Hi experts,
My customer is testing the MCSPI Loopback DMA example and they found there is large interval between words. The interval is significantly slowing down the communication, is there any way to improve this?
Regards,
Hang.
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.
Hi experts,
My customer is testing the MCSPI Loopback DMA example and they found there is large interval between words. The interval is significantly slowing down the communication, is there any way to improve this?
Regards,
Hang.
Hello Hang,
Thank you for the query.
Can you please request your customer to add additional details on the configuration.
Is customer testing on a custom board or on the EVM/SK.
Regards,
Sreenivasa
Hi Sreenivasa,
Since customer is looking a solution here, you may consider that they are testing the example in sdk using launchpad. If we can demonstrate a method to shorten the interval on Launchpad, they can migrate it to their setups.
Thanks,
Hang.
Hello Hang,
Thank you
Can you please request your customer to add additional details on the configuration.
Are you able to add some information on the configuration for me to check internally.
Regards,
Sreenivasa
Hi Sreenivasa,
They are testing MCSPI Loopback DMA example lld on launchpad. To better illustrate the issue, the transfer is looped forever. The SPI0_CLK and SPI0_D0 is measured in the above picture.
To reproduce the issue, you may use the mscpi_lookback_dma_lld example, modify it into looping forever like below:
Then measure the 47pin on J5 connector on the launchpad. You can see the interval between each cluster of the lock is 250 ns.
Regards,
Hang.
Hello Hang,
Thank you
Please review the below thread.
Regards,
Sreenivasa
Hi Sreenivasa,
Thanks for the info. .Going through the link, I see there's similar issue on AM62.
Andreas suggests using dma, which we already did, and using tube mode, which is not very helpful as per feedback. So, the root cause and solution are still not found
Andreas also offered some other solutions in the thread but that's for Linux and may not work for am243x.
I would forward the link to customer to help them understand it, but it looks to me that we still need to find a solution for the issue.
Thanks,
Hang.
Hello Hang,
Thank you
Andreas has been looking into this.
You might want to follow the thread.
I will also check with Andreas if he has some suggestions.
Regards,
Sreenivasa
Hello Hang,
I see you closed the thread. Did you make some progress?
Regards,
Sreenivasa
Hi Sreenivasa,
Customer read earlier from the link in the previous response https://e2e.ti.com/support/processors-group/processors/f/processors-forum/1310321/sk-am62-improve-poor-bandwidth-utilization-on-mcspi-bus
that there is no good way to improve this for the moment, so they don't prioritize this issue anymore. Since I have not heard from them on this issue for 2 months. I'd like to close this thread for the moment. We may open another thread on the same issue/ open this thread if they raise the issue again.
Regards,
Hang.