• State
  • Locked Locked
  • Replies 3 replies
  • Subscribers 82 subscribers
  • Views 858 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.

TMS320F28377S: performance CLA vs performance CPU

Lionel Franco
Prodigy 50 points
Part Number: TMS320F28377S
Other Parts Discussed in Thread: TMS320C28344

Hi,

I am working on a new design based on the Delfino TMS320F28377S (200Mhz).

I would like to reuse existing code which is running on the Delfino TMS320C28344 (300Mhz).

This code is written in assembly and quite efficient. I have benchmarked it on my existing platform and I am afraid that the 200Mhz of the TMS320F28377S, it will not be sufficient.

However, it might do the trick if I make good use of the CLA. This means that I will have to rewrite the code (different set of instructions) but it might still be acceptable.

What I am wondering about is "how fast will it run on the CLA?". How does the performance of the CLA compare to the performance of the Delfino CPU?

Thanks

  • 0
    Guru 192875 points

    Hi,

    Lionel Franco said:
    What I am wondering about is "how fast will it run on the CLA?". How does the performance of the CLA compare to the performance of the Delfino CPU?

    Check this wiki article:

    Regards,

    Gautam

  • 0 in reply to Gautam Iyer
    Prodigy 50 points
    Hi

    Is there any quantified information about the performance of the CLA running fixed-point operations compared to main CPU of the Delfino?

    Thanks

    Lionel
  • 0 in reply to Lionel Franco
    TI__Mastermind 20850 points
    Hi Lionel,

    We don't have any performance metrics on fixed point operations but, i can tell you right away it is going to be slow on the CLA. Most of the CLA instructions are floating point instructions, with a few to do integer math (ADD, SUB, CMP, LSL, LSR); there is no integer multiply.

    If you cannot do the entire algorithm in float you would have to convert your IQ number to float, normalize, do your computation and then convert back to IQ. The CLA is great for straight line computation, not so good for branching code (if-elseif-else, switch, ()?():()).