• State TI Thinks Resolved
  • Locked Locked
  • Replies 2 replies
  • Answers 1 answer
  • Subscribers 83 subscribers
  • Views 174 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.

TMS320F28384D: --fp_mode=relaxed with --float-support=fpu64

Patrick Sullivan1
Prodigy 105 points

Part Number: TMS320F28384D

Tool/software:

The compiler manual for c2000 states "Relaxed floating-point mode causes double-precision floating-point computations and storage to be converted to single-precision floating-point where possible." I would like to know if this conversion will still take place when --float_support=fpu64 is also set. Also, is any further explanation available for what is meant by "where possible". 

On a variant such as the TMS320F28384D that has hardware support for fpu64 are single precision floating point operations still expected to be significantly faster? 

  • 0
    Prodigy 105 points

    After reading the documentation description of fp_mode again I think I understand better now that the bulleted list describes the only situations where the double to single conversion take place. Is this correct?

  • 0 in reply to Patrick Sullivan1
    TI__Guru**** 252520 points

    Use --float_support=fpu64 because your CPU supports it.  

    You are welcome use --fp_mode=relaxed as well.  We don't have any tests which show, when combined with --float_support=fpu64, whether code runs faster, by how much, etc.  That said, our guess is that --fp_mode=relaxed causes a small improvement.

    Thanks and regards,

    -George