DSP-type instructions.  In fact, when I joined C2000 in 2007, we called ourselves a DSC (Digital Signal Controller), a hybrid-term.  This had its own set of issues.

    Around 2010, we chose to market ourselves as an MCU.  This suits us pretty well.  

    Note that our marketing toward the DSP portion of the C2000's DNA can be found at the below:
    http://www.ti.com/microcontrollers/c2000-real-time-control-mcus/applications/sensing-signal-processing.html

    ===

    Regardless of what we call ourselves, our core markets are those involved in the control of power electronics.  Doing this well (and better) is a focus for C2000.  This involves improving the peripherals - specifically the ADC's throughput and accuracy, performance of its analog comparators, and enabling the generation of advanced PWM waveforms.  At the same time it also involves improving processing capabilities. One example of this is the TMU accelerator which can be found in many of our most recent device families.

    ===

    With respect to your technical question, yes, the C28x core does have signal processing instructions (MAC, etc).  Please see the following:
    http://www.ti.com/lit/spru430

    Hopefully this helps!


    Thank you,
    Brett

  • Yes, I consider it a DSP. I started using TI with the TMS320F240 programming is assembly, when they called it a dsp because it did shift-add, etc in a single instruction for sum-of-products implementation.