We want to process real-time data that is continuously being collected from the ADRV9009/ADRV9008 card using the ZCU102 card.
We succeeded in capturing the data of ADRV9009 using No-Os. We are currently performing non-real-time processing by collecting a large data frame (via the axi dmac transfer function in the headless.c file).
By real-time means, we want to process the data frames (small) within the time between the two data frames. The project no-os which we are using stores the data in the DRAM. By experimenting, we discovered that the processor in zynq executes instructions at a relatively slow speed, and there is a possibility of missing a few data frames because the processing time for the data frame is longer than the period between two data frames.
This issue is also discussed in another thread, the link to which is provided below.
Could you please advise us on how to solve the problem and carry out the processing in real-time?