• State
  • Locked Locked
  • Replies 5 replies
  • Subscribers 83 subscribers
  • Views 1819 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

Tms320f28027 is floating point or fixed point processor

Ashutosh Pailwan1
Expert 2480 points
Other Parts Discussed in Thread: TMS320F28027

I am confused to understand whether tms320f28027 is floating point or fixed point processor. If it is fixed point what are its implication

  • 0
    Expert 1500 points

    It is only speed, in case you use float variables s/w library will be used as no floating point hardware.

    As you can see in the data sheet, it is a fixed point processor. No hardware support for floating point operation.

  • 0 in reply to Joy Philipz
    Expert 2480 points
    So is it required to use IQ maths library for floating variable operation for example to Carry out 0.9824 * 0.6543 multiplication opeartion
  • 0 in reply to Ashutosh Pailwan1
    TI__Guru 61235 points

    Ashutosh,

    Joy is correct.  F28027 is a fixed-pt. processor.  It has no floating point hardware.  If you code in floating point in C, the operations will use a software library that will take a lot of cycles to execute.  For example, I recall a floating point multiply takes about 100 cycles to perform using the RTS compiler library.  A floating pt. add takes something like 140 cycles.

    You are not "Required" to use IQmath library, but it will be a lot more efficient to do so.  Multiplying two IQ values will consume a few cycles, compared to the 100 cycles I mentioned above for floating pt. with the RTS library.

    Regards,

    David

  • 0 in reply to David M. Alter
    Expert 2480 points

    Other than time consumption dose it improve accuracy if I use IQ math for floating point computation.

  • 0 in reply to Ashutosh Pailwan1
    TI__Guru 61235 points

    Ahustosh,

    Ashutosh Pailwan1 said:

    Other than time consumption dose it improve accuracy if I use IQ math for floating point computation.

    The answer to this is complex.  The short answer is maybe.
    IQmath trades off dynamic range for precision.  It depends on the IQmath format you choose (where you put the binary point, i.e., the decimal point).  If you use a lot of precision and less dynamic range (e.g., say I4Q28 format), you can easily overflow when you do math operations.  If you don't use enough precision (e.g., I28Q4), you get imprecise results.
    I can tell you this.  The F28027 is a fixed-pt processor.  Other than maybe an isolated computation, you don't want to be doing floating point calculations on it.  The cycle costs will be enormous.  If you need floating point, look at the F28069 processor and its family.
    Regards,
    David