• State
  • Locked Locked
  • Replies 1 reply
  • Subscribers 83 subscribers
  • Views 758 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.

TMS320F280049: F280049 Flash code execution performance

Jordan Zhou
Genius 9790 points
Part Number: TMS320F280049

Hi,

Customer is evaluating F28004x, and want to know the Flash code executing performance, comparing to RAM.

Because of lack of RAM size, we are thinking of putting the code in Flash, instead of copying to RAM and executing. What will be the efficiency gap between Flash and RAM? We can take FOC as an example.

What about the situation in F28x7x?

Another question about CLA performance? We found some gaps between C28x and CLA? The same code (pure fix point), CLA costs more time than C28x. What could be the issue?

What about the situation in F28x7x?

Thanks a lot.

Br, Jordan

  • 0
    TI__Guru* 87105 points

    Jordan,

    Regarding Flash execution performance compared to that of RAM:

    1) For F2837xD, Flash accesses incur waitstates after 50MHz (see Table 5-19. Flash Wait States in data manual).  However, sequential code (which is the case for most of the control code) will take advantage from the prefetch mechanism in the flash wrapper.  See below paragraph taken from data manual.    

    The on-chip flash memory is tightly integrated to the CPU, allowing code execution directly from flash through 128-bit-wide prefetch reads and a pipeline buffer. Flash performance for sequential code is equal to execution from RAM. Factoring in discontinuities, most applications will run with an efficiency of approximately 80% relative to code executing from RAM. This flash efficiency lets designers realize a 2× improvement in performance when migrating from the previous generation Delfino MCUs.

    2) F28004x Flash wrapper also supports the prefetch mechanism but waitstates are incurred for frequencies above 20MHz.  Please check datamanual for details.

    Thanks and regards,
    Vamsi