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.

Multiple Multi-Segment SRIO Message Queues are not Utilized

Hello,

In my quest for multi-segment SRIO messaging, I followed the recommendation in the user guide to have four queues configured for multi-segment operation with no letter masks in the mappers. My concern is that the FPGA which is driving the data into the C6472 on my custom board is backing up on letter slots, and only one of the four queues is ever used? How can that be?

Thanks!

  • Hi Nortan,

    We are working with expert to answer this post. Thank you for your patience.

  • What exactly to you mean by "backing up on the letter slots"?  If you have a single driving device (FPGA) in your case, an the driving device is only sending one message (mailbox+letter) at time, then the DSP can keep up with a single RX queue.  Maybe I'm not following your question though.

    Thanks,

    Travis

  • Well I don't think that it is sending only one mailbox + letter at a time. Is one at a time defined by waiting for all of the acks before sending the next mailbox + letter? If that is the definition, then no, we are not running only one at a time.


    The FPGA is currently only using mailbox 0, and it advances through the letters if it does not get an ack back from the C6472 in time.

    I get Illegal Transaction Errors relatively often (ID CAP: 0x00ff00ee, CTRL CAP: 0xb3c00000). Or I will get message timeout errors.

  • norton256 said:
    Well I don't think that it is sending only one mailbox + letter at a time. Is one at a time defined by waiting for all of the acks before sending the next mailbox + letter? If that is the definition, then no, we are not running only one at a time.

    No, I'm defining one at a time as the packets for two different messages aren't interleaved.  So essentially if you are saying that they are not interleaved but instead back-to-back (even though the message responses aren't received back to the FPGA), then the DSP can receive them to one receive queue.


    Regards,

    Travis