• State Resolved
  • Locked Locked
  • Replies 5 replies
  • Subscribers 31 subscribers
  • Views 1149 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.

TMS570LS0714: Lifetime of FEE on TMS570 chip?Is there any algorithm to make it equally used?

cc liu
Expert 1150 points
Part Number: TMS570LS0714

Dear All

We are using TMS570LS0714 for developing a safety-critical device.   I have several questions about FEE as follows:

How many times of FEE can be erased\writed？ Does it have same erased\writed cycles as flash(BANK0)?

Our assessor was worried about that, for example, we store some information on block1, and we use this block more frequently than others, will this block fatigue much faster than others?

But I find that the when I put some info on block1, everytime I write this info to block1, the address of this info changed. It seems that the data on FEE moves everytime when write.

Am I right?   If so, please verify my guess and please tell me where I can find some infomation about this technique, for example,tha name of this algorithm.

I am not sure I make my question clear. Please let me know if it is not clear.

Thanks.

Best Regards.

Leo

 

  • 0
    TI__Guru 59540 points

    Hello Leo,

    The FEE has some management of the W/E cycles built into it. I don't know the exact mechanism or if it requires you to implement certain levels of protection in your application code or not. I am copying our SW team lead so that he can address it from the SW side.

    From the Hardware side, there is a limit specified in the datasheet as shown below:

    Note there is a parallel table for program memory vs data memory above.

  • 0 in reply to Chuck Davenport
    Expert 1150 points
    Dear Chuck
    Thank you for your reply. I am looking forward to hear from SW experts.
    If it can endure 100000 times of write erase, and I divide my FEE into two blocks, each one can be writed 100000, so if I put my data into block1 and block2 in trun, I mean this time I put data on block1 and next time on block2, in a loop. Then FEE
    can be writed 200000 times before broken? Is that what I am to understand?
    Our exterior assessor require our device to operate more than million times. What I need is just some evidence in the datasheet or specification, to confirm them FEE in TMS570 can be written million times with some management mechanism you mentioned. I have to show them some textual evidences or a name of that mechanism if it is a well-known algorithm.
    Thanks and Regards,
    Leo
  • 0 in reply to cc liu
    Expert 1150 points
    PS:
    Maybe I should use this term “wear-leveling algorithms”. Do we have this built in already?
    Leo
  • 0 in reply to cc liu
    TI__Genius 14205 points

    Hi Leo,

    Device TMS570LS0714 Bank7 has 4 Sectors, which can support 2 EEPROM's through FEE driver. What this enables is the blocks which has frequent writes/updates can be put in EEPROM1 and the other Blocks can be put in EEPROM2. This way Sector 0 and 1 will be frequently used not Sector 2 and 3.  

    Also lesser the Block Size ( frequent update block) its better we can reduce Number of Erases. In order to figure out number of Sector Erase, Block size and freq of write is an important consideration. You can do the math based on your application and FEE useage.

    FEE Driver is not a true 'wear leveling algorithm' But you can relate working to Dynamic wear leveling until the Virtual sector is full, but Static wear leveling when the VIrtual sector is full, makes copy of all valid data to new virtual sector. 

  • 0 in reply to Prathap
    Expert 1150 points
    Thanks Prathap.
    Regards.