• State TI Thinks Resolved
  • Locked Locked
  • Replies 9 replies
  • Answers 1 answer
  • Subscribers 98 subscribers
  • Views 676 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.

RTOS/TMDSIDK437X: QSPI boot time optimization

Haxor
Intellectual 565 points
Part Number: TMDSIDK437X

Tool/software: TI-RTOS

Hi folks,

i am using the idkam437x with the EtherCat_Slave_Full_ Demo and im booting via Qspi from flash.

Now it takes approx. up to 5s from powerup to operational state in TwinCAT.

1. could someone confirm the approx. same time?

2. is there a possibility to speed up this timing behaviour ?

because we will have a robotic tool changer in our application where bootup time from powerup is essential.

thx in advance     

  • 0
    TI__Mastermind 43296 points

    Hi Haxor,

    Did you also measure the time of binary (app) boot to main() vs. TwinCAT detecting slave to OP state?

    Whats the maximum QSPI boot time required in your application? You are looking for not only the second boot loader optimization but also overall system optimization, correct?

    Regards,

    Garrett

  • 0 in reply to Garrett Ding
    TI__Mastermind 43296 points

    Haxor,

    If you look into the SBL source - pdk_am437x_1_0_14\packages\ti\starterware\bootloader\src, there is not much space for optimization to reduce the boot time. Are you able to conclude the boot timing requirement?

    Regards,

    Garrett

  • 0 in reply to Garrett Ding
    Intellectual 565 points

    Hi Garrett,

    you are correct we want too reduce the overall boot time wich for us is the time from startup till we are in Operational state in our PLC.

    we need to reach a maximum time of 3sec (better <3s) but right now we have approx. 5s.

    And this with just the etherCat example running right now, so i expect longer boot time when the code will be expanding in the future adding more features.

    do you see any way to reduce this time ?

       

  • 0 in reply to Haxor
    TI__Mastermind 43296 points

    Hi Haxor,

    I am currently out of the office, will continue to investigate this after I return to the office in the July.

    Regards,

    Garrett

  • 0 in reply to Garrett Ding
    TI__Mastermind 43296 points

    Haxor,

    Any progress/update from your side? I am back in the office, will be looking into this ...

    Regards,

    Garrett

  • 0 in reply to Garrett Ding
    Intellectual 565 points

    Hi, i didn't make any effort in this direction so far, any help is apprechiated 

    thx

    haxor

  • 0 in reply to Haxor
    TI__Mastermind 43296 points

    haxor,

    Which model of PLC are you using for the test? As this is 'seconds' level optimization, the QSPI boot time self may have minimum impact.

    Regards,
    Garrett

  • 0 in reply to Garrett Ding
    Intellectual 565 points

    Hi Garrett,

    i think i dont understand your Question correctly, but i try to answer we use Twincat3 as Ethercat Master

    running on a buscycle of 4ms with only one Slave the idkam4379x with an custom esi file with Testdata of 64Bytes Input and 64 Bytes Output (will be more later)

    Then i powerup the board and measure the time still the Slave is in Op 

    thx 

    Haxor

  • 0 in reply to Haxor
    TI__Mastermind 43296 points

    Haxor,

    Some optimization might be possible to reduce the boot up time, below is high level considerations:

    1. disable UART print

    2. remove unused functions like EOE, FOE...if applicable

    3. profile and benchmark the application to move time consuming functions to internal L2 memory from DDR. The application note may help configure the memory sections - www.ti.com/lit/an/sprac45a/sprac45a.pdf

    4. fine tune the compiler options

    I was able to confirm the SD card boot up time takes about the same 5 seconds from power cycle to OP state, where 2 seconds is from 'Frame returned -> force initialization' to 'Communication re-established' as shown in your snapshot.

    Regards,

    Garrett