Forum Discussion

Altera_Forum's avatar
Altera_Forum
Icon for Honored Contributor rankHonored Contributor
17 years ago

GMII <-> MII bridge in MAX II CPLD ?

Hi Folks,

In my board, I would like to use GMII to MII bridge in MAX II CPLD. But I don't have idea of

> Whether it can be used in MAX II CPLD (LE utilization and device electrical characteristics support)?

> If it is possible, Then any vendor can provide the ip core of this bridge ?

> Do any one have design documents relating this or any suggestions from yours.

Regards,

Vijayan

5 Replies

  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor

    What's the intended purpose of the design? GMII should be MII downward compatible by specification, normally no bridge would be necessary.

  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor

    Sorry, There is a correction. It is RGMII <-> GMII.

    And in the design, CPLD should make this bridge between the processor and the switch
  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor

    OK, RGMII to GMII....Please ignore this post!!!

    Hi,

    I don't see any reason why a GMII to MII interface could not be implemeneted in a MAX II part.

    The CPLD should be able to be clocked fast enough (MII runs at 25MHz, GMII runs at 125MHz for Gbit ot 25Mhz for 100M) I assume you will run the GMII side at 10/100M rates to remove the need for traffic shaping/buffering in the CPLD in the GMII to MII direction)

    The MZX II i/o should be suitable for both MII and GMII (Spec'ed as 3.3V LVTTL if memory serves)
  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor

    In terms of CPLD complexity, I expect that MAX II would be able to realize the interface. But RGMII is utilizing a DDR transmission (dual edge). MAXII unlike e.g. Cyclone FPGA has no dedicated DDR I/O circuit. So a combination of dual-edge synchronous and asynchronous logic must be used. I doubt, if the timing requirements can be met at 125 MHz with this technique.

  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor

    Thanks for all your response.

    Is there any other possible solutions we have...?