• State Resolved
  • Locked Locked
  • Replies 7 replies
  • Subscribers 98 subscribers
  • Views 604 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.

AM5718: QSPI configuration port access speed

Nomo
Guru 16800 points
Part Number: AM5718

Hello,

It seems that the access speed via the configuration port of QSPI is twelve times slower than via the memory mapped port.
We have two questions for this.

1.The cause will be the difference of the access port and procedure.
  Do you think there are any different causes for this slow access via the configuration port?

2.Is there any method to improve the access speed?

The SDK is PROCESSOR-SDK-LINUX-RT-AM57X and the version is v06_00_00_07.

Best Regards,
Nomo

  • 0
    TI__Mastermind 44635 points

    Hello Nomo,

    I am wondering if you could please provide more information on the process used to measure the access speed. 

    Regards,
    Krunal

  • 0 in reply to Krunal Bhargav34
    Guru 16800 points

    Hello Krunal-san,

     

    Thank you for your reply.

    We've monitored the signal between AM5718 and QSPI-Flash.

     

    In case of the communication via the memory mapped port, the communication seems burst transfer.

    However, in case of the communication via the configuration port, there is 500ns interval each 2byte transfer.

     

    Best Regards,

    Nomo

  • 0 in reply to Nomo
    TI__Mastermind 44635 points

    Hello Nomo,

    I am wondering if you could please provide more information on which driver you are using (ti_qspi/mx66l51235l or custom drivers) and are the read/write test performed in Uboot or Linux? 

    Regards,
    Krunal

  • 0 in reply to Krunal Bhargav34
    Guru 16800 points

    Hello Krunal-san,

    Thank you for your reply.
    The driver of "ti_qspi" is used to measure the access time.
    And, the command to measure the access time is the "time" command on Uboot.
    The results are following.

    % time sf read 0x83000000 0x00000000 0x01000000 device 0 offset 0x0, size 0x1000000
    SF: 16777216 bytes @ 0x0 Read: OK
    time: 0.724 seconds

    % time sf read 0x83000000 0x04000000 0x01000000 device 0 offset 0x4000000, size 0x1000000
    SF: 16777216 bytes @ 0x4000000 Read: OK
    time: 8.838 seconds

    Best Regards,
    Nomo

  • 0 in reply to Nomo
    TI__Mastermind 44635 points

    Hi Nomo,

    I am internally checking with our QSPI driver experts and I will keep you posted on any updates.

    Regards,
    Krunal

  • 0 in reply to Krunal Bhargav34
    TI__Mastermind 44635 points

    Hi Nomo,

    Based on my internal discussion, this is an excepted behavior. The config port does not use DMA and hence cannot use the entire buswidth available on the interconnect. The memory mapped mode uses DMA and is faster than the config port. We always recommend to use the memory mapped mode and I am wondering if there is a reason why are using the config port.

    Regards,
    Krunal

  • 0 in reply to Krunal Bhargav34
    Guru 16800 points

    Hello Krunal-san,

    Thank you for your reply.
    I could understand what you described.

    We want to use 128Mbyte QSPI flash.
    Memory mapped mode is only used for the 64Mbyte address space, so we think another 64Mbyte address space should be accessed via the config port.

    Best Regards,
    Nomo