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.

TMS320F28P650DH: Benchmark for addition, subtraction, multiplication, division, and trigonometric function operations in CPU and CLA

Part Number: TMS320F28P650DH

Tool/software:

Hi Champs,

This question is for F28P65 and F28P55.

Now my customer need to optimize their code and need the below benchmark in CPU and CLA. They need to know how many cycles are required to execute the instruction.

1. a+b

2. a-b

3. a×b

4. a/b

5. sin, cos, tan, asin, acos, atan

Please help provide the data, thank you.

Best Regards,

Julia

  • Hi Julia,

    Assuming you are referring to floating-point:

    For 1 to 3, refer to 1.5.2 of https://www.ti.com/lit/spruhs1 (from that list, you can go to the specific ADD SUB MPY instruction and look at cycles). Each is a 2p cycle operation. "p" refers to pipeline cycles i.e. the 2nd instruction can be an independent instruction unrelated to this operation.

    For 4 to 5 (sin, cos, atan), refer to 7.5.3 of the same doc. sin, cos, atan take 4p cycles each. Division takes 5p cycles.

    For asin, acos, you will leverage asin = atan (x/ sqrt(1-x^2)) and acos = atan (sqrt(1 - x^2/x), so they should take about 18 cycles each.

    For tan, you will do sin/cos, so that should be about 9 cycles.

    Please note all of the above only refer to the compute, not any associated loads and stores and moves. Those will add additional cycles.

    Thanks,

    Sira